TC210/Sec0898/INF October 2015

Transcription

1 CENELEC TC210/Sec0898/INF October 2015 EUROPEAN COMMITTEE FOR ELECTROTECHNICAL STANDARDISATION TECHNICAL COMMITTEE 210 EMC CLC/SC205A Study Report on Electromagnetic interference between Electrical Equipment/Systems in the Frequency Range below 150 khz Edition 3 See below:

2 Page 1 / 121 SC 205A MAINS COMMUNICATING SYSTEMS TF EMI STUDY REPORT Electromagnetic Interference between Electrical Equipment/Systems in the Frequency Range below 150 khz, Edition 3 October 2015

3 Page 2 / 121 SC 205A Study Report Electromagnetic Interference between Electrical Equipment / Systems in the Frequency Range below 150 khz, Edition 3 Content List Executive Summary... 5 Introduction General Specific situation in the frequency range khz Dimension of the EMC problem & Environment Situation of emission levels and EMI Specific EMC issues General MCS robustness The impedance behaviour General Variable impedance characteristic and its possible effects Impedance value, measurement and calculation Summary Long-term effects of EMI General Thermal effects on electronic components Increase of emissions due to ageing Summary Measurement issues General Status of standardization Characteristics of measurement quantities and requirements Classification of higher frequency phenomena Types of application General requirements Summary Emissions. Measurement and test results General Measurement and test results Large photovoltaic inverter installations Solar campus LV grid with multiple PV inverters Small PV inverter installations Measurements Measurements in Germany Measurements in Sweden Emission survey Lamps with electronic ballast... 36

4 Page 3 / General Measurements Measurements in Germany Measurements in Sweden Summary Electric vehicle charging Measurements in Germany Measurements in France Measurements in Switzerland Measurements in Austria Frequency-controlled heat pump Cola spender DVD player TV box Beer cooler Traveling circuses Power supply to fiber switches Power supply to a network router Plugin charger Single-phase PSU pretending a 3-phase problem Power supply to a PLC modem Microwave oven Desktop power supply with apfc Summary EMI cases. Measurement and test results General EMI due to conducted emissions EMI to NCE Residual current devices CNC mill disturbing household and commercial equipment Ceiling lighting disturbing studio equipment CNC machine Power supply to a TV box LED Construction light EMI with solid state electricy meter (accuracy) due to PV inverter Chirping noise in an office building EMI with MCS Power supply to a TV amplifier TV amplifier Elevator drive PC screen Punch-in clock Homeplug modem Voltage converter Voltage converter II Voltage converter to a broadband switch Central TV amplifier Voltage converter to a fiber switch Frequency-controlled ventilation in a school Frequency-controlled ventilation in an apartment building Rectifier inside a mobile site Undercounter bottle chillers EMI cases in Spain EMI cases in Finland Heat pumps Frequency control to a pellet boiler EMI due to radiated field strength from conducted NIE signals... 74

5 Page 4 / General Radiated EMI to telecom equipment Radiated EMI in higher frequency ranges Descriptions of EMI cases Conductive and radiated EMI from an EV battery charger Radiated EMI from an electric fence Summary Standardization, Legislation & Regulation General Present legislative & regulation situation Present standardization situation Summary. Needs for the future Conclusions Recommendations Annexes A: MCS robustness B: Measurement issues: Frequency domain vs. Time domain analysis C: Comparison of measurement methods for the frequency range khz D: Acronyms & abbreviations E: Bibliography Figures 14 Tables

6 Page 5 / 121 Executive Summary With this 3 rd edition of the SC 205A Study Reports on electromagnetic interference (EMI) in the frequency range below 150 khz, in the light of different positions on and in evaluating related electromagnetic compatibility (EMC) problems, with additional measurement results concerning emission levels in the supply network and results from investigations of additional proven EMI cases, the given problems are highlighted in more detail and recommendations for what to do in the future are provided. Following to the first two editions of these Study Reports [1, 2], the main findings can be summarized as follows: A further increased number of different types of electrical equipment, figuring as an EMI source or being susceptible to related emissions/figuring as an EMI victim has been recognized; so,.. The given EMC problem is a broader one, not focusing on the coexistence of general (non-mains communicating) equipment (NCE) with mains communicating systems (MCS) only, where such ones would be operated. Anyhow, EMC with MCS represents a particular problem but not only concerning such ones operating in CENELEC-band A, in Europe, but also concerning such ones operating in the higher range 150 khz 500 khz, outside Europe. To some extent, conducted emissions having been measured show quite high peak levels, while there are also devices with almost no relevant emissions on the market. In several cases, the emission level and/or the immunity seem to evolve with time, due to ageing of some components or the repetition of surges; this is mainly observed with small and cheap devices like power adapters, and is likely to increase with the development of common chargers and alike. For many devices, the principal impact of high frequencies above an acceptable level is not on functionality itself, but on the failure rate. Lab test have proved the incoexistence between NCE and MCS in case of parallel operation at the same time on same frequencies in case of higher non-intentional emission (NIE) levels. With regard to the recognized specific characteristics of emissions in the frequency range khz, for standardization, concerning ensurance of EMC between NCE and MCS, it may not be appropriate to consider only a certain margin between NIE and MCS signal levels over the whole frequency range. Again, cases of EMI via the radiated path are reported, thus confirming radiated EMI as an EMC issue over the whole frequency range khz, with EMI victims like broadcast time signal systems and telecom systems like xdsl. Considering different measurement methods being provided by some standards and different methods being applied at such measurements, measurement results are not clearly comparable. A unified measurement method, considering the specific needs of the frequency range khz and the specific purpose of measurement, would help to cope with this problem. Emission levels in the frequency range khz show a variability in time, which results in an insufficiency of the usual frequency domain measurement to comprehensively recognize a possible disturbing effect of an emission. Increased application of time domain measurement is recommended. The impact of the impedance in function of the frequency, in the supply network as well as in installations, represents an essential parameter for the propagation of emissions and, therefore, for there possibly disturbing effect to other electrical equipment. With regard to the limited knowledge of this aspect, further investigations appear a necessary, also for finding a basis for typical impedance values, thus being an essential basis for measurement parameters and EMC requirements. After having proceeded with investigations more and more, a lot of new findings having been obtained may lead to the anticipation that beyond closing gaps in standardization, for what work is going on, at least in a first step, major fields for more in-depth investigations appear as having been opened.

7 Page 6 / 121 Introduction In April 2010, SC 205A 1 published their first Study Report on Electromagnetic Interferences between Electrical Equipment in the Frequency Range 2 to 150 khz [1], providing first results of investigations on EMI in this frequency range, due to first EMI cases like such with Touch-dimmer lamps (TDLs) as an EMI victim, an inverter as an EMI source, and automated meter reading systems using powerline communication (AMR-PLC) figuring as EMI victims as well as sources. With a second edition of this Study Report, in 2013 [2], the specific situation in the frequency range khz and the broad relevance of recognized EMI for safeguarding EMC also in this frequency range was highlighted; that also with provision of results of measurements on the existing situation of emissions in the grid as well as with an overview of results of investigations on proven EMI cases. This 2 nd edition of the Study Report has been submitted for vote as a CENELEC TR (50627) and received a positive vote. Having proceeded with collection of related information, with this 3 rd edition of related SC 205A Study Reports, further extended information is provided about the given EMC problems in the frequency range khz, concerning EMC between electrical equipment in general as well as EMC between NCE and MCS as a particular issue the given situation of related emissions in the grid, with another number of measurement results EMI cases and related investigation results new findings on parameters to be considered when dealing with EMC in this frequency range, in particular related to o o the impact of the network impedance and its variation over time on the more or less disturbing effect of emissions in this frequency range the behaviour of emissions in this frequency range over time and the increasing need for performing also time domain measurements for comprehensively evaluating emissions and their disturbance potential the actual standardization situation needs for the future, concerning o measurement of related emissions o o o investigation on the impedance of the grid / in installations over time closing gaps in standardization installation guidelines and possibly regulatory measures related to the ageing effect. 1. General As the results of investigations having been made for the Study Reports [1, 2] show, utilization of the frequency range khz by electrical equipment shows several effects, which, compared with other frequency ranges, needs to be considered as a somehow specific situation; that resulting in the need for consideration of these effects when considering related EMC issues. In the following, extended information about this specific situation and measurement and investigation results on related emissions and EMI cases are provided. 2. Specific situation in the frequency range khz Concerning the origin of disturbing interactions between electrical equipment using frequencies in the range khz, there are to be considered NIE, being inherent with the operation of (power) electronic equipment / systems, increasingly being applied from large inverters and uninterruptible power supplies (UPS) down to small voltage converters 1 CENELEC SC 205A Mains communicating systems

8 Page 7 / 121 signals from MCS, being the intended means for transmission of information over the mains in general. In Europe, MCS technology is chosen for automated meter reading systems (AMR- PLC) to a large extent. Both types of emissions contribute to potential for causing disturbing effects to several types of electrical equipment. With regard to the increasing application of equipment using related technologies, increasing weight of the problem can be expected. Sometimes, such emissions are also called Supraharmonics [4], this term obviously being intended to describe the frequency situation of such emissions being above that one assigned to harmonics of the fundamental -- usually 2 khz or 2.5 khz. While the term harmonics is connected with the fundamental of the supply voltage, for Europe 50 Hz, emissions in the frequency range khz may have any frequency, without any connection to the fundamental. From the disturbing effects point of view, the following types of equipment appearing as being susceptible to such EMI and having been involved in related EMI cases have been recognized NCE including solid-state meters (without communication) MCS, also for AMR-PLC systems. Generally, when viewing at EMI in the frequency range khz, the following combinations of equipment / systems need to be considered A NCE NCE B NCE MCS C NCE MCS Fig. 1: Combinations of groups of equipment / systems to be considered related to EMI in the frequency range khz NCE General, non-mains communicating equipment / systems, e.g. power supplies, inverters as an EMI source telecom equipment, measuring part including power supply of a solid-state meters, TDLs as an EMI victim MCS Mains communicating systems as applied e.g. for distribution network operators (DNO) AMR-PLC systems, home automation While case A relates to disturbing interactions between NCE (see e.g ), EMC problems in cases B and C are based on the parallel utilization by NCE on the one side and by MCS on the other side, around the same frequencies, i.e. NCS switching frequencies (and its multiples) and MCS signal frequencies (out of the standardized CENELEC band A). It appears being necessary to get awareness of the need for a new, more complex view on phenomena in the supply voltage, concerning khz NIE: periodic occurrence of currents/voltages with spike shape at certain switching frequency or variable ones, stemming from NCE, not linked to the power frequency, in a frequency range between 2 khz and some tens of khz, and its multiples its EMI effect being mainly dependent on the voltage level resulting from emitted currents and the impedance situation. Signals (MCS): discontinuous currents/voltages at information-dependent frequencies not being linked to the power frequency, in the frequency range 3 khz to 95 khz (148.5 khz), with different signal durations the EMI effect to other equipment being dependent on the time behaviour of emissions. Mixture NIE + MCS: parallel operation of NIE and MCS in supply areas where MCS are applied, with possible EMI between the both groups of equipment. The understanding of EMI requires to

9 Page 8 / 121 analyze signals in the frequency domain as well as in the time domain, what is not considered in measurement methods being standardized today (see Chapter 6, Annexes B, C). Starting with emission levels and immunity performance, EMC is always a result of coincidence of several technical factors. These factors appear as being wider spread for the frequency range khz. Tab. 1 gives an exemplary overview of the different impact factors on EMC in this frequency range and its primary influence on the aforementioned groups of electrical equipment. Tab. 1: EMC factors and its relevance for EMI in the frequency range khz EMC relevance for EMI relevance for NCE MCS NCE MCS as a source as a victim Equipment design Emission Level frequency / frequencies frequency volatility time behaviour Immunity level in general level due to specific characteristics of victim Load impedances Nominal value Volatility in equipment application Network / installation impedance Mains design As the Table shows, impedances represent a major impact factor on EMC in the considered frequency range on equipment side as wells on supply network side, thus influencing the effect of emissions in the range khz in terms of attenuation of emissions frequency-dependent effect of the impedance of the grid / the installation and individual devices. (see also chapter 4, item 5.3). It is recommended to further investigate this effect, in particular the process leading to a remarkable reduction of the impedance the effect of mitigation measures for reducing emission levels effects on the impedance related to the appropriateness of avoiding measures resulting in an inappropriate reduction of the impedance, which should be considered by defining a related mask The above-mentioned relevance of the recognized EMC problem for several types of electrical equipment as well as the increase EMI origins in the frequency range below 150 khz gives rise to considering the issue as being of broader importance, featuring with effects like degradation of performance / malfunction, including the display of wrong meter register values in case of solid-state meters loss of function

10 Page 9 / 121 hidden effects like ageing of electronic components and reduction of lifetime of equipment see also Tab Dimension of the EMC problem & Environment As shown in Fig. 1, one part of the problem relates to EMI between NCE, thus without any involvement of MCS and representing a problem, which relates to electrical equipment worldwide. On the other side, EMI between NCE and MCS are to be considered, whose occurrence can be seen as limited to areas, where MCS operating in the frequency range khz (CENELEC bands A D) are applied, what is the case for Europe, but not only. Therefore, measures to ensure EMC in the frequency range khz need to be taken a) for the general case A: to consider an appropriately limited emission from as well as an appropriately high immunity of NCE, b) for the regionally limited cases B and C, a solution which ensures co-existence between NCE and MCS For the specific case b), with the previous editions of this Study Report, some overview is given of frequency utilization by MCS applications and tasks for MCS deployment of MCS, where, in Europe, in the huge majority of cases, for the communication between meter and data concentrator an MCS is chosen. has already been given, in particular related to AMR-PLC for the smart grid for electricity supply [1] 2, [2] 3. The EU set out energy and environmental policies to improve the energy efficiency by 2020: the Third Energy Package targets at 80 % of smart meter penetration by 2020 (see also item 9.2). Many Member States established complementary national regulation and are paying close attention to the required deployment of smart metering systems (see e.g. [5] Fig. 2 shows the typical architecture for smart metering systems [6]. Fig. 2: Typical smart metering architecture AMMS Automated Maintenance Management System HHU... Hand-held Unit Further, multi-utility applications of PLC technology are being persued in Norway [7] and The Netherlands [8]. After having transferred the water and gas meter data via short range wireless communication from the related meters to the electricity meter, such systems would route the smart (electricity) meter s PLC communication to transmit all meter data to the head-end system, from which the data would be 2 Study Report I, item Study Report II, items 2.2, 6.1.3

11 Page 10 / 121 forwarded to the designated parties. That together with integrating signalling for home automating functions in the course of services offered by the one or the another utility to network users (Fig. 3 [9]). Fig. 3: Multi-utility application of PLC technology Fig. 4 gives an overview of the actual status of smart meter projects in Europe [6, 10], where more than 50 M smart meters are already installed, in the huge majority of cases, an MCS being chosen for the communication between meter and data concentrator (AMR-PLC), that solution being even presented as the preferred communication technology by national regulators. Fig. 4: Smart Meter rollout status in Europe [6, 10] (Source: European Commission, DG ENER, 2014) In two countries, Italy and Sweden, the rollout is completed, with ~ 0,6 M smart meters installed in Sweden and more than 35 M smart meters installed in Italy.

12 Page 11 / 121 Altogether, with roll-outs in 19 Member States, smart meters are expected to be provided to around 72 % of European electricity consumers by A report from the European Commission, released in June 2014 [11], measures progress on the deployment of smart meters across the EU. To date, Member States have committed to rolling out close to 200 M smart meters for electricity and 45 million for gas by 2020 at a total potential investment of 45 billion. By 2020, it is expected that almost 72% of European consumers will have a smart meter for electricity while 40% will have one for gas. provides an overview of the results of cost-benefit analyses (CBA) concerning smart metering systems for electricity and gas in the EU. An overview of the legal and regulatory framework for metering services and the smart metering landscape in Europe, which is strongly fostered by European policy makers, can be taken from [12]. Beyond that, as already mentioned in Study Report I [1], a number of related bigger AMR system rollouts have been started or are under way in Russia, North America, South Africa, Australia and Asia. Tab. 2 provides information about major ongoing mass-roll-out projects of smart metering in Europe [6]: Tab 2: Mass-roll-outs of smart metering in Europe (November 2014) Country DNO Meters to be Meters Project scale installed installed France ERDF 35 M Pilot test and rollout Italy ENEL D. Muntenia -- Pilot test 864 Italy ENEL 32 M Rollout completed Portugal EDP 0,1 M Demo Spain Gas Natural Fenosa 4 M Rollout Spain Hidrocantabrico 0.73 M Rollout Spain Endesa 13 M Rollout Spain Iberdrola 10 M Rollout When looking at disturbed equipment, in cases of EMI, careful consideration is advisable when talking about the (more or less) limited number of affected equipment; here, the sole consideration of percentages of the overall number of equipment being installed over a country, one continent or even worldwide may result into misleading, biased conclusions. That may be explained as follows: EMI cases have been identified with different types of electrical equipment (NCE as well as MCS; see Tables 4, 5). Reports about lower numbers of such EMI cases may lead to considerations for the taking of general (standardizational) measures not being justified and for taking case-by-case solutions appearing as the preferable solution. Apart from the solid-state meter case, which is sensible in terms of tarification, it is not the more or less limited number of reported EMI cases, in particular in terms of percentages of the overall installed numbers of equipment, which needs to be solely considered for evaluating the dimension of the problem, but the aspects of diversity of types of equipment, showing a broader spectrum of affected types the number of customers suffering from one of the aforementioned EMI effects and possibly complaining; a quite higher number of EMI cases, not being recognized by network users as such but resulting in a quality loss in operating electrical equipment at the time of interaction or, with some delay, in a reduction of lifetime, through ageing effects the efforts of DNOs, who are in most cases the primary instance for receiving related com-plaints, to analyze the related EMI cases and to find the underlying causes respectively the polluter taking measures for solving the single problem, e.g. replacing one element of equipment being involved in the EMI or a punctual mitigation measure, with consent by the affected customer and with an accepted solution concerning the bearing of related costs

13 Page 12 / 121 publicity effects when customers publicize related EMI problems they are suffering, referring to their purchase of good quality equipment (in Europe: with reference to the CE mark not related to quality but to conformity with essential requirements of Directives), A specific case relates to EMI between NCE and MCS. Beside cases of EMI caused by MCS to NCE (case B of Fig. 1), which has been verified as being quite lower than vice verse, values of 1 % (estimated rate of 1 % of the installed meters relevant for mitigation, from the Netherlands) to ~1 % to ~26 % (disturbed meter communication, numbers stemming from a survey covering the experiences from several Finnish DNOs [13]) are reported. [1, 14], values of some tens of percent up to 100 % of the smart meters of an AMR-PLC system assigned to a certain data concentrator have been reported from several countries to have lost communication with the system, thus being unreachable and therefore unreadable for the DNO for the duration of a disturbance [2, 3]. In the first case, the percentage numbers may be considered as quite small, but the effect to the DNO, whose operational means are affected in the order of magnitude of some devices may better highlight the dimension of the problem economically as well as related to the effect opposite to the customers. In the second case, the loss of significant parts of an AMR-PLC system may highlight the severe deterioration of the operation of such systems s operational means for DNOs 4. Situation of emission levels and EMI As related investigations have shown, at present, NIE are occurring on the mains with levels of up to 160 dbµv (~141 V) and even more, thus reaching and even significantly exceeding the standardized signal injection limits for MCS, 134 dbµv (5 V). (see also [15, 16]). As having been found when analyzing different EMI cases as well as in laboratory investigations generally, EMC is not ensured when high emission levels, of whatever type, are affecting insufficiently immune electrical equipment concerning the parallel operation of power electronics equipment, with its (switching) frequencies (and its multiples) and signal frequencies (out of the standardized CENELEC band A), with contemporaneous operation of power electronics and MCS around the same frequencies, proper function of MCS cannot economically be ensured. Whilst, referring to Fig. 1, for case A, beyond more or less economic solutions with case-by-case mitigation, the problem might be solved by setting appropriately low limits for NIE from NCE figuring as a potential EMI source; appropriately high immunity requirements, which consider also beyond levels the specific EMI interaction mechanisms in the frequency range khz, for case C sufficient consideration of standardized signal level limits for MCS at design of NCE is needed, case B appears as not being solvable by the sole setting of levels; that with regard to the recognized need for a remarkable signal-to-noise ratio (SNR) between MCS signals and NIE. For the related test conditions, lab investigations at ERDF, on the performance of two particular implementations (G3- PLC technology see ITU-T G.9903 [123] and ITU-T G.9901 [124] and ERDF s S-FSK PLC technology based on IEC [125]) resulted in the recognition, that for proper MCS function a margin of (at least) 45 to 50 db for the frequency range 9 95 khz respectively (at least) 52 to 56 db for the frequency range khz seems to be required (Fig. 2), that in particular when considering the attenuation of signals along the mains. Ensurance of such a margin is to be seen also together with economic constraints and for equipment design and manufacturing.

14 Page 13 / 121 Fig. 2: MCS signal levels vs. NIE: Margin requirements for proper MCS function with Maximum PLC signal level : corresponding to the maximum output level of the transmitter. Minimum PLC signal level without errors : corresponding to the minimum level required by the receiver such as no communication errors occur (0% < FER < 5%) for a given noise level. PLC link breaking level : corresponding to the signal level at the receiver below which no PLC communication is possible (FER = 100%) for a given noise level. Noise level : corresponds to the measured noise level at the receiver Further, from the point of view of correct MCS operation, possible near-far-problems are to be considered, where a modem of an MCS is facing two components in the supply voltage a signal from its MCS, having been quite attenuated along the mains, i.e. with a level of e.g dbµv an NIE from a nearby situated NCE, therefore having experienced quite low(er) attenuation along the mains Beyond the issue of signal and disturbance levels and a related margin in between, when dealing with case B (Fig. 1) problems, also the quite important impact of the impedances of connected equipment needs to be considered; that relating to the ever changing overall impedance value of all equipment/systems connected to a certain supply area/installation and its effect to MCS, that including, the impedance behaviour of possible measures for mitigating disturbances to signals from MCS the attenuation of MCS signals along the mains, through the resulting network impedance, is to be estimated between 40 db and 100 db typically 60 db --, thus resulting in accordingly lower signal levels at a certain point in the supply network compared with the signal injection level, which is limited by IEC [18] / [19] and CENELEC EN [20] over the frequencies in khz respectively at signal frequencies of MCS in particular. Finally, as new investigations with analyzing emissions in time domain show, dependent on the characteristic of an NIE respectively a signal, as well as dependent on the parameters of the signal analysis, deviations from the measurement result in frequency domain with an order of magnitude of some tens of db may be recognizable, thus indicating a higher disturbing effect of the analyzed emissions than having been recognized from frequency domain measurement only. Already in Study Report II [2] some overview has been given on main groups of EMI effects and examples for equipment figuring as EMI sources or victims as well as for effects of EMI to equipment in the frequency range khz. Considering additional information having been obtained during work on this 3 rd Study Report, the related Tables are completed as follows:

15 Page 14 / 121 Table 3: Main groups of EMI effects (Non-intentional) Emissions from network users equipment at or close to frequencies used for MCS interfere with intentional MCS signals, leading to disturbance or loss of MCS communication Multiples of (non-intentional) emissions from network users equipment, being close to frequencies applied for MCS may cause interference with the MCS resulting in failed communication Distortion of the supply voltage due to discontinuous (non-intentional) currents/ voltages from network users equipment or signal voltages from MCS may lead to degraded performance or maloperation of network users equipment Network users equipment representing a low-impedance path at frequencies used for MCS lead to an attenuation of the intentional MCS signal which might disturb or interrupt communication ( shunting effect ) (Non-intentional) emissions from network users equipment or (intentional) MCS signal voltages may result in somehow higher currents, leading to overheating and accelerated ageing of components in network users equipment Table 4: Equipment figuring as a source of EMI, Examples Inverters (e.g. photovoltaic infeed inverters, drives for elevators, rectifiers) and variable speed drives (VSD) (e.g. in elevator drives, ski lift drives, heating system circulation pumps, air ventilation systems, household equipment) Switched-mode power supplies, e.g., PCs, faulty external power supplies of PCs consumer electronic / home entertainment equipment (e.g. TVs, DVD players) power supplies of TV antenna amplifiers, lighting devices with light emitting diodes (LED), commercial equipment like punch-in clock, camera surveillance systems, ICT equipment (e.g. fiber switches with built-in power supply, WLAN switches, 3G/4G base stations), uninterruptible power supplies (UPS)), charging devices (e.g. for electric vehicles (EV)) Lighting equipment (e.g. fluorescent lamps, compact lamps, LED lighting equipment) Household equipment (e.g. induction cookers, washing machines, electric shavers) Active filters Portable mains operated tools AMR-PLC Table 5: Equipment figuring as an EMI victim, Examples AMR-PLC Solid state meters Electronic control, e.g. touch-controlled equipment like Touch dimmer lamps (TDL), alarm systems, traffic control systems, studio recording control units and mixing desks, traffic lights,

16 Page 15 / 121 in heating systems, for street lighting, in urinals, for doors, in kitchen appliances (e.g. steam irons, coffee machines, ceramic hobs) Communication systems, e. g. PABX, Ethernet-systems, ISDN-, ADSL-modems, IP network branch exchange, Routers, LAN, Home telephones Telephone systems including inductive train radio systems Earth leakage circuit breakers (ELB) Contactless magnetic card readers, credit card terminals Keyless entry systems Notebooks (cursor position) Broadcast standard time-signal systems (e.g. DCF77, Japanese system) Road vehicle smart keys TV and radio receivers Mobile radio Radio systems (Amateur radio, for railway control) Table 6 provides information about different effects of EMI to certain equipment in the considered frequency range. Table 6: Effects of EMI to equipment in the frequency range 2 khz 150 khz, Examples TDLs Street lighting Traffic lights Traffic control system for public transportation buses Solid state meters MCS Heat control with time basis through DCF77 signal Radio system for railway control Contactless magnetic card reader Home telephone Heating systems ADSL modem Routers Notebooks Ceramic hobs Coffee cooker Professional hair-dryers Steam irons Washing machines Electronic ballasts Induction cookers Unintentional switching (between light steps, OFF, also ON) Unintentional switch-on and off Malfunction Malfunction Displaying wrong meter register values Temporary or quasi-permanent loss of communication Malfunction Malfunction Malfunction of reading function Malfunction of ringing Incorrect alarms due to sensor faults Loss of link, CRC error Loss of synchronization (40, 50 and 70 khz) to the network Disturbed cursor position (37 khz) Incorrect relay switching Incorrect control lamp function and malfunction Switching ON/OFF spontaneously Insufficient heat, water loss, incorrect control lamp blinking Self-restart (some hours) after end of operation phase Audible noise Audible noise

17 Page 16 / 121 PCs and Lamps PABX Telephone (analogue) TV and radio receiver Video telecom circuits Inductive Train Radio System Keyless entry systems DC link capacitors in rectifier circuits Increased emissions through dryout electrolytic capacitors Automatic urinal water control Broadcast standard time-signal systems Amateur radio Audible noise Audible noise Audible noise Audible noise (up to 20 khz) Audible noise Audible noise Malfunction Thermal stress, Ageing, Lifetime reduction through larger HF currents Ageing, Lifetime reduction of other equipment Switching to permanent operation Electronic clocks: being fast (gaining up to 15 mins per day), Malfunction of control circuits fed by the time-signal Disturbed reception of distant transmitters Tables 4-6 may need further amendments in future, following further investigations. 5. Specific EMC issues 5.1 General EMC problems in the frequency range have been and still are an item for extensive discussion in standardization committees like IEC SC 77A WG8. This discussion can be divided into two basic options To strive for standardized emission limitation and immunity requirements, which together may ensure disturbance-free operation of electrical equipment. As discussions show, setting of limits and requirements that way may cause high costs for equipment in general, o o possibly without being justified by the number of occurring EMI cases in the case of MCS being involved in related EMI connected with the question of relevance of the problem for areas with MCS application only. To agree upon less stringent emission limits and immunity requirements, leaving the solution of occurring EMI cases up to related mitigation measures to be taken. This option leaves open questions concerning o o the appropriate choice of mitigation measures who would need to bear the costs for such measures Of course, different parties, like manufacturers and appliers of different types of equipment which may interfere with each other, have different positions in terms of preference. It appears as quite difficult to evaluate the questions which of the two options would be the more appropriate in an objective manner. In any case it is appropriate to distinguish in related discussions between the general EMC matter, related to EMC between NCE the specific EMC matter, related to EMC between NCE and MCS Concerning the latter, as also said in reports from TU Dresden (TUD) [21], it should be guaranteed, that NIE as an example Battery electric vehicles (BEV) chargers are mentioned as originators are lower than the intentional emissions from MCS in order to ensure its proper functionality. Analyses made by ERDF indicate an overall margin of around 50 db between maximum PLC signal level and noise level to ensure proper MCS functionality (see chapter 4, Fig. 5).

18 Page 17 / MCS robustness Different MCS design results in different EMC behaviour of MCS, also related to immunity to emissions from NCE, i.e. disturbances. The robustness is the ability of a system to function in an acceptable way in the presence of hazards. It is also the degree to which a system can function After investigating all the interferences caused on the mains in the frequency band khz, an ERDF paper introduces different techniques and methods that could be implemented at the chipset, meter and network level to improve the system s robustness at network level against potential noise and interferences [23]. This paper also covers the propagation channel which is the principal contributor to many problems and limitations that beset Power Line Communications (PLC). As main components of EMI to MCS are figuring high noise levels (low) impedance Influence of network situation between TX and RX of MCS over time. Here, the time dependability is to be distinguished between ad noise: change of attenuation over time, due to switch-on and off of loads periodic change of cumulative load impedance behaviour over time, within the 50-Hz-cycle peaks of noise depending on the phase relations of components at different frequencies (possible MCS input circuit saturation due to such peaks) (see also item 6). A noise is a bundle of electromagnetic interferences of any kind and from any source, that usually does not carry information, and that may overlap with the main signal at any point in the trans-mission channel. The robustness against noise is defined by the SNR. ad impedance: Dependent on the total impedance situation (supply network, loads (see item 5.3)), which is frequency and time dependent, signals from MCS are more or less attenuated. The resulting signal level at a certain network point is to be considered in relation to the local noise level. The attenuation or loss is the decrease of the amplitude of a signal during its transmission. It is defined by ratio of the output magnitude over the input. The robustness against attenuations is defined by a minimum threshold of reception at the receiver side in the absence of noise. Options for improving the robustness of MCS are given at chipset and meter level at modem/analog Front End transceiver module, by improving the receiver sensitivity and filtering unwanted out-of-band emissions adaptation of reflection factor to the network impedance application of digital low-pass filters as a part of the digital signal processor (DSP) reception optimization at DSP level with adaptation to the given noise conditions by applying appropriate statistical models at mains level application of analog filters/chokes behind the smart meter

19 Page 18 / 121 Such installation can raise several questions: - Which party needs to pay the power consumption (DNO or client)? - Which party needs to pay for the filter (purchase, installation, maintenance)? - Has the DNO the right to install a filter behind the smart meter? - Does the DNO need to protect the filter? Further, Annex A provides a set of principles for field proven effective PLC systems with regard to MCS robustness, operating in CENELEC bands. Concerning the mentioned value for attenuation of signals on the public supply network, attention may be drawn to up until now limited knowledge on related values and the impedance behaviour over time (see also item 5.3) 5.3 The impedance behaviour General As highlighted in chapter 2 and item 5.3 (see also [26]), the impedance of the grid / the installation with the connected loads represents a major factor for EMC in the frequency range khz. The influence of impedances is well-known from the area around the fundamental frequency of the supply voltage; it is considered for setting of EMC requirements, that based on a probabilistic approach, with reference values for the network impedance at power frequency, in IEC/TR [24] and the proposed impedance for emission measurements between 2 khz and 9 khz, in the Informative Annex B to IEC [25]. For khz, the impact of the impedance on the effect of emissions in this frequency range needs to be considered in a more detailed way, as it shows a quite more complex behaviour, based on a more systematic approach [26, 117]; that with regard to the higher frequencies and its effect to the impedances of the grid / installation as well as of the connected loads over the frequency range khz over time the complex waveshapes being inherent to power electronics application as well as from MCS and leading to the recognition of today, that for the considered frequency range, emissions and their possibly disturbing -- effect on electrical equipment increasingly needs, beyond, frequency domain analysis, investigation in time domain Variable impedance characteristic and its possible effects As an example, from an ongoing investigation, Fig. 6 shows the result of measurement of the network impedance in a certain public LV supply network during a period of the network frequency 50 Hz, in a spectrogram. Fig. 6: Network impedance behaviour in khz across one cycle of the 50-Hz-supply voltage

20 Page 19 / 121 There can be recognized the change of impedance values during a cycle, at a certain frequency between 2 khz and 150 khz a somehow contrary tendency of the network impedance when proceeding in time when comparing higher and lower frequencies, i.e. that due to the lot of electronic equipment being connected to the network, the network impedance is at voltage maximum/minimum lower than around the zero crossings. This behaviour of impedances in function of the frequency, in the supply network as well as in installations, represents an essential parameter for the propagation of emissions in the considered frequency range (see also [26]) and its effects to electrical equipment, which will need to be considered at setting EMC requirements. In the light of the increasing application of devices using inverters for their control, generating NIE in this frequency range, the effect of EMC filters applied for reduction of the levels of the aforementioned NIE and the contribution of these impedances, varying in function of the frequency, to the impedance of the supply network / installation the EMI potential of such devices including EMC filters to MCS with operating frequencies close to the switching frequency of a converter connected to the mains nearby, which may reach up to notching effects to the MCS communication the discussion of voltage distortion compatibility levels up to 150 khz, an improved knowledge and understanding of propagation mechanisms in the frequency range becomes more and more important. Unfortunately, related measurement results are limited up until now, so that more investigations appear as being needed Impedance values, measurement and calculation Over all, the aforementioned need concerns a) the behaviour of impedance in the supply network as well as in installations as it may be expected, also for appropriately setting compatibility levels and, based on that, emission limits and immunity requirements; that together with appropriately chosen approaches for the loads connected to the mains, with consideration of the daytime aspect. b) the measurement method for obtaining comparable results from related impedance measurements c) calculation methods for obtaining a quick overview of the impedance situation without measurement From what is available up until now, as examples, the following investigation results may be mentioned: ad a) Grid/installation impedance values and EMI appear to be a quite variable matter. Anticipating that it would be quite wrong to produce THE impedance or noise measurement of THE grid at khz, that appears to require a statistical summary as being practiced for the 50-Hz reference impedance. Nevertheless, as improved knowledge and understanding of propagation mechanisms in the frequency range khz becomes increasingly important, it would be beneficial to collect; information about network/installation impedances referring to specific configurations and topologies (grid components, loads) and times. Measurement values between 0.5 Ω and 50 Ω are known It may be safe to expect the grid impedance amounting to values between 1 Ω and 30 Ω.

21 Page 20 / 121 Compared with this range of impedance values, providing an example artificial network for measurements in the frequency ranges 9 95 khz and khz, EN [20], referring to EN [27], Fig. 7 Example of artificial network 9-95 khz and khz (EN , EN ) indicates impedance values of ~ 4 Ω to ~ 24 Ω for the frequency range 9 95 khz and ~24 Ω to ~ 34 Ω for the frequency range khz Irrespective of the recognized temporary variability of impedance values as mentioned above (see Fig. 6), some measurement results have been provided e.g. with [28]. Fig. 8 shows impedance values measured by Technical University of Dresden (TUD) in the LV network of the university, to be seen as a strong urban business area; in other networks, e.g. in the rural area, considerably higher impedances may occur. 25 Z / Ω IEC Ed. 2 Amendment 1 Extrapolierter Extrapolated Verlauf 100 % 95 % 50 % 10 % 90 % f / khz Fig. 8: Impedance values measured in the LV network of TUD 5 % 0 % Further, Fig. 9 shows preliminary results of measurements of the impedance, performed with a network analyzer, at concentrator level in two ERDF networks.

22 Page 21 / Frequency (khz) IZI [Ω] IZI [Ω] Frequency (Hz) Fig. 9: Impedance values in ERDF LV networks Preliminary measurement results For the frequency range 30 khz khz these measurement results show typical impedance values from 1 Ω to 10 Ω (Fig. 9, upper chart), but also values up to 40 Ω are observed (Fig. 9, lower chart). a linear increase of the impedance with frequency almost the same impedance values on each phase For these preliminary measurements, similar results are observed at meter level. Maybe re-consideration of the impedance values for related measurement equipment towards possibly lower impedance values, for better mirroring the reality in supply networks / installations will be recommendable. ad b) Grid/Load impedance measurement Beyond an overview of the PLC measurement procedure being practiced at Laborelec, providing information about different hard- and software aspects of such measurements together with related safety measures, a paper from Laborelec [29], having been provided to SC 205A for the purpose of this Study Report provides also information about advanced calculations to visualize the characteristics of a grid or a load. Clause 4 of this paper deals with

23 Page 22 / 121 measurements of the impedance of loads, based on the injection of a signal into the EUT and further calculation of the relation between voltage and current in function of the signal frequency the effect of the grid on different frequencies between 20 khz and 120 khz, based on the generation of a related frequency sweep and afterward calculation of the impedance using the injected current and the measured voltage, thus providing an impedance value, =,, being valid for certain frequency values each and a certain location. In the calculation for the Impedance a division is made between the FFT of the voltage Ux and the current Ix. This technique can also be used on loads like home appliances, resistances. Sub-Clauses & 2 of this paper provide more information about treatment of the impedance measurements on With regard to high frequency resolution / low time resolution (Stationary Impedance) and low frequency resolution / high time resolution (Periodic Impedance) new power quality issues as well as EMI between converters 4 and control signals transmitted over the supply lines (PLC) which we are facing due to increasing application of Active Infeed Converters (AIC), thus using the advantages in terms of cost, flexibility and efficiency, the dependence of the emission voltage at an In-plant Point of Coupling (IPC) at the frequency of the emission voltage, the complexity of the low-voltage distribution grid and in most cases the complete lack of knowledge of the line layout and of devices connected, regarding that expressing some reluctance in an appropriate calculation of the spectral impedance of a distribution grid section, the new dimension of complexity of EMC due to the increase of frequency ranges and of the density of electronic equipment connected to the same grid, leading to the need for a more systematic approach,, the recognition that none of the solutions for the measurement of the grid / installation impedance, having been proposed in the literature up until now. [30-33], was covering the whole frequency range between 2 and 150 khz. a Swiss paper [34] presents an on-line grid impedance meter designed for accurate and fast on-line measurement of the spectral impedance of low voltage distribution grids over the frequency range between 2 and 150 khz, having been developed within the IGOR project (Interferences Generated by Inverters On the Grid) as a useful application of grid impedance measurement over the considered frequency range, the analysis of interferences between two photovoltaic (PV) inverters. ad b) & c) Measurement and calculation Effects of power electronic devices including related EMC filters can be analyzed based on the measured or calculated impedance situation in a certain grid area / installation. In the following, as an example for results of related measurement on the one side and simulation on the other side, results of investigations [35] on the propagation of the sidebands of the operating frequency of a PV inverter at 16 khz across a rural LV network, are provided. In a first step, voltage distortion was measured at 4 different points in the network (MP 1 4), starting from the terminals of the inverter (MP1) up to the terminals of the incoming transformer 4 The term converter here being used as an umbrella term for converters and inverters

24 Page 23 / 121 (MP4), showing, that the max. voltage of the 16-kHz-sidebands of the PV inverter in percent of the inverter terminal voltage attenuated significantly over a distance of a few hundred meters (see Table 7, line 1). Following simulation with a computer program with consideration of the frequency-dependent line parameters of the LV distribution network resulted in quite different values (see Table 7, line 2). Obviously, the network model used is not suitable for achieving reasonable results in this frequency range for the type of network considered. In order to improve the results, RFI ( Radio-Frequency Interference ) capacitors as being used in all electronic LV devices in order to fulfil radio-frequency emission standards and are typically connected line-to-neutral directly at the incoming point of the power supply, were introduced in a formal way into the network model: For each household, at the connection point with the public network, one RFI capacitor of 1µF was introduced per phase connected via a 5 m line, that representing around three electronic devices plugged in. Note. All capacitors were connected between line and neutral; on real products both neutral to ground and phase to ground (~22 nf) will also be found. Further investigation showed, that introduction of two RFI capacitors per household resulted in simulation results with levels lower than what was measured in the field (see Table 7, line 3). With this simple improvement in the model, the results according to line 3 of Table 7 were obtained: Table 7: Max. voltage of the 16-kHz-sidebands of a PV inverter in percent of the inverter terminal voltage Results of measurement and of calculation As can be seen, even without any detailed knowledge of the individual customers loads, reasonable simulation results, in comparison with the measured values, were obtained. the PV noise having been measured in a distance of 250 m of public network mixing overhead line and underground cable was measured with a highly attenuated value of 6% of the initial injected signal unsatisfying results of a first network simulation, with regard to the real behaviour of the supply network, thus leading to some resonances when looking at the impedance characteristics, in theory could be changed to satisfying results when using a model with introduction of an RFI capacitor of 1 µf per phase connected via a 5 m line and household. Over all it can be recognized, that RFI capacitors (or maybe also the DC link capacitance of electronic devices without power factor correction (PFC)) have a significant impact on the propagation of emissions in the frequency range above 2 khz; it may be noted that due to the application of different electronic equipment in every household the consideration of such capacitances seems to be justified for simulation purposes Summary Up until now, the impedance characteristic of the supply network / installations together with the connected loads was not considered to a larger extent. According to today s recognitions related to the frequency range khz, there may be summarized:

25 Page 24 / 121 The impact of the impedance in function of the frequency, in the supply network as well as in installations, represents an essential parameter for the propagation of emissions and, therefore, for there possibly disturbing effect to other electrical equipment. Emissions and their possibly disturbing -- effect on electrical equipment needs to be considered in a more detailed way, as it shows a quite more complex behaviour, based on a more systematic approach. Amongst others, beyond, frequency domain analysis, there appears to be an increasing need for investigations in time domain. Several existing proposals for measurement of the impedance in the grid / installation do not cover the whole frequency range khz. Further investigation are needed concerning appropriate methods for finding information about the impedance reality in supply networks and installations as well as for defining measurement and calculation methods considering the specific situation in the frequency range khz providing results being comparable with each other in general as well as between measurement in the lab and in situ Based on further investigations on the real impedance behaviour in khz in the supply network / installations, a check of existing specifications for measurement equipment impedance towards possibly lower impedance values, for better mirroring the reality in supply networks / installations, may become recommendable.. Results of further investigations will need to be considered at setting of EMC requirements respectively at maybe reviewing already standardized ones as far as available. 5.4 Long-term effects of EMI General In Study Report II ([2], item ) a possible longer-term effect of emissions in the frequency range khz when impacting electronic equipment was already highlighted. There are two different types of problems to be considered Higher emissions in the frequency range khz causing thermal stress to and following reduced lifetime of electronic equipment Occurrence of increased emission levels due to ageing of electronic components the first problem having potential to effect in the way of the second-mentioned problem (see Fig. 10) Fig. 10 Long-term EMI effects of EMI in khz

26 Page 25 / Thermal effects on electronic components TUD has carried out research on the systematic analysis of the impact of emissions in the frequency range khz, in particular on the additional thermal stress on electronic mass-market equipment [36]. One central part of all commonly used rectifier circuits is the DC-link capacitor that provides the smoothed DC voltage to the internal circuit. In order to supply the DC-link, a rectifier circuit is used. Circuits without PFC and circuits based on capacitive voltage dividers represent the simplest and most sensitive designs. For both designs capacitors are directly connected to the grid during the recharging interval. Due to the low impedance at higher frequencies, already small HF voltages can cause significant additional currents flowing through the capacitor. As an example, Fig. 11 presents a rectifier circuit without PFC, which is commonly used in compact fluorescent lamps (CFL) with rated power below 25 W. Comparing Figures 12 and 13, the additional HF component in the current flowing through the capacitor can be clearly seen in Fig. 13. Fig. 11: Simplified schematic diagram of a CFL (11W) without PFC Fig. 12: Voltage and current measurement at 230V (50Hz) Fig. 13: Voltage and current measurement at 230V (50Hz) + 2.3V (5kHz) The analysis is based on measurements using different lamps with integrated electronic ballast as example (see also item 7.2.3). Two different types of CFL (one without PFC, one with active power factor (PF)) and one LED lamp were selected for a detailed study of the temperature behaviour inside the lamps. As an initial step a purely sinusoidal voltage with 230 V was applied and, after stabilization, the temperature values were logged. In a second step a two-frequent voltage that consists of a fundamental component with 230 V and a HF component with variable magnitude up to 10 % of the fundamental voltage were applied. As results, HF voltages can have a significant impact on the thermal stress and subsequently on the lifetime of electronic equipment, which contains rectifiers with electrolytic capacitors as DC link. The temperature increase at the DC-link capacitors of the equipment under tests (EUTs) amounts up to 1 Kelvin per percent HF component compared to the measurements with sinusoidal voltage, thus leading to temperatures measured at different points of the EUTs (CFL, LED) amounting up to ~ 74 C.

27 Page 26 / 121 Due to the different mounting conditions and air streams at the ceiling, the temperature increase may be even higher under real conditions. Besides the additional heating, especially audible noise was observed for a lot of the analyzed equipment. This can occur already at HF voltage levels of about 1 %. The investigations were mainly, but not exclusively focused on the DC-link capacitor. In most cases not only the temperature of the DC link capacitor additionally increases due to the increased input current, but also other elements, like the rectifier diodes are affected. It is most likely that this additional thermal stress increases the ageing and finally reduces the lifetime of electronic equipment. Many manufacturers are not yet aware of this problem, which would become more evident, if HF emission levels in the grid would increase in the future. Further studies are carried out to quantify this impact Increase of emissions due to ageing On the other side, as pointed out in TR 50627, ageing of electronic components in electric equipment can cause increased emissions in general and, following to that, result in EMI to other devices; that resulting in a status of electrical equipment not further meeting the essential requirements (ERs) of the EMC- Directive (EMCD), although having been so at the time of being placed on the market Summary In related discussions some experts express the opinion, that this ageing problems were not an issue related to the EMCD [37]. In any case measures seem to be necessary for ensuring EMC and therefore conformity with the ERs of the EMCD also after some time after being placed on the market. Beyond the technical part of such measures, i.e. the application of technical mitigation measures, national legislation measures appear as a solution for easing to take enforcement action on interference from products after the action of placing on the market and putting into service. Related legislative measures to address situations outside the EMC Directive [37] and the Radio and Telecommunications Terminal Equipment (RTTE) Directive ([38]) are e.g. planned to be taken in the UK [39]. 6. Measurement issues 6.1 General An essential question for gaining measurement results on emissions in the frequency range khz and evaluating /comparing the measured values is that one of the applied measurement method. In general phenomena in the frequency range 2 khz to 150 khz can be analyzed either in time domain (high-pass filtered momentary waveform) or in frequency domain (spectrum). Both methods have their advantages and disadvantages. The choice of one of these methods as the one to be applied depends on the characteristic of the higher frequency content of the analyzed signal; in many cases a combination of both analyses is beneficial. As an example for time domain consideration, reference to the recognitions reported in Study Report I [1] related to EMI with TDLs may be made. In that case, from the spectrum the interference mechanism cannot be explained. Graphical presentation in time-domain is usually a plot of the high-pass filtered momentary waveform of the signal (Fig. 14a). For frequency domain most commonly a spectrum (Fig. 14b) is used for graphical visualization of a particular time instant. If in addition to frequency also the characteristic over time shall be included, a spectrogram (Fig. 14c) is used.

28 Page 27 / 121 Fig. 14a: Oscillogram of the filtered current Fig. 14b: Voltage and current spectrum Fig. 14c: Spectrogram of the voltage Fig. 95 Different types of graphical presentation in time- and frequency domain (Example: EV charger) For visualizing the frequency characteristic within one fundamental cycle, the time resolution of the spectrogram needs to be increased (see Fig. 30, item ). Further details can be found in [40].

29 Page 28 / Status of standardization Up until now, discussions on EMI investigations as well as on the setting of compatibility levels in the frequency range khz have mainly focused on frequency domain. When looking at what is available as standardized measurement methods being applicable for at least a part of -- the considered frequency range, there can be found EN [27], EN [25], EN Ed.3 [41] and EN [20]. The new EN [42] also covers the higher frequency range. However, it only specifies techniques of test equipment immunity to certain types of distortion. It does not require the user to perform measurements alongside the immunity tests and therefore does not contain or refer to any measurement methods. Table 8 shows a list of standards recently available concerning measurement the frequency range from 2 khz to 150 khz. Table 8: Comparison of today s standards regarding the frequency range from 2 khz to 150 khz Standard EN (Annex B) EN (Annex C) EN EN Purpose Method Frequency range in khz Application Emission levels from equipment, in the grid () Discrete Fourier transform (DFT), 200-ms-interval, aggregation into 200-Hz-bands per 200 ms Levels in grid DFT, 0.5-msintervals, aggregation into avg and max spectra per 10/12 cycles Levels in grid and emission from equipment Spectrum analyzer with tunable filter and detector (peak/ quasipeak/ rms) Emission from equipment Spectrum analyzer with tunable filter and detector Emission levels in the grid, Emission levels of equipment Emission levels in the grid Bandwidth 200 Hz 2 khz 200 Hz ± 100 Hz (-6 db) Scan times 200 ms 10/12 cycles of fundamental Parameter RMS values RMS values (Max and average) 14.1 s or 47 min 100 Hz (sloped) Peak values Status Informative annex Informative annex Normative Normative For power quality (PQ) purposes, in an Informative Annex C of EN [41], three options for measurement methods are provided. Following to that, concerning related CISPR standardization, EN indicates CISPR emphasizes immunity and emission measurements for equipment under test (EUT), and may not be optimized for in-situ power quality measurements. For the purpose of in-situ power quality investigations and surveys, the measurement methods in CISPR 16 may be complex or expensive to implement, due to their gapless measurements and accuracy requirements. The measurement methods of CISPR 16 may provide a large amount of data in an in-situ power quality context. However, the amount of data for in-situ measurements specified by CISPR 16 may be required for coordination with levels defined by various IEC standards.

30 Page 29 / 121 Some specific attention appears advisable to be paid to measurement parameters for smart meter immunity testing. As has been highlighted with investigations at TU Hamburg [43], the appropriate choice of frequency steps for such tests may be crucial for results providing serious information about the immunity of the EUT. As shown, the susceptibility of actual smart meter devices to emissions in the frequency range khz can be narrowband to an extent which makes it necessary to use a much finer frequency stepping than recently being considered [42]; with regard to the resulting increase of testing time, as an alternative, application of broadband test signals are proposed, already having been used in the standards IEC [44] and IEC [45], which are prescribed for smart meter devices. 6.3 Characteristics of measurement quantities and requirements Classification of higher frequency phenomena The characteristic of measurement quantities of different types of phenomena in the frequency range 2 khz to 150 khz can by classified according to their behaviour on short time scale (within a single fundamental cycle) and long time scale (occurrence over longer time periods, like hours or days). Regarding the short-time behaviour it has to be distinguished between Constant emission: (emission does not significantly change in magnitude and/or frequency within one fundamental cycle) (very rare) Varying emission level: The emission changes noticeable within one fundamental cycle. (e.g. PV inverters, PV chargers, many other types of electronic equipment) Transients The emission occurs only during a very short time of the fundamental period, like the spikes generated by a zero-crossing oscillation (e.g. EV chargers, CFLs) Transients can occur in combination with varying and constant emission levels. While time-domain analysis should be selected for transient phenomena, frequency-domain analysis is usually preferred for constant emission levels in order to assess the severity of their effects adequately. On long-time scale it has to be distinguished between (Quasi-) continuous occurrence The higher frequency emission appears over longer time intervals (multiple hours) Discontinuous occurrence The higher frequency emission appears over shorter time intervals (few milliseconds) Discontinuous phenomena may require sophisticated trigger options to be captured. This is usually not necessary for phenomena which appear for longer times. Here the pre-processing and storage capabilities of the measurement instrument are of higher importance. This classification is also supported by CISPR. Regarding discontinuous emissions, CISPR defines discontinuous disturbance as for counted clicks, disturbance with duration of less than 200 ms at the intermediate-frequency- (IF-) port of a measuring receiver, which causes a transient deflection on the meter of a measuring receiver in quasi-peak detection mode. When saying in item 7.2 Continuous radio-frequency disturbances are predominantly measured in terms of frequency domain parameters. Discontinuous disturbances are also measured in terms of frequency domain parameters but may need additional time domain measurements., for emissions in the frequency range below 150 khz, there may be anticipated, that with regard to the disturbing effects having been recognized with emissions in this frequency range, discontinuous emissions appear to need consideration with lower durations.

31 Page 30 / Types of application The choice of an adequate measurement method also depends on the purpose of the measurement. In terms of EMC three major types of application are distinguished: 1. Measuring emission levels in the grid: The main purpose of such measurements is e.g. the assessment of the present levels in the frequency range 2 khz to 150 khz or the investigation of interference cases. 2. Measuring individual equipment emission: These measurements are carried out to test compliance of devices with existing emission limits. This is already mandatory e.g. for lighting equipment. 3. Verification of equipment immunity: Immunity tests are performed to ensure that a device complies with existing immunity standards like IEC [42]. Even if the additional verification of the test level is normally not required by the standard, such measurements can ensure that the designated test levels are really applied to the EUT. Depending on the objective different requirements apply to the different aspects of the measurements. In order to ensure the comparability of results, at least within each of the above mentioned groups a comparable measurement method should be applied. With regard to the spectrum of measurement conditions being applied for measuring emission levels with and without EMI background, as reported in Study Reports I [1] and II [2] as well as in chapters 7 and 8 of this report, the results cannot be taken as exactly comparable but may fulfil the task of providing an overview of the existing situation General requirements For each of the applications a few general requirements need to be considered. Reproducibility / Comparability One major requirement is the comparability of measurement results. Measuring a certain measuring quantity should provide comparable measurement results or a clear and reproducible way of conversion needs to be specified. Accuracy The term accuracy is quite complex when it comes to the evaluation of spectra and is influenced by systematic and random errors. First, there is the accuracy of the measurement hardware, which includes all systematic errors. These are the constant and relative errors of an analog-digital converter, as well as the effect that the measurement hardware may have on the system it is connected to. Second, the signal is affected by noise, which includes all random errors. It is determined by the quality of the signal conditioning hardware, the used measurement range, the analyzed frequency range, the resolution of the analog-digital converter as well as the length of the measurement window that is used for the discrete Fourier transform if applicable. Signal filtering Since all measurements are limited in bandwidth, the signal must be bandwidth limited as well. Therefore, any measurement device must be equipped with a suitable anti-aliasing low-pass filter. This applies to any measurement method. Depending on the method, the use of other signal filters is advisable or required.

32 Page 31 / Summary It appears as advisable to establish a new consistent framework for measurement and assessment of voltages and currents in the frequency range khz [4], thus providing a unification of the relevant measurement parameters for ensuring comparability and appropriate accuracy of measurement and evaluation results without requiring extensive processing performance. for the appropriate setting of emission limits, compatibility levels and immunity requirements For establishing such, it appears recommendable (see also [22]) to consider different types of application of measurements like o measurement of emission levels in the lab as well as in situ o analysis of electromagnetic (EM) interaction of equipment / EMI cases o immunity testing to define the types of data needed to be considered to analyze the impact of measurement parameters on related results to make the application of a high-pass filter, either digital or analog, mandatory for sufficient suppression of the fundamental component and to avoid leakage, thus ensuring an adequate accuracy across the full frequency range to ensure application of a common bandwidth, where a value higher than 200 Hz could be beneficial o Higher frequency resolution provides more details on magnitudes o Maximum values should be used in case of time-varying emissions (PLC) to ensure o gapless measurement (should be used, if very short emission is expected) or o measurement with gaps not being synchronized to power frequency, for achieving a reliable representation of the signal characteristics to consider possible problems with spectra with sharp boundaries between individual bands to apply quadratic averaging (RMS values) for aggregation of several spectra over a longer time period The appropriate choice of a measurement method appears as highly dependent on the kind of higher frequency distortion that shall be measured and evaluated. This may mean that, finally, different measurement and/or evaluation methods could be appropriate to be defined for different types of disturbance issues in the grid. More detailed information is provided in Annexes B and C. Investigations on the specification of a measurement method for the frequency range khz having been already made on research level [29, 43, 46, 47, 48, 49] may provide support in finding an appropriate standardized solution. 7. Emissions. Measurement and test results 7.1 General Extending related information having been provided with [1, 2], this chapter provides additional results of measurements on different types of electrical devices without having been involved in EMI cases before. Related information and measurement results have been provided from Austria, France, Germany, Sweden, Switzerland and EMC ADCO 5. 5 EMC Administrative Co-operation Group

33 Page 32 / Measurement and test results Large photovoltaic inverter installations Solar campus Extensive measurements were carried out by experts from Austria, Germany and Italy at a 1 MW photovoltaic (PV) installation [50, 51] with focus on the analysis of harmonic emission both for classical harmonics below 2.5 khz and higher frequency emission in the range of khz. The installation of this solar campus consists of nine large inverters with rated power of 100 kva and eleven small inverters with rated power in the range between 1 kva and 10 kva (total output of nearly 50 kva), connected via a dedicated 1.25-MVA-transformer to the MV network. All large inverters are three phase while all small inverters are single phase. For R&D purposes, an MCS is installed. Fig. 15 shows the behaviour of the solar campus over 16 hours, over night. There can be recognized the signaling situation of the MCS (40 90 khz), showing higher, almost unattenuated levels during night, while inverters and their filters are out of operation. the emissions from inverters, in particular in the lowest part of the chart, at daylight and no emissions during night; the residual emissions being visible in this chart are stemming from other loads. Fig. 15: Time variation of emissions and signal attenuation in a Solar Campus, measured at the busbar Beside the measurement of total emission of all inverters, in one part of the measurement the inverters were switched off and on stepwise following a predefined schedule. This provides a comprehensive basis for a detailed characterization of the interactions between the inverters in terms of harmonic emission. Table 9 provides the times of switch-off of inverters, Tab. 9: Switch-off sequence of inverters Fig. 16 shows the response in terms of emissions measured from the systems (inverters, MCS) remaining in operation.

34 Page 33 / 121 Fig. 16 Spectrogram of emissions from a set of inverters switched off step by step Amongst others, it can be recognized, that the switch-off of small inverters at the end of the measurement results in some increase of emissions (see red circle), obviously due to reduced attenuation effects by filters built-in in the inverters LV grid with multiple PV inverters Different from the situation shown with Fig. 15 for the Solar Campus, Fig. 17 shows the emission situation of the external LV grid, where the influence of the inverters on the MCS signal is not to be recognized. Fig. 17: Time variation of emissions and signal attenuation in the external LV grid, measured at the busbar Results of additional measurement at a small-scale roof-top-installed PV system are shown in Fig. 18. Fig. 18: Time variation of emissions and signal attenuation in the external LV grid,

35 Page 34 / 121 It can be recognized that emissions from the inverter of the roof-pv disappear during night hours no attenuation effect on the MCS signal can be observed also during inverter operation hours there is no impact of the Solar Campus inverters, which are connected to the MV grid and electrically sufficiently far away situated Small PV inverter installations Measurements Measurements in Germany Measurements have been carried out at three PV inverters from different manufacturers with rated power around 5 kva in a laboratory environment. The spectra in Fig. 19 show that inverters differ in their switching frequencies (15-20 khz) and magnitude at switching frequency. Those frequencies are selected as good compromise between low switching losses and low audible noise. Fig. 19 Spectra of three PV inverters for home use These spectra were measured with sinusoidal supply voltage and no additional impedance network between the amplifier simulating the grid and the PV inverter. Further measurements have been carried out with different grid impedance networks in order to identify the source behaviour of the inverters. Fig. 20 presents the results, not showing a consistent behaviour of all three inverters in terms of switching frequency. Inverter A behaves for voltages up to 5 V nearly as current source, while Inverter C can be represented with good approximation as voltage source. Inverter B can be represented as voltage or current source with internal impedance, also called the PV inverters' source impedance. Subsequently a simple model as voltage or current source seems not to be sufficient and more information about the source impedance is required. See [47] for more details. Fig. 20 Impedance characteristic of inverters

36 Page 35 / 121 The source impedance can be identified by systematically changing the network impedance in a wide range. If the relationship between current and voltage emission at switching frequency is linear, an identification of the source voltage and the source impedance of the emission at switching frequency is possible and a model for the switching frequency emission can be developed. Fig. 21 exemplarily shows such a linear relationship for another small PV inverter D. The validation of the model using the reference impedance recommended in EN in normal and reverse operation is shown in Fig. 22. The error of the model is less than 10%. Fig. 21: Voltage and current values at switching frequency, if reactance of the network impedance is changed Fig. 22: Comparison between measured and predicted emission from inverter D for two network impedances Measurements in Sweden Solar panels are becoming more and more deployed and, as e.g. Swedish measurements on such ones have shown, have been discovered to figure as an EMI source. Fig. 23 shows the results of related Swedish measurements [52]: Peak values Momentary values Fig. 23: Emissions from a solar panel As can be seen, peak values of the emissions are again showing values above ~112 dbµv over a broad frequency range, from 9 khz to 150 khz, with max. values of around 128 dbµv around ~17 khz.

37 Page 36 / Emission survey Using an inverter, electricity being generated from sunlight by solar PV modules can be fed into the electrical supply of a building or directly into the public electricity supply network. Grid-connected solar panel systems intended to be used by consumers are already widely used throughout Europe and the market is growing continuously; that determined the EMC Administrative Co-operation Working Group to dedicate their 6 th EMC Market Surveillance Campaign, 2014, to assess the compliance of samples of related equipment taken from the market with the provisions of the EMCD. 14 national Market Surveillance Authorities (MSA) EMC ADCO members participated in the campaign (Austria, Cyprus, Finland, Germany, Ireland, Lithuania, Luxembourg, Malta, The Netherlands, Romania, Slovenia, Sweden, Switzerland and the United Kingdom) A total of 55 solar panel inverters mainly from EU (58 %) and China / Taiwan (35 %) origin --, all of them but one CE-marked, were selected and assessed between the 1st January 2014 and the 30th June products were assessed for emissions. Concerning emissions in the frequency range below 150 khz, the related EMC ADCO report [53] provides the following information: With regard to having become apparent, that emissions below 150 khz can be very high from some types of products and these can produce significant disturbance on the mains network, amongst others, this campaign was taken as an opportunity to gather information on emission levels in the range khz from a variety of manufacturers products, to increase under-standing of current emission levels. In the absence of an own product family standard for solar panel inverters, for the purposes of the campaign, EN [54] -- providing limits at mains terminals in the frequency range 9 khz khz for induction cooking devices, and therefore not applicable from a standard s point of view, but providing the test method (in Table 8) and limits that could be used for information gathering (and reported separately to avoid confusion) -- has been applied. 9 out of 14 MSAs took part at the study of emissions below 150 khz, testing overall 37 EUTs. The reduced number of participating MSAs was explained with regard to the increased challenge in performing this campaign because of the nature of the product and complexity of measurements / study of emissions below 150 khz (and of the DC side). From the results obtained of the solar panel inverters under test, the majority did not meet the harmonized standards as aforementioned, that would provide a presumption of conformity with the EMCD. Concerning khz, 37 EUTs were tested, and 14 out of them (38 %) were found compliant to EN [54] limits at mains terminals. These results of the optional study (below 150 khz (and DC side)) were recommended to be forwarded to the European Standardization bodies in order to be take into account at development of future standards for solar panel inverters. A similar campaign should be considered on the same basis after a certain period to assess the effect on the market Lamps with electronic ballast General Modern lighting technologies require a DC supply, which is provided either by internal or external electronic ballast. Depending on the design of the electronic ballast (with/without power factor correction; passive/active power factor correction) and the technology (compact fluorescent lamp (CFL)/LED) these lamps generate emissions in the frequency range above the harmonic area, between 2 khz and 150 khz Measurements Measurements in Germany A data base, called PANDA (equipment harmonic Database) and hosted by the Technical University of Dresden, Germany (TUD), primarily dedicated to the collection of measurement results on harmonic

38 Page 37 / 121 emissions from single phase household equipment with rated current below 16 A (magnitudes and phase angles of the voltage and current harmonic spectra of all kinds of electronic equipment up to at least the 40 th harmonic), provides also raw waveform data of 140 CFL and LED lamps (as per April 2014). This database is accessible via Internet ( [55]. The following contribution is based on the analysis of these data. CFLs require a resonance circuit to provide the high voltage for the discharge process. It usually operates at frequencies in the range between 20 khz and 60 khz. CFLs with rated power below 25 W normally use very simple rectifier circuits without PFC and depending on the efficiency of the implemented filter circuit a certain amount of the resulting emission is injected into the grid (example in Fig 24). CFLs with rated power above 25 W need to implement an active PFC (apfc) circuit in order to comply with the requirements for harmonic emission at frequencies below 2 khz. These circuits are based on high frequency switching and inject emissions in the frequency range khz into the grid (example in Fig 25). Figures 24 and 25 show the current waveform and time behaviour of such a CFL without and with an active power correction circuit (apfc). a) time domain b) frequency domain (200-Hz-bands, RMS) Fig. 24: Example of CFL without PFC circuit Current emission in time and frequency domain High-pass filtered waveform a) time domain b) frequency domain (200-Hz-bands, RMS) Fig. 25: Example of CFL with apfc circuit Current emission in time and frequency domain High-pass filtered waveform The waveform of the apfc CFL in Fig. 25a additionally shows a zero crossing oscillation, which is caused by the short interruption of HF switching of the apfc circuit around the zero crossing of the supply voltage. It is hidden in the spectrum and can be clearly identified by time domain view only. Electronic ballasts for LED lamps show a much higher diversity compared to CFL. It ranges from circuits as simple as a capacitive voltage divider with subsequent rectifier bridge to complex apfc circuits.

39 Page 38 / 121 Therefore also the emissions of those types of lamps varies significantly in terms of frequency and magnitude. As an example, Fig. 26 shows the distortion of a LED lamp bought in Germany and measured in the laboratory at TUD. Its current emissions amount to about 90dBµA at ~53.5 khz and still ~88 dbµa at ~107 khz. Fig. 26: Example of an LED lamp Current emission in time and frequency domain High-pass filtered waveform It may be mentioned that the tested lamp, being CE marked, does not comply with the limits given by CISPR 15 [56]. Summarizing, there can be said: Measurements in frequency domain and comparison of results between CFLs without and with apfc show o concentration of more or less higher emission levels to the frequency ranges of ~ 2 9 khz and ~30 40 khz in general o emissions, with current levels up to ~ 80 dbµa in both cases. While in frequency domain, some similar behaviour of CFLs without and with apfc can be seen, in time domain some differences can be recognized: o Additional zero-crossing oscillation for apfc CFLs, caused by the short interruption of HF switching of the apfc circuit around the zero crossing of the supply voltage o By far broader time ranges of higher current emissions at CFLs with apfc, compared with such from CFLs without For all measured lamps Fig. 27 provides an overview of switching frequencies and levels of related emissions higher than 60 dbµa (11% of the lamps (CFL, LED)). Fig. 27: Switching frequencies and emission levels for different lamps with electronic ballast (emission levels calculated as 200-Hz-bands, RMS)

40 Page 39 / Measurements in Sweden Measurements having been conducted at the Swedish Lulea University of Technology, Skellefteå [122], with support from Skellefteå Kraft Elnät, Elforsk and the Swedish Energy Administration, were dedicated to the distortion of the supply voltage caused by fluorescent lamps with high-frequency (HF) ballast, being used, amongst others, for their higher energy efficiency and longer lifetime compared with incandescent lamps. The 58-W-lamps were equipped with a HF-ballast with a voltage range of 230 V to 250 V and a nominal current between 0.33 A and 0.29 A and a power factor of In line with the recognitions highlighted in this Study Report, the emissions from the HF ballasts figuring as a number of different HF components recurrent transients, narrow-band time-invariant signals, time dependent signals are discussed not only in frequency domain, but also in time domain and in time / frequency domain. Summarizing the results of the measurements with one lamp shows: When considering classical harmonics up to 2 khz only, the current drawn by fluorescent lamps with HF-ballast shows a low total harmonic distortion (THD). However, this low distortion at lower frequencies, commonly obtained by means of apfc circuit using e.g. variable switching frequency strategies, goes at the expense of higher frequency components in the current. In time domain, the current drawn by the lamp shows kind of notches around its zero crossing, also HF ripple as well as HF ripple around the current maximum With applying a (2 nd order Butterworth) filter, for getting a better view of the current, the notches show up as recurrent oscillations every 10 ms on the current waveform and the HF ripple is visible as well Fig. 28 shows the current drawn by the lamp without and with filtering a) without filtering b) with filtering Fig. 28: Current drawn by the lamp, measured in time domain in frequency domain, the spectrum (in 200-Hz-bands) of the current drawn by the lamp shows two main emissions above 20 khz, around 28 khz and upwards from 41 khz (Fig. 29). Fig. 29: Current drawn by the lamp, measured in frequency domain

41 Page 40 / 121 Appearing also in the voltage spectrum when the lamp is in operation but disappearing when being switched off, these emissions give rise to the assumption that they are generated by the HF-ballast. in time / frequency domain, spectrograms of voltage and current show (Fig. 30) o many of the higher frequencies being synchronized with the fundamental frequency o o o around 28 khz an almost constant frequency, with only minor changes in amplitude for the emissions around 41 khz an arc-like function, probably due to variable switching frequencies the aforementioned notches as small lines in the lower frequency range between the variable switching frequencies, probably being the background for the above-mentioned HF ripple. a) filtered voltage b) filtered current Fig. 30: Time / Frequency domain a completely different behaviour in the frequency ranges 2 9 khz and khz, with o high oscillations occurring twice per cycle and lesser oscillations four times per cycle o a number of narrow-band distortions as well as a broadband distortion being present as well as broadband distortions starting at ~41 khz Measurements on an increasing number of lamps, from one to nine, showed a different THD behaviour for 2 9 khz on the one side and for khz on the other side, obviously due to different mechanisms in the two frequency ranges being responsible for the distortion. While for 2 9 khz the distortion of current increases somewhat less than linearly with the number of lamps it remains about constant in the frequency range khz. no significant impact of the number of lamps on the current waveform other than that the amplitude of the recurrent oscillations increases and the HF noise between the notches is relatively higher with a lower number of lamps Summary With regard to the different behaviour of CFL emissions having been recognized in the frequency ranges 2 9 khz and khz the just growing awareness of the importance of time-domain analysis and time-frequencyconsideration the different behaviour of CFLs without and with apfc, being recognizable in time domain in particular the differences in emission behaviour between CFLs and LEDs

42 Page 41 / 121 further investigations should be made including simulations and additional measurements to get a better understanding of the origin of the high-frequency distortion both for individual lamps and for the interaction between multiple lamps Electric vehicle charging EVs are a promising technology for remarkably reducing the environmental burden of road transport. They have potential for significant contributions towards achieving the climate protection goals in the transport sector, therefore the massive EV roll out will have a significant impact on the DNO s. Therefore, DNOs should be aware of the impact that EVs could have on their electrical grid. Besides PV inverters, also charging units for BEV or plug-in hybrid ones, using technologies with switched power electronics like apfc or pulse-width-modulation- (PWM-) controlled rectifiers are generators of emissions in the frequency range khz. Therefore, analysis of the higher frequency behaviour of EV chargers is needed, e.g. how the emission is generated and how the emission of multiple devices adds. Concerning standardization, the development of standards for EV charging shows up until now, that a need for consideration of emissions in the frequency range khz is not seen in general. Results of investigations in Germany (TUD), France (ERDF) and Switzerland (ewz and TUD) provide an overview of related emissions Measurements in Germany Results of measurements performed by TUD [21] provide a brief overview of emissions in the frequency range between 2 khz and 150 khz caused by six BEVs with onboard charger. The highest emission levels at switching frequency of the rectifier circuit usually occur during the time when the highest charging current is drawn by the EV. Fig. 31 presents the spectra of 4 selected EVs measured in the lab. Fig. 31: Current spectra of 4 BEV chargers (200-Hz-bands, RMS) Up until now more than 19 different EV types (what covers more than 80% of the current German market) were measured under laboratory conditions and in real LV grids. Some of these measurements are also available in PANDA (see also ). In order to cover the whole energy emitted by the BEV chargers at their respective switching frequencies, the results of the FFT are aggregated into 800-Hz-bands and presented in a magnitude vs. switching frequency plot (Fig. 32)

43 Page 42 / 121 Fig. 32: Switching frequencies and emission levels for different EVs (800-Hz-bands, RMS) The switching frequencies range from ~ 8 khz up to ~100 khz. Due to the different impedance conditions, emission levels measured in the lab are in some cases higher than the levels measured with the same EV when connected to the real grid. While about five of the EV types did not show any significant emission above 60 dbµa in the frequency range 2 khz khz, other EV types emitted up to 114 dbµa (about 0.5 A) when connected to the grid. In the laboratory, emission levels of up to 125 dbµa (about 1.7 A) at switching frequency were observed. Some of the EVs with switching frequencies below 15 khz caused significant audible noise, which becomes annoying if several EVs of the same type are being charged at the same time. Further analysis of the measurement data shows that emission levels and switching frequencies of the EVs also depend on the electrical design of the charger, the charging state and the network impedance at the connection point Measurements in France Measurements on disturbances during the charging process of EVs have been performed by ERDF [57]. Two different types of private EVs, as being intended for future mass roll-out in the upcoming years, have been investigated in a lab environment with an artificial mains network (AMN) according to CISPR-16. The measurements have been performed with an Anritsu Spectrum Analyzer, with the following setup: RBW: 300 Hz VBW: 100 Hz EV Type 1: Reference Level Offset: -22 db (to compensate the 22 db loss caused by the attenuator (20 db) and the coupling effect (2 db)) Center Frequency: 79.5 khz Span: 141 khz Start Frequency: 9 khz Stop Frequency: 150 khz Spectrum analysis showed the following pattern:

44 Page 43 / 121 Fig. 33: EV1 charging pattern (Yellow) / Not charging pattern (Green) There can be recognized a wide-band disturbance due to emissions from EV1 between 9 khz and 65 khz, with peaks ranging from 105 dbµv at 30 khz to dbµv at 61.6 khz, not any disturbing emissions from EV 1 on from ~74 khz obviously, the switching frequency of the charging unit is 61 khz in the case of the highest peak, at the switching frequency 61 khz, an increase of the disturbance level, caused by charging EV 1, by ~58 db EV Type 2: From the pattern being showed by spectrum analysis (Fig. 34) Fig. 34: EV2 charging pattern (Yellow) / Not charging pattern (Green) there can be recognized a wide-band disturbance due to emissions from EV2 between 9 khz and 100 khz, with peaks ranging from ~81 dbµv at ~30 khz to 90.1 ~dbµv at ~100 khz

45 Page 44 / 121 not any disturbing emission from EV 2 on from 100 khz the frequency band > 100 khz the charging unit s switching frequency is 100 khz in the case of the highest peak, at the switching frequency 100 khz, an increase of the disturbance level, caused by charging EV 2, by ~27 db Measurements in Switzerland For their EV fleet, a Swiss municipal utility operates a central charging infrastructure with 15 charging points, including a three-phase off-board charger for DC charging of one EV type at 18 kw. While the measurements of the emission of the on-board chargers are already included in the results of item , here, the off-board charger is addressed. The high-pass filtered current and voltage waveforms in phase L1 for one fundamental period (time instant within the interval of highest charging current) are shown in Fig. 35. The content of emissions in the frequency range khz varies significantly in magnitude. a) current b) voltage Fig. 35 High-pass filtered current and voltage waveforms of the BEV charging station Current maxima of up to 3 Apk result in voltage maxima of up to 5 Vpk. Fig. 36 presents the corresponding spectra. At the off-board charger s switching frequency of 45 khz the current amounts to 120 dbµa, which causes a voltage of almost 125 dbµv at the connection point of the charger [58] Fig. 36 High-pass filtered current and voltage spectra of the BEV charging station (200-Hz-bands, RMS)

46 Page 45 / Measurements in Austria Measurements have been carried out on an EV, with application of frequency as well as time domain, using a national Instrument Data Acquisition Unit (NI USB-6251) with three channels, each one sampled with 333 khz with applying a high-pass filter (2 khz). Fig. 37 shows the measurement results in frequency and in time domain. Fig. 37: Measurement results of EV charging in time domain vs. frequency domain Fig. 37a shows the charging currents (RMS) and resulting voltages at the connection point of the charging plug. Fig. 37b and c show the voltages in time domain (over 50 ms), where Fig. 37c shows details over a time of ~ 4 ms. Fig. 37d shows the spectrum of the emissions calculated from the values shown in Fig. 37b, representing aggregated 200-Hz-bands. It can be recognized a maximum value at 10 khz (switching frequency of the charger) (Fig. 37d), resulting in the visible period of 100 µs of the oscillations shown in Fig. 59c a superimposed oscillation with 1.6 khz, visible in Fig. 59c (violet) with a corresponding maximum value in Fig. 37d. It may be noted, that this particular EV charger was of same type as BEV1 in Fig. 31 (item ) Frequency-controlled heat pump Swedish measurements of an inverter-controlled air/water 6 kw heat pump indoor unit [59] showed emission levels according to Fig. 38:

47 Page 46 / 121 Peak values Momentary values Fig. 38: Emissions from an inverter-controlled heat pump The chart shows emission peak levels above 100 dbµv for just the whole frequency range 9 khz to 150 khz for the frequency range ~18 khz to ~27 khz emission peak levels exceeding 130 dbµv for the frequency range ~38 khz to ~48 khz emission peak levels still being close to the limits for intended MCS signals [20] Cola spender Measurements on a Cola spender [118], having been made due to interferences, resulted in finding this device being the EMI source due to higher disturbance levels (Fig. 39), Fig. 39: Emissions from a Cola spender showing a quite variable situation of emission levels, with several peaks across the frequency range, with values up to ~110 dbµv at frequencies of around 43 khz, 48 khz, 71 khz, 85 khz. Installation of a single-phase filter led to acceptable disturbance levels.

48 Page 47 / DVD player The emissions shown in Fig. 40 might be assumed as such from an MCS. Indeed, these emissions have been measured in Sweden at a DVD player [119], showing once more emission levels above 100 khz over nearly the whole frequency range below 150 khz, reaching up to 125 khz two distinct peak areas, with levels up to 115 dbµv (momentary) at ~79 khz and (momentary) at ~145 khz a) b) peak momentary Fig. 40: Emissions from a DVD player before (a)) and after (b)) installing a single-phase filter (35 khz - 95 khz) Besides the quite high peak emission values over the whole frequency range it may be noted, that the value at 150 khz appears as being remarkably higher than the maximum value specified in EN [60] for this frequency (see also item ) TV box Measurement on a TV box [61] showed the following emission levels: With emission levels generally above 120 dbµv reaching values close and even above the signal limits for MCS in a broader frequency range of ~35 khz to ~95 khz the obtained measurement results may explain the started symptoms of arriving at the end of the device s lifecycle, therefore of the occurrence of a defect show a remarkable potential for EMI to other electrical equipment, also and in particular to MCS Figure 41: Emissions from a TV box

49 Page 48 / Beer cooler Measurements on a beer cooler, carried out in Sweden [62], showed, that even new types of such devices may generate remarkable emissions (Fig. 42a), with levels up to ~114 dbµv at ~66 khz. The emission situation after switching off the cooler is shown in Fig. 42b. a) b) Figure 42: Emissions from a beer cooler in operation (a)) and after switch-off (b)) Traveling circuses As measurements in Sweden [63] show, also mobile emission sources, e.g. travelling circuses, can be the background for the temporary occurrence of high emission levels in the frequency range khz and, following to that, of EMI. This kind of problem is not permanent, and after waiting a few days, the disturbance is definitely gone. Figure 43 shows the emission levels generated by such a traveling circus equipment, measured on one phase: Peak values Average values Momentary values Fig. 43 Emissions from a travelling circus equipment The measured momentary values of related emissions reach levels above 111 dbµv in a frequency range ~61 ~81 khz and up to ~127 dbµv at frequencies around 68 khz. So, when dealing with EMI and facing some uncertainty about the origin of high emission levels, the possibility of mobile disturbance sources should be taken into account Power supply to fiber switches In Study Report II [2], with item , an EMI case was reported from Sweden, where a new fiber switch caused loss of communication of 89 electricity meters (~94 %) out of 94 ones within an AMR-PLC system around a concentrator.

50 Page 49 / 121 With another report from Sweden [64], further attention is drawn to fiber switches, whose built-in power supplies may represent a more or less EMI source, thus jeopardizing proper communication within an AMR-PLC system from the meters for collecting data, when the EMI source being near a related data concentrator. Fig. 44 shows emission levels having been measured at such fiber switches with built-in power supplies, with high levels frequencies up to ~95 khz peak values at 23 khz, 37 khz, 72 khz and 96 khz highest peak levels of ~124 dbµv at ~37 khz and ~122 dbµv at ~72 khz. Fig. 44: Emissions from a fiber switch with built-in power supply It is pointed to the fact, that a growing number of energy companies are looking to strengthen their substation infrastructure using fiber optics. In cases where fiber switches with built-in power supply are used as aforementioned, a kind of homegrown EMI situation may be the result for the utility. For mitigation, the application of filters was recommended Power supply to a network router Fig. 45 shows the emissions of a power supply to a network router, which Swedish measurements [65] showed it generating quite high emissions: a) b) Fig. 45: Emissions from a power supply to a network router before (a)) and after (b)) replacement This emission situation might have developed due to ageing effects within the somehow older power supply. Besides the quite high peak emission values over the whole frequency range it may be noted, that the emission at 150 khz appears as being remarkably higher than the maximum value specified in EN [60] for this frequency Plugin charger Measurement of a plugin charger [66] resulted in emission levels as shown in Fig. 46:

51 Page 50 / 121 peak momentary Fig. 46: Emissions from a plugin charger Higher emission levels can be recognized on from ~3 khz, with a first peak of 120 dbµv at ~27 khz and levels increasing from ~55 khz to 150 khz, from ~95 dbµv to ~122 dbµv Single-phase PSU pretending a 3-phase problem At first sight, measurements on all three phases of an installation [67] appeared to show an EMC problem being caused by a three-phase EMI source. Fig. 47 shows the emissions measured on the three phases. The spectra are showing max. values up to 128 dbµv on all three phase the max. values appearing on different frequency positions across the phases o at ~72, ~86 and ~102 khz on L1 o o at ~20, ~40, ~60 and ~72 khz on L2 at ~23, ~45, ~67 and ~89 khz on L3 Fig. 47: Emissions from a PSU, pretending a three-phase EMC problem After closer analysis it turned out, that it was a simple single-phase PSU which gave the appearance of a major three-phase problem. Replacing the PSU solved the problem.

52 Page 51 / Power supply to a PLC modem Fig. 48 shows the results of Swedish measurements of emissions from the PSU of a PLC modem [68]. Peak average momentary Fig. 48: Emissions from a PLC modem From this result, showing large frequency ranges with emission levels above 100 dbµv, below ~43 khz and from ~60 khz to ~95 khz, it was anticipated, that this power supply were starting to give up Microwave oven Swedish measurements on a microwave oven [69] resulted in emissions as shown in Fig. 49. peak average momentary Fig. 49: Emissions from a microwave oven With regard to the measured emissions behaviour, with increasing levels over frequency relatively high peak and momentary values showing these emissions when the device only being plugged in, the related manufacturer has been contacted.

53 Page 52 / Desktop power supply with apfc In order to reduce the harmonic emission below 2 khz and to increase their energy efficiency, modern computer power supplies commonly utilize active PFC (apfc) circuits. This results in emissions at the switching frequency of the electronic switch of the active PFC circuit, which is normally several ten khz. The following measurement results have been obtained within TUD. Fig. 50 presents the high-pass filtered waveform for one fundamental period. In the current emission spectrum (Fig. 51), the switching frequency can be clearly identified with 66 khz, with a magnitude of ~100 dbµa. The interruption of the high frequency switching close to the zero crossing, which results in a slight further increase of energy-efficiency, is clearly visible, but doesn t lead to a recognizable zero-crossing oscillation, like it is shown for the CFL with apfc reported in , Fig. 25a. Fig. 50 High-pass filtered current waveform for one cycle of the fundamental Fig. 51 Current emission spectrum (200-Hz-bands, RMS) 7.3 Summary Beyond larger / commercial equipment & systems using power electronic technology, a lot of different types of electronic mass market equipment (e.g. PSUs, lamps, EV chargers) showing relevant emission levels in the considered frequency range can be recognized. It appears as not being sufficient to consider the emission behaviour per type of equipment, but as being necessary to consider the rated power, technology, brand, possible ageing effects. Examples of equipment are presented which have relevant emissions, thus reaching peak values of more than 130 dbµv. But there are also devices on the market with almost no relevant emissions.

54 Page 53 / 121 To some extent, conducted emissions having been measured show quite high peak levels, while there are also devices with almost no relevant emissions on the market. Emission measurement results have been obtained with different measurement methods. It results that there is a need for a unified measurement method particularly considering the needs of khz In many cases, measurement results obtained in frequency domain do not comprehensively mirror the emission behaviour of a device, thus not providing safe evaluation bases for possible EMI. Therefore, increased application of time domain measurement is recommended when performing measurements in khz. Further investigations should be made o on emissions in the considered frequency range in general o on the specific emission behaviour of equipment dependant on the applied technology 8. EMI cases. Measurement and test results Based on investigation results provided from Finland, France, Germany, Spain and Sweden, also considering some new data from Japan, following editions 1 and 2 of the Study Report [1, 2], this chapter provides additional examples of EMI, having been identified with electrical equipment in the frequency range khz. 8.1 General As already highlighted in Study Report II [2], not appearing as expectable from first sight, beside EMI due to conducted emissions, also such due to radiated emissions and hidden effects of EMI in the form of ageing and lifetime reduction need to be considered (Fig. 52). Fig. 52: Types of EMI effects in the frequency range khz In the following, additional examples for EMI having been investigated are provided, items & 2 being dedicated to EMI cases to be recognized by more or less spontaneous effects.like malfunction of affected equipment item being dedicated to hidden effects becoming recognizable only after some time

55 Page 54 / EMI due to conducted emissions EMI to NCE Residual current devices Following the first two editions of SC 205A Study Reports [1, 2] some interest has been expressed in whether residual current devices (RCD) would show susceptibility to emissions in the frequency range khz. There was found one information about an immunity test carried out with a 30-mA-RCD. When applying disturbances (up to 1 %) at frequencies between 2 khz and 9 khz, the EUT showed mechanical vibrations, those ones increasing with the disturbance level, which could be clearly heard and resulted in a damage of the EUT. Beyond that test result, related investigations resulted in no further reports about such EMI, but feedback from RCD manufacturers about tests performed with RCDs in the frequency range khz, concerning common-mode disturbances, in general the awareness of related EMC challenges and measures having been taken to avoid EMI from emitters in the considered frequency range, thus to better cope with the changed EMC environment efforts made in the recent past on digital technology level to ensure sufficient immunity of RCDs to emissions in the frequency range khz CNC mill disturbing household and commercial equipment Investigations have been made by TUD [47] in 2012 after several network users in an urban grid having complained to their network operator about malfunction of their electrical devices. These complaints related to the following problems: Household: Malfunction of a fully automated coffee maker, making it unusable during working hours Commercial: Malfunction of a hairdryer control, with autonomous switch-on and switch-off times a day Industry: Periodic malfunction of the control of a new CNC milling machine in a cogwheel factory, causing destruction of milling heads Figure 53 shows the related network scheme with points of customers, complaints and measurements. Fig. 53: Network scheme with locations of customers, complaints and measurement After receipt of the complaints, the DNO performed measurements. The requirements according to EN [70] were fully met. Then the network operator asked TUD to cooperate in the investigation to find the source of the disturbance.

56 Page 55 / 121 A preliminary measurement was carried out to see, whether the cause might be at a higher frequency than those ones covered by EN Fig. 54 shows the measured voltage spectrum at the end of the feeder with most complaints (site c). Fig. 54: Spectrum of voltage at site c (200 Hz bands, measurement) For a single 200-Hz-band around 8 khz, the spectrum shows high voltages up to an RMS value of 3 V ( 1.3%) unequally distributed among the three phases. It can be clearly seen that individual 200-Hzbands do not cover the whole energy emitted at switching frequency. Therefore an 800-Hz-band (as it is also suitable for PV inverter emission) should be used. The levels at 8 khz were higher than those of any harmonic at this location. These measurements were carried out for several days. The results show that the distortion was present only Monday through Friday from 6 a.m. to 10 p.m.. This suggested a source with industrial use. Through further investigation the malfunctioning CNC mill itself was identified as the cause of the disturbances. It was commissioned without an EMC filter. As a first mitigation measure, the factory feeder was relocated to another part of the grid. That solved the problem for the complaining customers I (coffee maker) and II (hairdresser s saloon), but not for the CNC mill itself. Then the commissioning company for the CNC mill installed a generic EMC filter, which reduced the disturbance by a factor of 3 to 4. However, it did not yet provide a solution for the machine malfunctions. That what was finally achieved by installation of a manufacturer specific filter, which reduced the disturbance level again by a factor of 2; this is shown in Fig. 55. Fig. 55: Inverter emissions measured at the end of the CNC mill feeder (site c) with / without filter installed, four spectral components around 8 khz combined to a single 800-Hz-band, boxes mark 5% and 95% quantiles and line marks median, only times with CNC mill in operation are included

57 Page 56 / Ceiling lighting disturbing studio equipment High-level disturbances from dimmable ceiling lighting equipment in a film- and sound studio having, with values up to 114 dbµv at around 70 khz have been recognized as the origin of EMI. Fig.56 shows the measured disturbance levels [71], which caused damage to control units in the studio and a breakdown of circuit boards of mixing desks, having been protected by active filters, several times, during recording. Fig. 56: Emissions from a dimmable ceiling lighting equipment CNC machine This EMI case [72] is an example for EMI from one NCE an automatic lathe -- to another NCE an induction cooker --, without degradation of performance of functionality of the latter, but generating audible noise at it, why the related network user complained to the DNO. A network user In the supply area of the German DNO Energiedienst Netze GmbH, now ED Netze GmbH (EDN), contacted the utility whether disturbing noise at his induction cooker could have its origin in the public electricity supply. After several complaints some ten years ago, about EMI with control equipment of heating systems, washing machines and coffee makers and no distinct disturbance source having been possible to be recognized, at a guess, the utility had undertaken some enhancement of the local supply network around a cabinet making connected in the same supply area; after this enhancement, first, no further complaints had been made to the DNO. Now, a check of the harmonic levels in the supply voltage resulted in values meeting the PQ values of EN [70] up to the 25 th harmonic as well as being below the compatibility levels for harmonics up to the 50 th harmonic, according to EN [73] and did not allow for a possible recognition of a disturbance source. Measurement of the noise level directly at the induction cooker, in the audible frequency range up to 20 khz, showed peak levels at 5 khz, 10 khz and 20 khz. Fig. 57: Noise level at an induction cooker

58 Page 57 / 121 After considering the related grid scheme, once more the industrial enterprise of the cabinet making got investigated in more detail, where 10 CNC automatic lathes were in operation. After changing the point of connection of the cabinet making, the noise at the induction cooker was not further audible, thus proving that one out of the 10 automatic leathes would be the EMI source. Selective switch-off of one of the automatic leathes each together with application of a network analyzer enabled the evaluation of its emissions, showing only one of these automatic leathes generating remarkable emissions above 2.5 khz. After a bridge rectifier circuit at the voltage input of the automatic leathes, rectifying the three phase voltages to DC the supply voltage of the automatic leathes results from a pulse width modulation (PWM) with sine evaluation, which fragments the DC voltage with a certain pulse frequency, towards pulses with a certain pulse-pause-time percentage. The pulse frequency of the pulse inverter and the related side bands were to be recognized in the measured currents and supply voltages. Fig. 58: Spectrum FFT-voltage DC to 20, measured at the connection point of the automatic leathe For mitigation, between the two options of filtering to block the disturbance at the point of connection (POC) of the induction cooker, with an active network filter, to filter the disturbing emission directly at the automatic leathe, following to a related utility s initiative, after coordination between the industrial enterprise and the manufacturer of the automatic leathes, a network filter was installed. As measurement then showed, while remarkably reducing the disturbance levels on from 10 khz, the impact of this filter on the disturbance levels around 5 khz was insignificant. Essentially, the noise emission from the induction cooker was mainly attributed to frequencies around 5 khz, that resulting in the perpetuation of the disturbance in principle. Additional measures like modification of the installed network filter or additional installation of a trap circuit at the induction cooker have been considered.

59 Page 58 / Power supply to a TV box In Sweden, a power adapter for a Digital TV box with built-in HD has been found as an EMI source on many occasions. After having caused EMI and, finally, having stopped working, measurement on such an external power supply [120] showed an emission spectrum as can be seen in Fig. 59 (see also item 7.2.8): Fig. 59: Emissions from a power adapter to a TV box This adapter showed high emissions over the whole frequency range khz peak values up to ~125 dbµv at ~46 khz and ~50 khz as well as ~122 dbµv at ~86 khz and ~121 dbµv at ~96 khz particularly high emissions in the frequency range 30 khz to 95 khz, i.e. the main frequency area where MCS are operated, with more than 112 dbµv LED Construction light In 2014, an EMI case was observed in Sweden, where 48 out of total 270 electricity meters connected to a substation lost communication with their AMR-PLC system [90]. As the EMI source a construction light with light emitting diodes (LED) was found, as it is publicly sold to customers. Several devices were found on the worksite, having been bought at the same time. All lamps showed equal EMI behaviour. The emission spectrum of these lamps is shown in Fig. 60. Fig. 60: Emissions from an LED construction light As can be seen, the spectrum shows broader ranges of emission levels close to the limits of MCS signalswith exceeding 110 dbµv in the ranges from ~31 khz to ~52 khz, ~72 khz to 84 khz and ~109 to ~130 khz reaching max. values of ~136 dbµv at ~38 khz and ~132 dbµv at ~ 74 khz and ~147 khz on from ~31 khz, the emissions of these CE-marked lamps were quite above the limits of EN [56], violating these limits by up to ~44 db. After completion of the analysis and dialogue with the owner, the lamps were disconnected and returned to the store by the owner. On from disconnection, all meters communicated with their AMR system again.

60 Page 59 / EMI with solid state electricity meter (accuracy) due to PV inverter In Germany, a static electricity meter was discovered to register wrong energy values in 2011 when a single-phase, transformer-less PV inverter with 5 kw output power was operating on phase L3. The EMI case occurred in a residential installation and was reported to the utility from the network user, as he noticed remarkable differences to the energy reading of an electro-mechanical meter in the same installation. On-site measurements next to the meter using an oscilloscope showed significant high frequency current emissions on phase L3 at about 20 khz, as shown in Fig. 61. The measured RMS values of the current and voltage were 1.4 A and 230 V, the components at 20 khz were 0.5 A and 0.7 V, respectively; that resulting in THD values of 39 % for current and 0.3 % for voltage [74]. The origin of these conducted emissions and the EMI case was clearly identified to be the PV inverter. At these conditions, the power at mains frequency generated by the inverter was only about 300 W due to cloudy atmospheric conditions. Measurements in other PV installations showed that high frequency current emissions from PV inverters are not strongly correlated to the amount of solar radiation and can be significant also in stand-by mode. Fig. 61: Signal current waveforms on phase L3 connected to a single phase photovoltaic inverter, showing significant high frequency current emissions at about 20 khz This static electricity meter showed large errors in energy measurement, clearly exceeding the accuracy class of the meter. The issue was resolved by replacing the static electricity meter with another type, which was immune against these high frequency current emissions. CLC/TR [75] (see also item 9.3), having been published later on in 2012, requires a significantly higher immunity level than the level of conducted emissions in this installation. Meters fulfilling CLC/TR would therefore correctly operate in this installation. Another possible EMI to electricity meters can also cause the display of wrong meterng values due to additional zero crossings of the supply voltage; that with regrad to some types of metering equipment synchronizing the sample rate with the power frequency, by considering the zero-crossings Chirping noise in an office building Measurements were carried out in Germany by a local DNO after receiving complaints from customers in an office building about chirping noise coming from the lighting installation and from active loudspeakers. Measurements showed a pulsing emission at 4 khz with maximum levels of 122 dbµv (RMS) [76]. Related waveforms are shown in Fig. 62.

61 Page 60 / 121 Voltage Current Minimum Maximum Fig. 62: Current and voltage waveform at minimum and maximum levels of the disturbing emission As origin of the EMI two newly installed 500 kva uninterruptable power supply (UPS) systems could be identified, whose interaction generated a disturbing emission at 4 khz. As a quick mitigation measure, the supply of the complaining customers was separated from the UPS systems. Being aware of the problem, the UPS manufacturer has promised to provide a suitable filter. Due to the lack of compatibility levels, the discussion with manufacturer and operator of the UPS systems and the final conviction for installing the additional filter was difficult. It should be noted that the aforementioned problems are observed in many other EMI cases with only NCE being involved, what underlines the urgent need for a comprehensive standardization framework EMI with MCS Power supply to a TV amplifier Following to communication problems between AMR-PLC and data concentrator, measurements have been made in the suburb of Lyon, France [77]. Fig. 63 shows the related situation of the network. Fig. 63: Network situation at measurements in Caluire et Cuire

62 Page 61 / 121 From an MV/LV substation with one transformer (apparent power = 630 kva), 8 LV feeders were feeding residential installations in three buildings (A C), with 30 AMR-PLC devices being installed in building B. Having been discovered by the data concentrator during the installation phase, afterwards communication failed completely. As the EMI source, the amplifier of a collective TV antenna has been found. For investigating the problem, measurements of the supply voltage spectrum have been made (Max hold measurement results see Fig. 64) with the smart metering system disabled and the TV antenna amplifier a) being switched off, for measuring the noise at different locations in the building (lower. green curve) b) being enabled.(upper, yellow curve) Fig. 64: Max holds measurement of supply voltage spectrum Markers 1 & 2 show the channel frequencies used in the applied S-FSK PLC system (63,3 & 74 khz) Markers 3, 4, 5 & 6 show the peaks of the disturbing emissions generated by the TV amplifier. As can be seen, two of the several peaks of the TV amplifier s emission are close to (Marker 1) or identical with (Marker 2) the operating channels of the applied PLC, system, with an emission level there of around 132 dbµv the afore-mentioned peaks (Markers 1 & 2) of noise floor + TV amplifier are above/close to the signal limit for MCS [20]. After replacing the amplifier of the collective TV antenna, measurements having been performed like before showed a reduction of the output noise level of the TV amplifier by some 40 db, with that reduction of the overall disturbing emission levels quite below the signal emission limits of EN [20] allowing for proper operation of the MCS and the related data collection again TV amplifier Elevator drive PC screen Lab and field measurements have been made by ERDF [78] for analyzing the impact of higher level emissions from NCE on the proper function of AMR-PLC, also for finding criteria for the needed margin between signal and noise levels; that related to MCS S-FSK PLC technology used by ERDF, operating with two carriers, at 63.3 khz and 74 khz.

63 Page 62 / 121 Figures show the measured emissions from (the power supply of) a TV amplifier, a variable speed drive of an elevator and of (the power supply of) a PC screen. Fig 65: TV amplifier (up to ~ 110 dbµv noise floor) Fig 66: Variable speed drive of an elevator (80 dbµv noise floor with peaks 92.1 dbµv (48.3 khz, dbµv (65 khz), dbµv (72.6 khz) Fig 67: PC screen (100 dbµv noise floor with peak dbµv at 65.7 khz) In each of these EMI cases, the PLC communication links were broken (FER = 100%, see chapter 4., Fig. 5). Given the attenuated PLC signal level at the receiver (after traversing the grid between TX and RX) and the high noise level at 72,6 khz ( dbµv), the binary data carried at 74 khz cannot be demodulated properly. Yet, in case one carrier is obscured by noise, the S-FSK PLC technology still allows to work in ASK (Amplitude Shift Keying) mode, using exclusively the remaining carrier (the carrier s presence corresponds to a binary 1 and its absence to a binary 0, or conversely). However, the moderate noise peak centered at 65 khz (85.96 dbµv) still does not allow proper PLC operation using the second carrier (63.3 khz) only. This behaviour is explained by the attenuation encountered on the grid between TX and RX resulting in a too low signal-to-noise ratio at the receiver. For the occurrence of EMI in case of MCS involvement, the following can be concluded, also for the appropriate setting of compatibility levels: Dependence on the network impedance situation between MCS transmitter and receiver and following to that on the attenuation of the MCS signal, in coincidence with NCE emission levels Dependence on the impedance behaviour of the installation and the degree of decrease of NIE till the metering point

64 Page 63 / 121 Dependence on the signalling structure of the MCS as well as on the frequency situation of highlevel NCE emissions Higher NCE emission levels together with higher attenuation between TX and RX of the MCS would not allow proper PLC operation in the field Punch-in clock In Sweden [79] EMI from a digital punch in-clock to an AMR system was observed. In a light industrial facility, 18 out of the 25 electricity meters connected to the related substation lost communication with their AMR system. Fig. 68 shows the result of the frequency-domain measurement at the substation as well as at the incoming power cable; measurements at the three phases showed similar patterns. a) at the substation b) at incoming power cable Fig. 68: Power supply voltage (Phase L1) before any action As source of the disturbing emissions a digital punch-in clock was found, whose built-in AC/DC converter appeared as being the origin. After eliminating this device by the related customer, all meters properly communicated with the AMR system again. It can be recognized disturbance levels > 100 dbµv in frequency ranges khz and 78 - ~ 100 khz, that to be compared with the transmission channels of the AMR-system (75 khz, 86 khz) emission peaks of ~112 dbµv at 62 khz (substation) and ~117 dbµv at 61 khz respectively ~110 dbµv at 91 khz (incoming power cable) Fig. 69 shows the supply voltage after disconnection of the EMI source, followed by a reduction of the disturbance level peak values by ~20 db in the lower frequency range and ~40 db in the upper frequency range. a) at the substation b) at incoming power cable Fig. 69: Power supply voltage (Phase L1) after disconnection of the EMI source

65 Page 64 / Homeplug modem In a Swedish supply area, 11 out of 104 electricity meters lost communication with their AMR system, i.e. with the concentrator in the related substation. Measurements have been made [80] on all three phases at the substation, at the incoming power cable as well as inside the customer s premises at the socket where the device having been finally anticipated as being the EMI source a homeplug modem -- was connected to the mains. Fig. 70 shows the spectra of emissions measured on L1 each before taking action. Fig. 70: Peak levels of emissions from a PLC homeplug modem, measured at different points While at the substation, there appears one peak only, with ~ 99 dbµv at ~65 khz, measurement at the incoming cable resulted in values between ~90 dbµv and ~104 dbµv for the major part of the frequency range between ~3 khz and ~77 khz. At the socket inside the residence, where a homeplug modem was connected to the mains, levels between 90 dbµv and ~110 dbµv, with max. values of up to 124 dbµv (at ~17 khz) were measured. The EMI source was identified with a CE-marked PLC homeplug modem, which, according to the customer, was fully operational, but whose transmitter as well as receiver showed evidence of overheating, recognizable also through discolouration on the enclosure. After the customer having disconnected the modem and stopped using it, all electricity meters communicated with their AMR system again Voltage converter In Sweden, a voltage converter (Input: V / A; Output 24 V DC, 2 A), being installed in a cabinet for a street light system inside the substation, was identified as the source of EMI to an AMR-PLC system, resulting in a 100 % loss of communication, with all 230 electricity meters connected to a substation having lost communication to their concentrator [81]. From measurement on all three phases whose results show similar emission spectra, Fig. 71 shows the emission spectra measured on phase 2, showing the worst situation in terms of levels, before and after taking action.

66 Page 65 / 121 a) before taking action b)after replacement Fig. 71: Peak levels of emissions from a voltage converter After power line analysis was completed and the EMI source was eliminated, all electricity meters communicated with their system again. The owner of the installation replaced the unit with a new one Voltage converter II In another Swedish EMI case, a voltage converter (input: V / 0.4 A; output 12 V DC, 1.25 A), located in a private residence for power supply of a DSL router, was also identified as the source of EMI to an AMR-PLC system, resulting in a 75 % loss of communication, with 64 out of 85 electricity meters connected to a substation having lost communication to their concentrator. From measurement [82] on all three phases whose results show similar emission spectra, Figure 72 shows the emission spectra measured on phase 2, showing the worst situation in terms of levels, before taking action. Fig. 72: Emission peak levels from a voltage converter before taking action After replacement of the voltage converter with a new one, all electricity meters communicated with their AMR system again. Fig. 73 shows the related emission situation: Fig. 73: Emission peak levels from a voltage converter after replacement with a new one

67 Page 66 / Voltage converter to a broadband switch In another Swedish EMI case, a non-european, CE-marked voltage converter (input: V / 0.4 A; output 18 V DC, 0.8 A), located in a commercial property for power supply of a broadband switch, was also identified as the source of EMI to an AMR-PLC system, resulting in 200 out of 495 electricity meters having lost communication to their concentrator. From measurement [83] on all three phases, the emission spectra with highest levels each are provided below. Figure 74 shows emission spectra before taking action. a) before taking action b) after replacement with a new one Fig. 74: Emissions from a voltage converter to a broadband switch As can be seen on the left-hand charts, with increasing closeness to the wall-socket, where the voltage converter was connected to the mains, the emission levels were increasing, being close to the emission limits for signals according to EN [20] for a frequency range from 75 khz to 82 khz, at the incoming power cable

68 Page 67 / 121 Even exceeding these limits for a frequency range from ~48 khz to ~127 khz at the related wallsocket Central TV amplifier Another EMI case having been observed in Sweden [84] showed a common central television amplifier, ( MHz, Gain 35 db) installed in the basement of an apartment building, causing about 100 electricity meters losing communication with their AMR-PLC system. Like in item , measurement of the emission levels at the substation and at the incoming cable showed an increasing tendency when approaching the EMI source connected in the customer s premises. Fig. 75 shows the emission situation measured on phase L2, representing the worst one out of the three phases as well as the emission levels on this phase after replacing the device by a new one by the owner. a) before taking action b) after replacement Fig. 75: Emissions from a central TV amplifier measured on phase L2 at the incoming cable Voltage converter to a fiber switch Several CE-marked AC adapters (Input: V AC, 1.5 A; Output: 48 V DC, 1.5 A) of the same make and mode, in the reported cases used for power supply to fiber switches and installed in the basement of apartment buildings, have been found in Sweden causing loss of communication between electricity meters and their AMR-PLC system. Fig. 76: Emissions from a voltage converter at the substation In one case a communication loss for around 100 meters out of 228 ones connected to a related substation [85]. The units were reported to the broadband provider who replaced them with new ones. Fig. 76 shows the emission situation measured at the substation on phase 2, being similar on the other two phases Fig. 77 shows the emission situation at the incoming cables of two different apartment buildings connected to the related substation related to phase 2, which appears as being representative also for the other two ones.

69 Page 68 / 121 a) apartment building 1 b) apartment building 2 connected to the same substation Fig. 77: Emissions from a voltage converter at the incoming cable of two different apartment buildings It can be recognized max. emission levels of up to and even exceeding the limits for MCS signals according to EN [20] different situations of frequency ranges with emission levels o > 100 dbµv, in the ranges ~70 khz to ~105 khz and ~85 khz to ~90 khz at apartment building 1 in the range ~13 khz to ~94 khz at apartment building 2 o > 110 dbµv in the range ~85 khz to ~90 khz at apartment building 1 in the range ~16 khz to ~87 khz at apartment building 2 a distance of ~41 khz between max. peaks of emission levels in apartment building 2. After replacement of these adapters by the broadband provider with new ones, emissions returned to a quite lower level, followed by normal communication of all electricity meters with their AMR-PLC system Frequency-controlled ventilation in a school A frequency-controlled ventilation in a climatization room of a school was found as figuring as an EMI source [86], causing loss of communication of 33 electricity meters with their AMR-PLC system out of 69 ones being connected to the related substation. Fig. 78a shows the emission levels measured on phase L3 at the incoming cable to the climatization room before and after taking action, phase 3 representing the phase with the highest emission levels in this case, the other two phases showing similar emission behaviour. a) before taking action b) after replacement of the drive Fig. 78: Emissions from a frequency-controlled ventilation

70 Page 69 / 121 As can be seen, besides a peak with ~110 dbµv at ~28 khz, across the frequency range below 95 khz, the emission levels at the incoming cable are quite within ~83 dbµv and ~103 dbµv; around the frequencies used by the MCS being operated in the related supply area, 75 khz and 86 khz, NIE levels are between ~8 dbµv and ~102 dbµv; the margin to the MCS signal limits of EN [20] is obviously not sufficient to allow EM co-existence between the ventilation system and the MCS without taking measures. It may be anticipated that time-domain analysis of this EMI case would provide more insight into the interaction being responsible for the EMI. After dialogue with the owner of the property, he decided to replace the 10-years old drive with a new one, after what all meters communicated with their AMR-PLC system again (Fig. 78b) Frequency-controlled ventilation in an apartment building Another EMI case with a frequency-controlled ventilation system, this time installed in a climatization room in an apartment building, caused a 100% loss of communication for all 302 electricity meters connected to a substation [87]. There were two drives that caused the same EMI effect. Fig. 79 shows the emission situation measured at the substation as well as at the incoming power cable. Increasing emission levels makes visible the approach to the EMI source. a) at the substation b) at incoming power cable Fig. 79: Emissions from a frequency-controlled ventilation in an apartment building Also in this case, the emission level of around 109 dbµv to 119 dbµv around 80 khz was sufficiently high to cause the aforementioned EMI to the MCS operating with transmission channels on 75 khz and 86 khz After dialog with the owner of the property, an EMC filter was installed on the drives, followed by regular communication of all electricity meters with their AMR-PLC system again Rectifier inside a mobile site Inside a mobile site, two different types of rectifiers have been found as being the source of EMI to an AMR-PLC system, causing loss of communication of around 100 electricity meters out of over all 244 ones connected to a substation [88]. Fig. 80 shows the emission situation at the substation and at the incoming power cable at phase L2 before taking action, the spectra of L1 and L3 being quite similar.

71 Page 70 / 121 Fig. 80: Emissions from rectifiers inside a mobile site before taking action Representing the measurement point closer to the EMI source, the right chart shows a somehow modified pattern, not only with higher maxima (~132 dbµv at ~45 khz, ~115 dbµv at ~65 khz, ~107 dbµv at ~ 90 khz) and lower minima, down to ~52 dbµv) but also a move of the maxima in their frequency position Despite the lower emission levels at the incoming cable in the operational frequency range of the applied MCS (75 khz, 86 khz), compared with those measured at the substation, with a range of ~70 dbµv to ~79 dbµv, the disturbing effect on the MCS were to be observed. Fig. 81 shows the measurement results obtained at the same measurement points after installation of a filter, in his case supplied by the electricity company to the site owner. Fig. 81: Emissions from rectifiers inside a mobile site after installation of a filter Also here, the afore-mentioned differences in emission spectra between the substation and the incoming power cable can be recognized. But obviously the attenuation of the emissions at the power cable down to levels from ~58 dbµv ~70 dbµv in the range of the transmission channels of the applied MCS (75 khz, 86 khz) contributed sufficiently to proper function of the MCS. Also in this case, time domain analysis should provide more information about the EM interaction Undercounter bottle chillers Analysis of another EMI case [89], where the DNO lost communication with the electricity meters of his AMR-PLC system (transmission channels 75 khz and 86 khz) resulted in the finding of some under-

72 Page 71 / 121 counter bottle chillers in a bar of a hotel as the EMI source. The fridges were new and had been delivered by the beverages distributor. As the originator of the EMI the LED lighting inside the fridges were assumed. Two of them caused the most severe disturbances and with switching off these two devices all meters communicated well again. Fig. 82 shows the results of measurement of the emissions at the incoming power cable before and after taking action, on the phase with the highest emissions each: As a mitigation measure, EMC filters were installed. And the installer has since then further EMC filters ordered to be installed with every fridge. a) before taking action b) after installing an EMC filter Fig. 82: Emissions from undercounter display fridges EMI cases in Spain During the deployment of Smart Meters using OFDM-based PLC in Spain, some particular equipments have been found to generate remarkable EMI on the frequencies used for the AMR-PLC system (40 90 khz), thus deteriorating the performance of the smart meters and in some extreme cases leading to loss of communication and therefore loss of remote readability of it. Measurements for related investigations have been performed [115] in real field installations with connecting the measurement equipment just before the meter in the metering cabinet, which is not always near the equipment which is generating EMI, so the emission levels at the EMI source could be even higher. The impedance to be considered should be the one existent in the network, which is there typically 2 Ω in average. The following figures show the emission levels measured during investigation of seven EMI cases. As already mentioned, in all these cases, these emission levels caused EMI with AMR-PLC. Amplifier A Amplifier B Fig. 83: Power supplies of TV antenna amplifiers

73 Page 72 / 121 The power supply of TV antenna amplifier A showed peak emission levels at ~42 khz, ~60 khz and ~84 khz, with levels up to > 95 dbµv. The emissions of TV antenna amplifier B were measured with values up to ~85 dbµv, the maximum value situated at around 68 khz, showing higher emission leavels aroung 68 khz and a bandwith of > 5 khz BTS A BTS B Fig. 84: Power supplies of Base Transceiver Stations (BTS) In the considered frequency range (40 90 khz), the emissions of BTS A showed levels up to 92 dbµv, with its maximum at around 57 khz. BTS B showed peak values at ~45 khz, ~67.5 khz and ~90 khz, with a maximum value of ~75 dbµv at ~45 khz. /A: Elevator drive control Fig. 85: Inverter emissions Frequency inverter of a heating system An elevator drive showed emissions with peaks at ~25 khz, ~50 khz and ~100 khz, with a maximum value of around 77 dbµv at 50 khz (Fig. 85, left-hand). In the considered frequency range (40 90 khz), a frequency inverter of a heating system showed one emission peak, with a value of ~92 dbµv at around 32 khz; another peak emission was measured at ~64 khz. (Fig. 85, right-hand) Fig. 86: LED lighting emssions

74 Page 73 / 121 Measurements on some LED lighting showed kind of white noise with high levels over the whole considered frequency range, with values uo to 97 dbµv (Fig. 86) EMI cases in Finland From Finland, concretely the biggest DNO in Finland, Caruna, EMI cases caused by different NCE to their AMR-PLC system, operating on frequencies 75 khz and 86 khz are reported; the following types of equipment have been recognized as main EMI sources, mainly being represented by inverters or switchmode power supply units: TV antenna amplifiers UPS Camera surveillance systems TVs DVD players WLAN switches Computers 3G/4G base stations Chargers Inverters LED lighting Most common causes for EMI in the equipment were due to a broken or bad quality power or rectifier circuit in motor connections were wrong types of cables applied in inverters were occurring due to the installation guides not always having been followed Heat pumps The use of frequency controlled pumps for home heating is increasing. A number of such devices, having a reversing valve, are reported to have been found to generate disturbances to the grid [91, 92], resulting e.g. in loss of communication of electricity meters with their AMR-PLC system. Fig. 87: Frequency controlled heat pumps As is reported, obviously, the manufacturer had a filter for this, but the filter was not included in new deliveries. After related EMI analyses and having notified the manufacturer about the type of problem they were generating, the manufacturer mounted their own filter and the meters could be read again Frequency control to a pellet boiler Analysis of another EMI case with electricity meters having lost communication with their AMR-PLC system resulted in finding the frequency control to a pellet boiler as the EMI source [93]. Fig. 88 shows the related emission situation as measured on phase L2 where the highest emission levels were found.

75 Page 74 / 121 a) before taking action b) after installation of an EMC filter Fig. 88: Emissions from a frequency control to a pellet boiler In the right-hand chart, the frequency range having been found as that part of CENELEC band A (3 95 khz), which is most used by MCS [94] (see also item 9.3), is marked with a red rectangle. 8.3 EMI due to radiated field strength from conducted NIE/signals General As already reported in the 2nd edition of the Study Reports [2] (see in particular item 4.3, also radiated field strength from conducted emissions/signals is to be considered when dealing with EMC in the frequency range khz; several EMI cases, with (protected) broadcast time-signal systems, a contactless magnetic card reader and a traffic control system as EMI victims are reported there. In the following, information about EMI having been provided from Japan can be found, dealing with proved EMI cases with equipment in the telecom field. These EMI cases are referring to EMI sources causing emissions in the frequency range below 150 khz or khz, the interference being based on the effect of radiated field strength stemming from the emissions produced by inverters and power driving circuits; thus also highlighting the fact, that in the frequency range above 150 khz, being used in Japan for narrow-band PLC (see [95]), being apart from the frequency range used for MCS in Europe, avoiding co-existence problems with NIE equipment in khz, without any LW broadcast services being active in Japan other problems related to EMC need to be considered Radiated EMI to telecom equipment Examples of radiated EMI cases have been reported from Japan in the late nineties, with the source in the frequency range below 150 khz, affecting telecom systems as a victim. Tab. 10: EMI due to radiated disturbances to equipment in the telecom field. Examples. (Japan, )

76 Page 75 / 121 In all cases, EMI was caused by radiated disturbances generated by inverters and power driving circuits installed equipment figuring as EMI sources. Related EMI have been recognized from different equipment to telecom systems like Optical Network Units (ONU), Home Gateway connected to ONU (HGW) and DSL). Even when considering the given increase of mobile users and use of CATV as well as FTTH, with regard to still > 10 M DSL users in Japan, the spectrum of EMI effects (see Fig. 89a) and of equipment affecting the function of telecom equipment via the radiated path (see Fig. 89b) may be of interest. a) Disturbance effects b) Disturbance sources Fig. 89 Radiated EMI to telecom equipment. Sources and effects Additionally, the Japanese research report of 1998 provides an overview of recognized EMI caused by Grid Connected Power Converters (GCPC) (see Tab. 111, [96]): Tab. 11: Overview of EMI cases caused by GCPCs (Japan, 1998) A related report [97], presented to the annual IEEJ 6 meeting in 1998, providing information about the results of the related study including disturbance suppression measures, says: 6 IEE Japan

77 Page 76 / 121 Recently, Power electronics equipment using inverters (e.g., an elevator, a lift, and an uninterruptible power supply) come into wider use. These equipment have become widespread not only in factories, but also in commercial and residential area. In general, These equipment have a switching circuit operating at the switching frequency from several kilohertz to a few hundred of kilohertz. Large pulse current flows in the switching circuit, and high frequency noise occur because of the current. When noise suppression is not implemented, or is implemented for the limited frequency range, large noise current flows into ground lines and power lines connected to the equipment. Thus, electromagnetic waves are radiated around the equipment and create the potential for trouble Radiated EMI in higher frequency ranges Based on data from the Japanese Ministry of Internal Affairs and Communications from 2012, Japanese EMC experts reported about EMI effecting in much higher frequency ranges than the one where the related conducted NIE are generated (below 150 khz), due to the radiating effect of multiples of the fundamental switching frequencies below 150 khz in the EMI sources. Table 12: Examples of EMI cases due to emissions below 150 khz, causing radiated disturbances in higher frequency ranges (Japan, 2012). Even though the switching frequency of such circuits is below 150 khz, correlated common mode (CM) disturbance up to several hundred MHz (or up to a few GHz) is generated. Fig. 90 shows, how, according to the Japanese investigations, frequencies which can cause EMI to other equipment and related malfunctions, may be assigned to the ranges below 150 khz and from 150 khz to 500 khz: Fig. 90: Assignment of disturbance frequencies to the ranges DC to 150 khz and above 150 khz

78 Page 77 / 121 From the aforementioned recognitions, it derives also from the Japanese point of view, that specification of limits and methods of measurement for both conducted and radiated disturbance in the frequency range below 150 khz are needed for every type of electric/electronic equipment and systems with EMI potential in the considered frequency range(s) in order to avoid serious interferences and degradation of EM environment Descriptions of EMI cases Conductive and radiated EMI from an EV battery charger Investigations were made in Japan [98] on an EMI to different equipment/systems. Fig. 91 shows the transmission paths for conducted EMI having been anticipated to be the background for the EMI to ADSL and transmission speed of PC before taking measures the three types of equipment getting disturbed o o o ADSL modems showing transmission (bit) errors PCs showing reduced transmission speed radio receivers showing noise sound. Fig.91: Conducted EMI from an EV battery charger to different devise in customer premises Fig.92 provides the measurement results of the disturbances having been measured at the ADSL modem. Fig. 92 Disturbance shapes at ADSL modem before taking measures

79 Page 78 / 121 Fig.93 shows the spectra of emissions with and without EV charging in operation. Fig. 93 Emission spectra measured at point B (telecom line, L1 - E) of the ADSL modem with and without EV charging After taking mitigation measures as shown in Fig.94, emission level measurements EMI to the ADSL modem / transmission from PC was stopped. Fig. 94 EMI situation after taking mitigation measures There was still noise sound caused in a radio, thus showing the effect of EMI caused by the EV charger also via a radiated path Radiated EMI from an electric fence Another Japanese EMI case, not dealing with emissions from a device in the classical sense, was caused by induction due to fault currents from an electric fence (EF) to a telecom line, initiated by animals coming

80 Page 79 / 121 into contact with the EF. As a consequence, ADSL transmission was disturbed (bit errors) and linkdown of PCs was observed (Fig. 95). Fig. 95: EMI due to fault currents from an electric fence Measurement of the conducted disturbance levels on the telecom line [98] showed during occurrence of such fault currents (Fig. 96) oscillation bundles with up to 8,5 Vp-p increased levels in a broad frequency range from 10 khz up to at least 500 khz, thus also covering o the ADSL signal band to some extent o the frequency band used for PLC in Japan ( khz) Fig. 96: Conducted disturbance levels in a telecom line due to fault currents from an electric fence Measurement of the radiated emissions in the considered area showed an increase of emission levels by up to 50 db when the EF being ON. As a first mitigation measure, earthing of the cable sheat of the telecom line as applied. Thus measure resulted in a reduction of the maximum values of the afore-mentioned oscillation bundles, from 8,5 Vp-p to 2 Vp-p, thus in a decrease of the noise level, however linkdown still occurred. Additional application of a (DSL_MJS) CM filter led to a decrease of bit errors, but, again, many linkdowns still occurred.

81 Page 80 / 121 Obviously, some conversion of CM disturbance to DM occurred, due to insufficient cable symmetry and to long cables. Further mitigation, by adding a low-pass filter between L1 and L2 was followed by a decrease of the throughput, however the linkdowns did not further occur. 8.4 Summary Also the investigation results being reported in this chapter have been obtained with different measurement methods, so that a need for a unified measurement method particularly considering the needs of khz is seen. Emphasis of EMI has been recognized with devices using switch-mode technology, to a large extent e.g. PSU, converters As already explained in Study Report II [2], also the radiated interference path needs to be considered for EMI and its avoidance (e.g. by appropriate installation design). Again, this overview highlights, that EMI in the frequency range is not a sole issue of interference between NCE and MCS, but likewise such of EMI between NCE and NCE (see Fig. 1). Considering the experiences having been made in Europe as well as in Japan, o o with EMC in the frequency range below and above 150 khz, with regard to conducted as well as radiated disturbances there can be summarized: o o o o o o Switching circuits are widely used in electric/electronic equipment, and these circuits usually generate NIE having potential for causing disturbing effects to other equipment below 150 khz as well as above 150 khz. Therefore, a lot of different types of equipment/systems are to be considered as potential EMI sources. These emissions, at present being limited only to a small extent, may propagate on the mains, telecom cables as well as air, causing disturbance to other equipment, possibly causing degradation of performance, severe malfunctions or damages on telecom equipment as well as electric/electronic equipment in general. Concerning the above-reported EMI cases with AC adapters, it may be noted, that related EMI cases have been reported, beyond Sweden, also from several other countries, like UK (see Study Report II [2]) and France. It appears that due to quality reasons or to ageing effects, this mass-market device may represent a quite simple, everywhere used device with high potential to cause EMI in general and to MCS in particular. Likewise, a lot of types of electric equipment has been recognized as being sensitive to such emissions, be it arrived at their point of application on the conducted or on the radiated path Broadband services, xdsl and PLC etc. are widely used in customers premises, e.g. ADSL using a wide frequency range between 26 khz up to several MHz and being sensitive to radiated disturbances from multiples of NCE emissions on switching frequencies. MCS, using frequencies below 150 khz in Europe, as well as NCE generating NIE with fundamentals up to some tens of khz, re facing a thorough co-existence problem With Fig. 61 of Study Report II [2] an overview of the measurement values having been obtained concerning emissions in the frequency range khz without and with EMI with other electrical equipment be it NCE, be it MCS in comparison with existing standards levels has been given. Additionally to this chart, based on the results from measurements for Study Report II, being copied to Fig. 97a here, Figures 97b1 b3 provide overviews of results of emission measurements and investigations on EMI, related to small power supplies for ICT equipment (Fig. 97b1), on power drives (Fig.97b2) and on equipment like inverters, chargers, lighting and others (Fig. 97b3). Lines connecting measurement values indicate that the connected peak values are results from one and the same measurement, at different frequencies.

82 SC205A/Sec0400/R Page 81 / 121 Fig. 97 Emission peak levels from measurements on emission levels and investigations on EMI cases a) from Study Report II [2] b) for this edition of the Study Report, related to b1) small power supplies for ICT equipment b2) power drives b3) on equipment like inverters, chargers, lighting and others in relation to existing standardized limits and levels

83 Page 82 / 121 Amongst others, there can be recognized, that o all types of equipment show emission levels also exceeding the limits for intended signals (EN [20] o NIE from PSU to ICT equipment show quite high emission levels, in particular at higher frequencies, while emissions from power drives and other sources are more equally distributed over the frequency range 9. Standardization, Legislation & Regulation 9.1 General Power electronics as well as MCS play an important role in the task of ensuring energy efficient application of electric energy. Due to the increasing application of high frequency switching techniques (e.g. in PV inverters or active PFC circuits) a shift of non-intentional emissions from classical harmonics below 2 khz to higher frequencies in the range 2 khz to 150 khz the increasing application of MCS for smart metering/grid purposes, as successors of the previous ripple control systems, increasing application of the frequency range khz for signaling purposes. can be observed. Concerning utilization of the frequency range by the two aforementioned groups of electrical equipment, Fig. 98 gives an overview of the recent situation of utilization of the frequency range khz, highlighting also some differences in the utilization for narrow-band PLC systems in Europe and worldwide. The increasing number of EMI linked to higher frequencies and the different types of EMI effects (e.g. malfunction of equipment, falsification of meter readings, audible noise, reduction of lifetime (see Table 6)) lead to the recognition of an urgent need for the development of an appropriate standardization framework consisting of related compatibility levels, emission limits and sufficiently extended immunity requirements and measurement method (see items 6, 9) as well as to discussions about the appropriateness of / need for additional regulatory/legislative measures (see item 9.2). Concerning the utilization of frequencies for MCS (see Figure 98), Fig. 98: Utilization of frequencies in the range khz

84 Page 83 / 121 with regard to the position being taken from some part of experts to consider leaving CENELEC bands A D and to move narrow-band PLC to higher frequencies, up to 500 khz, it may be of interest, that for Europe, utilization of frequencies in the range khz is restricted to LW broadcast several (protected) radio services like flight and sea navigation e.g. in Japan there is no LW broadcast, but there are also a lot of flight navigation systems in the frequency range 160 khz to khz Together with information from the markets, it can be summarized: Further increasing application of power electronic equipment/systems, with switching frequencies also in the frequency band khz Tendency to further increase of switching frequency values Application also of software-controllable switching frequencies in inverters, thus with changing values of frequencies of related NIE Frequency range above khz (up to 500 khz) not applicable for MCS in Europe (radio region 1 of ITU) Main frequency range used for MCS for electricity supply network purposes in Europe is khz [94] Following to the information provided with the previous editions of this Report [1, 2], in the following an overview of the last actual development in standardization and related environment is given. 9.2 Present legislative & regulation situation The general framework related to EMC on legislative level is given by the Electromagnetic Compatibility Directive (EMCD) [37], with (in Article 3, Definitions) its definition of EMC (4) electromagnetic compatibility means the ability of equipment to function satisfactorily in its electromagnetic environment without introducing intolerable electromagnetic disturbances to other equipment in that environment; Here, the term intolerable appears as being a somehow undefined criterion, being essential for appropriate application of the EMC definition. its Essential Requirements (ER) in Annex I, 1. General requirements, requiring Equipment shall be so designed and manufactured, having regard to the state of the art, as to ensure that: (a) the electromagnetic disturbance generated does not exceed the level above which radio and telecommunications equipment or other equipment cannot operate as intended; (b) it has a level of immunity to the electromagnetic disturbance to be expected in its intended use which allows it to operate without unacceptable degradation of its intended use. As has been clarified by the EMC Consultant and confirmed by the European Commission (EC), with NCE as well as MCS are falling under the scope of the EMCD and therefore under the protection through its ERs. As mentioned before, through operation of NCE and MCS in parallel, a specific EMC & EMI situation is given for the frequency range khz. With regard to that, there are discussions about the definition of

85 Page 84 / 121 (5) electromagnetic disturbance means any electromagnetic phenomenon which may degrade the performance of equipment; an electromagnetic disturbance may be electromagnetic noise, an unwanted signal or a change in the propagation medium itself; and the -- given or not given need for distinction between disturbances and (wanted) signals. The definition above does not explicitly mention wanted signals, as being emitted from MCS by intention, having potential for disturbing effects but as an intended emission for signaling purposes gives rise for considering the appropriateness to see it as a disturbance. Beyond the EMCD, several other EU Directives and Standardization Mandates from the EC have impact on the development of EMC in electricity supply networks and between electrical equipment (see below). The Third Energy Package [99] requires Member States to ensure implementation of intelligent metering systems for the long-term benefit of consumers. This implementation may be conditional on a positive economic assessment of the long-term costs and benefits (CBA) to be completed by 3 September For electricity, there is a target of rolling out at least 80% by 2020, of the positively assessed cases. Furthermore, in line with the spirit, and complementing the provisions of the Third Package, the Energy Efficiency Directive [100] supports the development of energy services based on data from smart meters, demand response] and dynamic prices. It does that while respecting and promoting individuals right to the protection of personal data as enshrined in Article 8 of the Charter of Fundamental Rights of the European Union (the Charter), as well as ensuring a high level of consumer protection (Article 38 of the Charter). EEOC, a forum formed by ESMIG 7, Eurelectric 8, CEN 9 /CENELEC 10 and Orgalime 11, has analyzed the background situation for the given EMC problems in khz. Several Directives and Standardisation Mandates from the European Commission, aiming at increased energy efficiency on the one side and at deployment of smart metering on the other side, need to be considered in such an analysis: Electromagnetic Compatibility Directive 2014/30/EU (EMCD) [37] Energy Efficiency Directive 2012/27/EU (EED) [99] Third Energy Package 2009/72/EC (IEMD) [100] Ecodesign Directive 2009/125/EC (EDD) [101] Measuring Instruments Directive 2004/22/EC (MID) [102] Mandate Implementation of the Ecodesign Directive (M/495) [103] Mandate Measuring Instruments (M/441) [104] Mandate Electric vehicles charging (M/468) [105] Mandate Smart Grid Deployment (M/490) [106] It was recognized, that the development of increasing application of technologies meeting the requirements of these Directives and Mandates resulted in (increasing) problems concerning proper function of NIE, solid state meters and MCS; that due to the utilization of the frequency range khz, by NIE and MCS in parallel (Figure 99), without related regulation or comprehensive standards related to EMC being available. 7 European Smart Metering Industry Group 8 European Union of the Electricity Industry 9 European Committee for Standardization 10 European Committee for Electrotechnical Standardization 11 European Engineering Industries Association

86 Page 85 / 121 Fig. 99: Background to EMC problems in the frequency range khz The Commission and their EMC Working Party have been accordingly informed, and at present on political level trust is expressed to solve the problem with appropriate measures on standardization level (see also item 9.3) A specific issue is EMI to broadcast time-signal systems, as having been already described in SR II [2, 3] and the question of (regulatory) protection of such from harmful interference. Additionally to what is reported in Study Report II, the following may be of interest: In 2012, on the World Radiocommunication Conference (WRC-12), with amendments to Article 5 Frequency allocations [107], ITU 12 implemented some modifications, which with MOD 5.53, 5.54, 5.A116, 5.B116, 5.56, 5.67B) are relating to parts of the frequency range khz, dealing amongst others with frequency allocations for scientific research using frequencies below 8.3 khz meteorological aids service in khz radionavigation in khz in different countries maritime radionavigation in khz in China Further, particular, attention is to be drawn to MOD 5.56, which assigns protection from harmful interferences to stations of services for the transmission of standard frequency and time signals operating in the frequency ranges khz, khz and in Region I also khz and khz. Further, information is provided, that in, the frequencies 25 khz and 50 khz will be used for this purpose in several countries under the same conditions (see also Table 7 in Study Report II. Tab. 13: Protected frequency bands for standard frequency and time signal services [107] Frequency [khz] Region khz khz worldwide khz khz Region I 25 khz Armenia, Azerbaijan, Belarus, the Russian Federation, 50 khz Georgia, Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan 12 International Telecommunication Union

87 Page 86 / 121 With regard to different positions being taken from different parties like NCE manufacturers, MCS and smart meter manufacturers, DNOs as well as to several questions as mentioned above, there are different views on how to solve the recognized EMC problems, possibly by taking regulatory/legislative measures: NCE manufacturers, through Orgalime do not see a need for regulatory measures or a modification of the EMCD. As far as technical problems exist, they need to be solved based on the given legislative framework, through standardization [108]. DNOs, through Eurelectric, as well as Meter/AMR-PLC manufacturers, through ESMIG, o recognize a strong need for having regulatory protection of the CENELEC A frequency band (defined in the EN [20]) at European level; o feel that on standards level alone, maybe no sufficient protection can be achieved in order to protect against disturbances caused by other devices connected to the electricity grid [6, 121]. Some European Regulators have expressed their view of not feeling responsible for frequency management in the considered frequency range and related regulatory activity The EC is trusting on standardization to appropriately solve the problems, leaving it open -- only for the case of failing this goal -- to take measures like harmonization of relevant standards / on legislative level. 9.3 Present standardization situation Compatibility levels: Due to a related task given by IEC ACEC / SMB to IEC SC 77A WG 8 13, work on a completion of IEC [73] and -12 [109] with the specification of compatibility levels (CL) for khz, as a basis for future standardization of related emission limits and immunity requirements for products, is under way in WG 8. With regard to the quite different positions being taken from the different interested groups, related discussions on the appropriate setting of such CLs, takes more time and showed also the need for more detailed investigations on different parameters needed to be considered for this frequency range in particular (see also [1, 2] and items 2, 3, 5 and 6 of this report). In a first step, with regard to the specific situation in the frequency range khz, with parallel, contemporary use of the network/installation by NCE and MCS to the given situation of available standards dealing with emission limits and immunity requirements, discussions focussed on discussions focused on the recognition of a need for splitting the CL specification into two ones, considering the need for o o a higher value as a basis for setting immunity for general electrical equipment a lower value, as a basis for setting emission limits for NCE the consideration of an appropriate margin NIE and MCS signal levels, for ensuring proper function of MCS. An overview of several opinions related to the aforementioned needed margin shows Utilities and MCS/smart meter manufacturers, deducing a required margin of db (see chapter 4), depending on the frequency range NCE manufacturers, anticipating that CLs somehow equal with C1 curve of TS Ed 2.0 (see information about emission limits/values in this item) should not cause intolerable interference to PLC in the field in practice Others, like the Swedish consulter Swemet, indicating a maximum NCE level of - 40 db (3.5 V) as to be considered acceptable for proper MCS communication [110]. 13 WG8: Description of the electromagnetic environment associated with the disturbances present on electricity supply networks

88 Page 87 / 121 As mentioned earlier, the sole setting of CL values, focused on the frequency domain, appears as not being appropriate, thus not considering possible hided effects of emissions when considering its behaviour in time domain. Further, with regard to the aforementioned margin between NCE and MCS signal levels and related additional efforts for NCE to meet related requirements, setting of one continuous CL curve (for immunity, for emission) each, over the whole frequency range khz, appears as problematic approach. Therefore, as one of the approaches, a split of the frequency range khz into two sub-ranges has been considered, that considering the main utilization areas of the two groups of equipment/systems in question: 2 x khz, where the main switching frequencies of power electronics are situated x 150 khz, mainly used by MCS. That somehow similar to the shape of the emission limits curve in EN [54] & EN [56], with its sudden decrease of emission limit values at 50 khz. For the choice of a related split frequency, IEC TC 13 made an enquiry on the main utilization of frequencies for smart metering systems, resulting in their recommendation [94], supported by CENELEC SC 205A, of 30 khz for choosing this split frequency. Further discussions were considering another split for setting CLs, distinguishing, for the NIE-related CL, between the two environments public supply network on the one and installations on the other side. Emission limits/values: Up until now, due to still lacking CLs, the situation of emission limitation by standards has not changed compared with what has been described in Study Report II [2]. DTS Ed. 2.0 of TC 22, whose development has been described in the previous editions of this Study Report [1, 2] received a positive vote [117]. In its EMC related part, TS [117] provides emission values being recommended to be considered by active infeed converter (AIC) manufacturers for their products. In the discussion on setting CLs for khz, the values of its C1-curve are proposed as a basis for setting the emission-oriented CL, while DNOs are in favour of the spurious emission limits curve of EN [20]. Immunity requirements: Since the start of standardization work related to the frequency range khz, concerning immunity the following results have been obtained: For testing immunity of equipment to conducted, differential mode disturbances and signalling in the frequency range 2 khz to 150 khz at a.c. power ports, IEC SC 77A WG 6 has worked out a new standard. IEC/CENELEC EN [42] provides specifications for test waveforms; range of test levels; test equipment, test set-up, test procedures and verification procedures to demonstrate the immunity of electrical and electronic equipment operating at a mains supply voltage up to 280 V (from phase to neutral or phase to earth, if no neutral is used) and a frequency of 50 (IEC: 60 Hz) when subjected to conducted, differential mode disturbances such as those originating from power electronics and power line communication (PLC) systems CENELEC TR [75], making available a first specification comprehensively considering needs for immunity to conducted disturbances, for the specific product electricity meter, was worked out by CENELEC TC 13 and published in ad Emission limits & Immunity requirements: In the absence of standards specifying compatibility levels, immunity requirements or emission limits, product standards did not consider the frequency range khz, as there could not be referenced to related EMC standards. Consideration of this frequency range in Generic standards would ensure consideration in product standards whatever kind of, up to declaring that no specifications were to be applied for the related product, due to irrelevance for this product. But there are opinions that EMC below 150 khz is not an issue relevant for all products so that possibly unnecessary burdens for product manufacturing need to be avoided. Taking the immunity issue and the situation after publication of IEC/EN [42], it may be noted, that new product standards still do not consider this frequency range in general and this

89 Page 88 / 121 immunity test standard in particular. A new Draft IEC Ed. 1 (69/326/CD [111]) may be taken as an example. This draft, dealing with Electric vehicle onboard charger EMC requirements for conductive connection to an a.c./d.c. supply, still does not consider immunity in the frequency range below 150 khz. It may be the case, that immunity from conducted disturbances in khz is not a real issue for the related product. But emissions and their conducted as well as radiated effects, causing EMI to other electrical equipment (see investigation results from four European countries (item 7.2.4) and Japan) via the conducted as well as the radiated path, should be considered concerning their possibly disturbing effect. Therefore related consideration in product standards should be carefully taken into account. Measurement methods: An overview of recently available standards dealing (also) with measurement in the frequency range below khz, for different purposes and with somehow different measurement parameters, is provided in item. 6.2, together with some more detailed information on measurement parameters and the effects of related choices on the measurement results in Annex B. Power quality: Also TC 8X, considering the issue of khz as a PQ issue, started work on completion of the PQ standard EN [70], thus aiming at providing more comprehensive information about related phenomena, exceeding the recent part in EN 50160, which is limited to mains signalling, and that for the frequency range up to 95 khz only. 9.4 Summary. Needs for the future The existence of a gap in standardization related to EMC in the frequency range khz is a widely acknowledged fact now. From several sides, closure of this gap has been urgently recommended. Having been communicated the related problems to the European Commission and their EMC Working Party, trust has been expressed in standardization to appropriately solve the issue. In related discussions, from different parties quite different approaches are represented. It can be summarized: Controversial positions recognizing EMI problems in the frequency range Hz appearing to be o a purely European problem, with regard to involvement of MCS- vs. o a general EMC problem, with regard to - the entire spectrum of specific parameters to be considered related to EMC and their effect on reliable electrical equipment operation in general, also without considering MCS - wide variety of types of electrical equipment involved in related EMI cases to be considered worldwide Emission limits and measurement methods for conducted and radiated disturbance in the frequency range from 9 khz to 150 khz have still not been published except for particular equipment as induction cooking equipment and lighting equipment. There is a growing understanding for the urgent need for related EMC/Product standards (compatibility levels, emission limits, immunity requirements, measurement). As stated also from Japanese side [96] o There is a need for standardizing requirements for both conducted and radiated disturbance in the frequency range from 9 khz to 150 khz and 150 khz to 30 MHz taking account of: the variety of types of EMI sources like GCPC, LED lighting devices, DC charging for EV and wireless power transmission, which will be much more in commercial use in near future

90 Page 89 / 121 the wide spread of types of EMI victims including non-radio services as smart meter, xdsl and sensible electric/electronic medical equipment. o To define the aforementioned requirements, it is necessary to standardize a specific disturbance coupling model & measurement method, ensuring appropriate consideration of impedance behaviour of the supply network / installations consideration of the behaviour of emissions also in time domain comparability of measurement results Possible need for regulatory/legislative measures in addition to the EMCD as well as for a review of related rules like for installations. Need for distinction between non-intentional emissions and intended mains communicating signals when dealing with emissions having potential to affect reliable operation of electric equipment. Different views on the relative occurrence of related EMI problems, related to its share in the overall operation of equipment. While from one side such EMI cases are seen as occurring quite rarely, that leading to the preference to leave solutions up to case-by-case mitigation, the other party draws attention to local shares of degradation of performance or loss of function (in case of interference with MCS) of up to 100 % Conclusions 1. EMC in the frequency range is a general issue, related to EMC between NCE and NCE, and not limited to areas with MCS in operation and therefore to co-existence between NCE and MCS. Anyhow, interference potential between NCE and MCS represents a particular problem not only for the frequency range khz, but also in the frequency range above, from khz, which needs specific consideration in case of MCS operation in a certain supply area. 2. Beyond larger / commercial equipment & systems using power electronic technology, a lot of different types of electronic mass market equipment (e.g. PSU, lamps, EV chargers) showing relevant emission levels in the considered frequency range can be recognized. To some extent, conducted emissions having been measured show quite high peak levels. But there are also devices on the market with almost no relevant emissions. 3. In several cases, the emission level and/or the immunity seem to evolve with time, due to ageing of some components or the repetition of surges; this is mainly observed with small and cheap devices like power adapters, and is likely to increase with the development of common chargers and alike. 4. For many devices, the principal impact of high frequencies above an acceptable level is not on functionality itself, but on the failure rate. 5. Impedance of the grid and of installations (also due to varying situation of connected loads) in the frequency range khz shows varying values over time, beyond over frequency. This variation in time is an essential parameter to be considered related to its effect on propagation of related emissions. Recently, knowledge of the real impedance in the supply network and in the installation is limited; some single measurement activities ae under way. 6. Several existing proposals for measurement of the impedance in the grid / installation do not cover the whole frequency range khz. Further investigations are needed, concerning appropriate methods for finding information about the impedance reality in supply networks and installations as well as

91 Page 90 / 121 for defining measurement and calculation methods o considering the specific situation in the frequency range khz and o providing results being comparable with each other in general as well as between measurement in the lab and in situ. 7. Beyond the impedance issue, with regard to applied technologies for equipment, levels of emissions in the frequency range khz show variation over time, what is essential for the disturbing effect of such emissions to other equipment. 8. There is a need for distinction between emissions originally stemming from a device disturbances in the grid / the installation, considering the effect of impedances (resonances) on such emissions. For EMI analysis, the overall disturbances, consisting of originary emissions and responses from impedances need to be considered. 9. With regard to the aforesaid, the sole performance of frequency domain analysis does not provide sufficiently comprehensive results about the EMI potential of emissions in the frequency range khz. Time domain analysis getting increasing weight for analyzing emissions concerning their disturbing effect, appearing as increasingly indispensable, also with importance also for the setting of compatibility levels and, based on such, of EMC requirements. 10. Measurements have shown, that communication is not possible in the presence of NIE above a certain level. 11. Results of further investigations will need to be considered at setting of EMC requirements respectively at maybe reviewing already standardized ones as far as available. 12. In many cases, measurement results obtained in frequency domain do not comprehensively mirror the emission behaviour of a device, thus not providing safe evaluation bases for possible EMI. Therefore, increased application of time domain measurement is recommended when performing measurements in khz. 13. It appears as advisable to establish a new consistent framework for measurement and assessment of voltages and currents in the frequency range khz [4], thus providing a unification of the relevant measurement parameters for ensuring comparability and appropriate accuracy of measurement and evaluation results without requiring extensive processing performance. for the appropriate setting of emission limits, compatibility levels and immunity requirements 14. The appropriate choice of a measurement method appears as highly dependent on the kind of higher frequency distortion that shall be measured and evaluated. This may mean that, finally, different measurement and/or evaluation methods could be appropriate to be defined for different types of disturbance issues in the grid. 15. It may be impossible to ensure co-existence between NCE and MCS with sole consideration of a certain margin between NIE and MCS signal levels, thus focusing on a difference in terms of voltage level only. 16. On the one side, some larger, commercial equipment like EV and its charging equipment need more attention concerning EMI potential than household equipment. On the other side, together with the aforementioned ageing problem, quite small devices like voltage adapters, being widely used for the supply of electrical equipment, represent an essential EMI source.

92 Page 91 / To most extent, EMI investigations relate to conducted disturbances and signals. Beyond conducted emissions be it non-intentional or from MCS the radiated path is to be considered for EMI; that also and in particular for frequencies up to 500 khz. 18. The ageing problem is to be seen as twofold, as ageing of electronic components causes increased emissions from concerned electrical equipment, resulting in an increased potential for EMI to other electrical equipment thermal stress to other electronic equipment, resulting in a reduction of lifetime of such equipment 19. Even implementing techniques supporting MCS robustness, what is already done by the different PLC technologies being deployed in Europe, disturbances can generate problems to MCS 20. There is a need for standardizing requirements for both conducted and radiated disturbance in the frequency range from 9 khz to 150 khz and 150 khz to 30 MHz taking account of: the variety of types of EMI sources like GCPC, LED lighting devices, DC charging for EV and wireless power transmission, which will be much more in commercial use in near future the wide spread of types of EMI victims including non-radio services as smart meter, xdsl and sensible electric/electronic medical equipment. 21. There is a need for a review of installation rules, taking account of the findings of the now available Study Reports. Additionally, there may be a need for regulatory/legislative measures in addition to the EMCD, also and in particular concerning the problems arising due to the ageing effect. 22. With regard to the recognized EMC problems and loads in the household of today mostly being DC supplied Not a conclusion, but maybe a question to be considered for the future design of installations: Does it make sense to keep AC installations? While AC figures best for transportation of electric energy, wouldn t, in future, DC be the better basis for direct connection of loads, thus ensuring a clean grid and offering more freedom in voltage ranges on the grid. 11. Recommendations To consider EMC between electrical equipment in general when dealing with EMC in the frequency range khz, but taking account of the particular issue of EMC with MCS, for areas where MCS are operated. To develop a uniform measurement method for conducted emissions in the frequency range khz, considering also the effects of events of short durations, based on an identified set of measurement parameters and necessary measurements, needed for analysis of EMI and ensuring comparable results. Existing specifications like e.g. in EN [41] may serve as a basis. To consider also time domain measurement when performing EMI measurements and investigating EMI cases To consider protected broadcast-time signal frequencies with regard to radiated EMI caused by conducted disturbances in the frequency range below 150 khz To investigate the effect of and the process leading to a remarkable reduction of impedances over frequency in khz Consider EMC between NCE in general, with EMC between NCE and MCS as a specific problem. Develop a measurement method for khz, focused on the specific needs of given measurement tasks Consider the time domain at emission measurement and EMI investigation Consider broadcast-time signal systems & radiated EMI Investigate the impact of impedance behaviour over frequency

93 Page 92 / 121 To investigate in more depth, for khz, the real impedance levels in the supply network as well as in installations, for finding typical values, as an improved basis for setting of EMC levels and requirements. To consider time varying impedance behaviour (grid, installation, individual devices) and its impact on the propagation of emissions and on their effect to other electrical equipment. To take into account the effect of measures to reduce emission levels on the impedance behaviour To avoid measures to reduce the mains impedance over the frequency range below 150 khz below a certain low limit. To consider increasing occurrence of EMI also due to different political basics, aiming at energy efficiency, smart grids and the given application of energy efficient and mains communicating equipment the reported measurement and investigation results at future development of EMC and Product) standardization To consider the ageing problem at designing products and to check a need for related regulatory measures. To pay attention to the additional stress which is caused to electronic components and, following to that, to possible lifetime reduction, what should be considered at product design. To choose electronic components with a lifetime comparable with the usual lifetime of the equipment where it s applied. To further investigate on o o o emissions in the considered frequency range in general the specific emission behaviour of equipment, with variation over time, dependant on the applied technology the real / typical impedance of the grid and installations Investigate real impedance levels for finding typical values Consider time-variation of impedances in khz Consider mitigation measures effects on the impedance behaviour Limit inappropriate reduction of the mains impedance Consider measurement & investigation results at setting political goals and in standardization Consider the ageing problem at product design and at regulation Consider thermal stress to electronic components due to emissions in khz Proceed in investigating emission and impedance characteristics, in general as well as with regard to their behaviour over time

94 Page 93 / 121 Annex A: MCS robustness A set of basic principles for field proven effective PLC systems operating in the CENELEC A-Band 1) A preferred method is dual / multi carrier operation which is employed with automatic selection of the most robust carrier. 2) Carrier spacing is selected such that a single noisy device impairs just one of the carrier frequencies, allowing the other to function unimpeded (assuming common harmonic spacing of ~15 khz or greater). 3) Dual Carrier frequencies are selected. 4) High selectivity receive filters result in rejection of noise that is more than ~5 khz away from either carrier frequency. 5) Since both noise and communication signals are attenuated as they propagate through the mains network, a noisy device closest to a receiver tends to determine which carrier frequency may be impaired, and then which will provide the more robust communication. 6) Impulse noise cancellation is used to virtually eliminate its impact on communications. 7) Forward error correction is used to correct occasional data errors. 8) Acknowledged service with packet retransmission is utilized when an acknowledgement is not received. 9) Reliable communication between a central concentrating device and more distant meters is achieved when there is a positive signal to noise ratio of ~8 db (after impulse noise cancellation) for at least one of the carrier frequencies for each hop between adjacent devices. 10) The transmit level is ~128 dbµv when measured differentially (~122 dbµv when measured in accordance with EN [20]). 11) Signal attenuation on the mains distribution network is statistical but the median value has been measured to be ~40 db per 100 meters of distance. 12) Noise levels are also statistical yet the median value in Europe has been found to be ~60 dbµv with >95% of locations at <70 dbµv (in 3-kHz-bands centered on one manufacturers carrier frequencies). 13) This results in typical communication distances of ~150 meters per hop (SNR = 128 dbµv TX 60 db attn/150 meters 60 dbµv noise = 8 db SNR). 14) The above parameters have been found to yield reliable communications for millions of deployed meters throughout Europe with >99.7% coverage. 15) Note that any deployment of new devices producing noise above historically observed statistics would result in degraded coverage as an increased percentage of metering devices would then no longer be able to reach an adjacent device that could act as a repeater. 16) The same methodology can be applied to C-Band

95 Page 94 / 121 Annex B: Measurement issues: Frequency domain vs. Time domain analysis B.1 Frequency domain methods For measurement in frequency domain, generally two different principles exist: Heterodyne principle: - classical spectrum analyzer being used in communication technologies; - suitable for frequencies up to GHz range Discrete Fourier transform (DFT) principle: - classical PQ measurement instrument for harmonic measurements - suitable for frequencies up to several MHz Due to the different principles and the unavoidable variation of parameters, even one and the same emission at one and the same measurement point at one certain observation time results in different measurement results. B.1.1 Frequency analyzer based on DFT principle The methodology of an analyzer based on DFT principle is shown in Fig. B.1. Fig. B.1: general processing chain for DFT methods The processing chain starts with digitization of the signal. This process consist of the adaption of the signal amplitude to the desired level, the application of an anti-aliasing filter (low-pass) and the application of a signal filter (high-pass). The signal filter may be applied before or after the analog-to-digital conversion as an analogue filter or a digital filter respectively. The amplitude conditioning usually also realizes galvanic isolation from the network and overvoltage protection for the rest of the instrument. Then the analog signal is converted to a digital signal using sample-and-hold circuits in combination with an A/D converter. The digitized signal then undergoes further digital processing, which is specified according to IEC as pre-processing. A DFT is calculated for the digitized signal. This results in a complex spectrum, of which only the magnitude or absolute values are further considered. The phase angle is usually discarded. Depending on the selected frequency resolution for the DFT, the spectrum may be aggregated into wider bands. E.g. EN [aggregates the resulting spectra from 5-Hz-bands into 200-Hz-bands. If more than one spectrum is [25] recorded, these may be time-aggregated into longer intervals, e.g. from one spectrum per 200 ms to one spectrum per minute. This measurement data is then stored for later postprocessing.

96 Page 95 / 121 In case of time aggregation it is recommended to store in addition to the average values also the minimum and maximum value for each value of the spectrum. As e.g. having been verified by TUD [22] for the measurement methods according to EN [25] (Annex B) and EN [41] (Annex C), with several measurements of typical equipment emitting in the frequency range khz as well as on emission levels in LV networks, the different measurement parameters as described below, applied with the different methods, lead to different results. a) Amplitude conditioning The quality of the amplitude conditioning module of the measurement device is important regarding the SNR of the measurement system. The SNR of the amplifier has direct impact on the accuracy of the measurement results. b) Anti-aliasing filter According to the Nyquist Shannon sampling theorem the usable frequency range only extends up to the half sampling frequency f s /2. Spectral components above this frequency limit are mirrored at f s /2 and added to the spectrum below the limit. This makes any measurement result ambiguous, if no anti-aliasing filter is applied. The design of a suitable anti-aliasing filter is a trade-off between remaining impact of aliasing (level of damping), the complexity of the filter and the loss of accuracy in the desired measurement range. The limited steepness of any analog filter in the stop band means the edge of the pass band must be located below f s /2. For the specification of such filters usually frequency limits and the behaviour of the filter at these frequencies are described. As example measurements shall be carried out up to 150 khz at a sampling frequency of 1 MHz. The filter should reach -20 db at half sampling frequency and have a deviation from ideal transfer ratio G(f) = 1 of less than 0.5%. Choosing the commonly used Butterworth filter family, the minimal filter order to meet the requirements is 4. The 3-dB-cut-off frequency for this filter is set to 270 khz, but could be chosen between 269 khz and 282 khz. The frequency dependent damping behaviour of the filter is shown in Figure 98a and the deviation of the filter output from ideal transfer ratio is shown in Figure B.2b. a) Frequency dependent attenuation behaviour b) Accuracy in the measurement range Fig. B.2: Damping behaviour and measurement accuracy for the 4 th order Butterworth filter c) Signal filter Any device for measurements in the frequency range 2 khz to 150 khz using DFT methods must be equipped with a suitable high-pass signal filter to suppress the fundamental component as well as low

97 Page 96 / 121 order harmonics in the signal. Otherwise unwanted leakage of the fundamental component may result in less accuracy to a degree, where the signal might not be suitable for further analysis. Since the measurement intervals for measurements in the frequency range 2 khz to 150 khz are usually not synchronized to the fundamental frequency (i.e. constant 200-ms-window), the fundamental component will leak out, if it does not fit into the measurement window. Even frequency changes as low as 30 mhz, which are rather common in the European electricity network, can lead to a significant loss in SNR and therefore a loss in measurement accuracy [22] if no filter is applied. Figure B.3 shows the spectra of the voltage at the point of connection of a photovoltaic infeed converter at fundamentals different from the nominal power frequency under laboratory conditions. Without a high-pass filter (Figure B.3a) there is an increasing leakage caused by the varying power frequency. With a high-pass filter (Figure B.3b) the leakage is effectively suppressed. a) without high-pass filter b) with high-pass filter Fig. B.3: Spectra of voltage at point of connection of photovoltaic infeed converter in laboratory at different fundamental frequencies, without high-pass filter [22] A proposal for a suitable high pass filter should have several properties. It should: Provide high damping of the fundamental and low order harmonics Have high accuracy in the desired measurement range (e.g. < 0,5% error above 2 khz) Have a low order for improved stability and lower costs A practical example for such a filter can be found in [112]. The frequency response and error of the filter compared to the ideal transfer ratio are shown in Fig. B.4. a) Amplitude response b) Relative error Fig. B.4: Amplitude frequency response and relative error of the proposed filter d) Bandwidth As stated before, it is important that, results are comparable even if they were obtained using different methods. One of the most important aspects in this context is the frequency resolution or bandwidth of

98 Page 97 / 121 the measurement method. Fig. B.5 shows the spectrum of the voltage in a public low voltage network with photovoltaic inverters and MCS. The switching frequency of the inverter is 20 khz and the resulting component at this frequency is clearly visible in the spectrum. This component is rather narrow. The MCS operates in the range from 40 khz to 90 khz and the resulting components in the signal are rather broad. While the difference in the results is small for narrow signal components the difference is larger for broader components and reaches up to 10 db in this case. Theoretically, the difference can reach up to =20 log db when the results from measurements with bandwidths B 1 and B 2 are compared. The measurement devices according to CISPR [27] operate with a nominal bandwidth of 200 Hz, as also suggested in EN [25]. Figure B.5: Comparison of spectra of the same signal with different bandwidths e) Gaps in measurement PQ measurements in the range up to 2 khz according to EN [41] (see also EN [25]) are always gapless. There is even an overlapping defined in order to avoid gaps due to the transition from period-based to time-based aggregation (10-minute-clock-resynchronization). For the frequency range 2 khz to 9 khz, in an Informative Annex, also EN proposes gapless measurement, while one of the methods proposed in the Informative Annex C of EN Ed. 3 [41] allows large gaps in the measurement; as the latter specifies, 32 intervals with a duration of 0.5 ms should be analyzed per 10/12 cycles. This covers only 8 % of the signal as can be seen in Fig. B.6a. With regard to their position within the fundamental period it can be seen that always two intervals are located at the same position within a 200-ms-window and a fundamental frequency of 50 Hz. Therefore even considering the coverage related to one fundamental period still contains gaps (Fig. B.6b). a) regarding 10 fundamental cycles b) regarding one fundamental cycle Figure B.6: Location of the measurement intervals according to EN Ed.3 [41], Annex C

99 Page 98 / 121 Figure B.7 shows the value of the highest spectral component of the current of a LED lamp with apfc depending on the time of the start of a measurement according to the method in EN Ed. 2 [25] Annex B and EN Ed. 3 [41], Annex C. It shows that the measured value is constant for the first method, but varies ± 1 db or approximately ± 10% for the second method, depending on when the measurement was started. This is due to the modulated nature of the spectral component and that the measurement intervals are supposed to be synchronized to the fundamental frequency. Fig. B.7: Variation of the measurement results for the current of a LED lamp at 54 khz for the methods according to EN Ed.2, Annex B and EN Ed.3, Annex C depending on the starting time of the measurement The results improve if the measurements are desynchronized from the fundamental. This can be achieved by using random size gaps between the intervals. Therefore, measurement in the frequency range of 2 khz to 150 khz should be gapless. f) Measurement windows The DFT algorithm results in correct values only if the length of the measurement interval is an integer multiple of the period of all spectral components in the signal. For any component, for which this is not the case, leakage will occur. The magnitude of the leakage depends on the length of the measurement interval and the type of window. Further parameters Aggregation in frequency Fig. B.8 shows an excerpt from the spectrum of a photovoltaic infeed converter with a switching frequency 16 khz (solid blue line). The spectrum is aggregated in frequency into 200-Hz-bands according to the recommendation of EN [25] (blue x markers). The borders between two 200-Hz-bands are located at integer multiples of 200 Hz. Since the spectrum is symmetrical to the 16-kHz-limit, so are the resulting 200-Hz-bands. Most of the power is split evenly into the 15.9-kHz- and 16.1-kHz-bands, both of which reach about 1.9 V. If the switching frequency of the inverter is hypothetically increased by 70 Hz (solid orange line) and the 200-Hz-bands are calculated again, the two highest peaks are located in the same band (16.1 khz), which reaches 2.4 V. The two neighboring bands reach 0.9 V. While the total power of the two spectra remains the same, the power is split differently. Assuming a limit of 2 V is defined for each the 200-Hz-band, the first spectrum would be compliant and the second spectrum would not be compliant. This means that spectra with the same impact on devices may not be treated equally. The same issue applies for other bandwidth (e.g. 2 khz) too This issue needs to be addressed before emission levels are established.

100 Page 99 / 121 Figure B.8: Spectra of voltage in 5-Hz-resolution at the POC of a photovoltaic infeed converter with slightly shifted switching frequency (16 khz and khz) and the resulting 200-Hz-bands Measurement quantitites Spectral values can be stored either as peak or RMS values. RMS values represent the power of the spectral component and are normally used in Power Quality instruments. On the other hand spectrum analyzers can also provide peak values, which are not necessarily identical with peak values stemming from a DFT. It is important to indicate for a measurement value or a graph, whether RMS values or peak values are presented. Frequency aggregation is only meaningful applicable to RMS values. For spectra obtained from DFT calculation, the simple conversion between RMS values U and peak values Û can be used = 2, which is equivalent to a shift by 3 db in logarithmic scale. Aggregation in time Spectral analysis is commonly based on RMS values. For aggregation in time only quadratic average should be applied. The reason is that with quadratic averaging the resulting value is exactly the same as if the RMS calculation had been done for the full aggregation period. Future measurement standards should specifically mention quadratic averaging methods for the frequency range 2 khz to 150 khz, which is not explicitly done in EN Ed.3 [47]. Along with the average values, the maximum and minimum values are commonly calculated. If the measurement window is longer than a fundamental period (EN Ed. 2 [25], Annex B), these values are changing only very little for quasi-stationary signals. If the measurement window is much shorter than a fundamental period (EN Ed. 3, Annex C), minimum and maximum values can differ significantly. As an example, Figure B.9 shows the filtered waveform of the voltage from the point of connection of a photovoltaic infeed converter. For this waveform, the 200-Hz-bands according to EN Ed. 2, Annex B and the 2-kHz-bands according to EN Ed. 3 Annex C are calculated. For both methods, the different results for the time-aggregated maximum, quadratic average and minimum values over a time interval of 1 second are given in Table B-1. For the first method, all three parameters are virtually identical, while for the second method there are very large differences between the parameters. The first method hides the modulated nature of the signal. It allows for the analysis of the signal behaviour within the aggregation period. This may be of interest for the analysis of MCS, which may be turned on and off several times within a 10 min interval. The second method allows for the analysis of this signal behaviour in a much shorter time span, but it is blind for changes in operation state of the emitting equipment within the aggregation period.

101 Page 100 / 121 Fig. B.9: Filtered voltage waveform, at the point of connection of the PV inverter Table B-1: Comparison of maximum, quadratic average and minimum values for the waveform from Fig. B.9, 1 s duration Parameter EN Ed. 2 annex B EM Ed. 3 annex C U max [V] U avg [V] U min [V] The first method may be better suited for long term in situ measurements and the evaluation of the thermal impact on devices or noise generation. The second method may be better suited for measurements prompted by grid connected customers complaining about device malfunction. B.1.2 Analyzer based on heterodyne principle The information provided in this section is mainly derived from [113]. a) Overview A spectrum analyzer is a measuring instrument that displays an electrical signal according to its frequency. Each frequency component contained in the input signal is displayed as a signal level corresponding to that frequency. The usual spectrum analyzer measurements are: modulation, distortion, and noise. The most common type of spectrum analyzer is the swept-tuned receiver. It is the most widely accepted, general purpose tool for frequency-domain measurements. The technique most widely used is superheterodyne. Heterodyne means to mix - that is, to translate frequency - and super refers to superaudio frequencies, or frequencies above the audio range. Very basically, these analyzers "sweep" across the frequency range of interest, displaying all the frequency components present. The swept receiver technique enables frequency domain measurements to be made over a large dynamic range and a wide frequency range, thereby making significant contributions to frequency-domain signal analysis for numerous applications. Components of a Spectrum Analyzer Fig. B.10 shows the principle of a spectrum analyzer (Super-Heterodyne Method) Fig.B.10 Block Diagram of a spectrum analyzer (Super-Heterodyne Method)

102 Page 101 / 121 The major components in a spectrum analyzer are the RF input attenuator, mixer, IF (Intermediate Frequency) gain, IF filter, detector, video filter, local oscillator, sweep generator, and LCD. The Mixer A mixer is a three-port device that converts a signal from one frequency to another (sometimes called a frequency translation device). We apply the input signal to one input port, and the Local Oscillator signal to the other. By definition, a mixer is a non-linear device, meaning that there will be frequencies at the output that were not present at the input. The output frequencies that will be produced by the mixer are the original input signals, plus the sum and difference frequencies of these two signals. It is the difference frequency that is of interest in the spectrum analyzer. This signal is called the IF signal, or Intermediate Frequency signal. The IF filter The IF filter is a bandpass filter which is used as the "window" for detecting signals. Its bandwidth is also called the resolution bandwidth (RBW) of the analyzer and can be changed via the front panel of the analyzer. With a broad range of variable resolution bandwidth settings, the instrument can be optimized for the sweep and signal conditions, giving the possibility to trade-off frequency selectivity (the ability to resolve signals), SNR, and measurement speed. The Detector The spectrum analyzer must covert the IF signal to a baseband or video signal so it can be digitized and then viewed on the analyzer display. This is accomplished with an envelope detector whose video output is then digitized with an analog-to-digital converter (ADC). The digitized output of the ADC is then represented as the signal s amplitude on the Y-axis of the display. This allows for several different detector modes that dramatically affect how the signal is displayed. In positive detection mode, we take the peak value of the signal over the duration of one trace element, whereas in negative detection mode, it is the minimum value. Positive detection mode is typically used when analyzing sinusoids, but is not good for displaying noise, since it will not show the true randomness of the noise. In sample detection, a random value for each bin is produced. This is best for looking at noise or noiselike signals. For burst or narrowband signals, it is not a good mode to use, as the analyzer might miss the signals of interest. When displaying both signals and noise, the best mode is the normal mode or the rosenfell mode. This is a "smart" mode, which will dynamically change depending upon the input signal. For example, if the signal both rose and fell within a sampling bin, it assumes it is noise and will use positive and negative detection alternately. If it continues to rise, it assumes a signal and uses positive peak detection. The Video Filter The video filter is a low-pass filter that is located after the envelope detector and before the ADC. This filter determines the bandwidth of the video amplifier, and is used to average or smooth the trace seen on the screen. The spectrum analyzer displays signal-plus-noise so that the closer a signal is to the noise level, the more the noise makes the signal more difficult to read. By changing the video bandwidth (VBW) setting, we can decrease the peak-to-peak variations of noise. This type of display smoothing can be used to help find signals that otherwise might be obscured in the noise. Other components The local oscillator (LO) is a Voltage-Controlled Oscillator (VCO) which in effect tunes the analyzer. The sweep generator indeed tunes the LO so that its frequency changes in proportion to the ramp voltage. The sampling of the video signal by the ADC is also synchronized with the sweep generator to create the frequency domain on the x-axis. Because the relationship between the local oscillator and the input signal is known, the horizontal axis of the display can be calibrated in terms of the input signal s frequency.

103 Page 102 / 121 The RF input attenuator is a step attenuator located between the input connector and the first mixer. It is also called the RF attenuator. This is used to adjust the level of the signal incident upon the first mixer. This is important in order to prevent mixer gain compression and distortion due to high-level and/or broadband signals. The IF gain is located after the mixer but before the IF or RBW filter. This is used to adjust the vertical position of signals on the display without affecting the signal level at the input mixer. When changed, the value of the reference level is changed accordingly. Since we do not want the reference level to change (i.e. the vertical position of displayed signals) when we change the input attenuator, these two components are tied together. The IF gain will automatically be changed to compensate for input attenuator changes, so signals remain stationary on the LCD, and the reference level is not changed. b) Impact of measurement parameters on spectrum analyzer performances 1. Resolution The resolution is an important specification when trying to measure signals that are close together and wanting to be able to distinguish them from each other. As mentioned before, the IF filter bandwidth is also known as the resolution bandwidth (RBW). This is because it is the IF filter bandwidth and shape that determine the resolvability between signals. In addition to filter bandwidth, the selectivity, filter type, residual FM, and noise sidebands are factors to consider in determining useful resolution. Resolution bandwidth (RBW) A signal cannot be displayed as an infinitely narrow line. It has some width associated with it. This shape is the analyzer's tracing of its own IF filter shape as it tunes past a signal. Thus, if we change the filter bandwidth, we change the width of the displayed response. Some manufacturers specify the 3 db bandwidth, some others specify the 6 db bandwidth. Fig. B.11 Illustration of a 10 khz-rbw for two signals of equal amplitude When measuring two signals of equal-amplitude, the value of the selected RBW tells us how close together they can be and still be distinguishable from one another (by a 3-dB-'dip'). For example, if two signals are 10 khz apart, a 10-kHz-RBW will easily separate the responses. A wider RBW may make the two signals appear as one. In general, two equal-amplitude signals can be resolved if their separation is greater than or equal to the 3-dB-bandwidth of the selected resolution bandwidth filter. RBW Type and Selectivity Selectivity is the important characteristic for determining the resolvability of unequal amplitude signals. Selectivity is the ratio of the 60 db to 3 db filter bandwidth. Typical selectivity ranges from 11:1 to 15:1 for analog filters, and 5:1 for digital filters. Usually we will be looking at signals of unequal amplitudes. Since both signals will trace out the filter shape, it is possible for the smaller signal to be buried under the filter skirt of the larger one.

104 Page 103 / 121 The greater the amplitude difference, the more a lower signal gets buried under the skirt of its neighbour's response. This is significant, because most close-in signals you deal with are distortion or modulation products and, by nature, are quite different in amplitude from the parent signal. RBW determines measurement time Fig. B.12 RBW type and selectivity When narrowing the resolution bandwidths for better resolution, it takes longer to sweep through them because they require a finite time to respond fully. When the sweeptime is too short, the RBW filters cannot fully respond, and the displayed response becomes uncalibrated both in amplitude and frequency - the amplitude is too low and the frequency is too high (shifts upwards) due to delay through the filter. Fig. B.13: RBW determines measurement time Spectrum analyzers have auto-coupled sweeptime which automatically chooses the fastest allowable sweeptime based upon selected Span, RBW, and VBW. When selecting the RBW, there is usually a 1-10 or a sequence of RBWs available (some spectrum analyzers even have 10% steps). More RBWs are better because this allows choosing just enough resolution to make the measurement at the fastest possible sweeptime. For example, if 1 khz resolution (1 sec sweeptime) is not sufficient, a sequence analyzer can make the measurement in a 300-Hz-RBW (10 s sweeptime), whereas the 1-10 sequence analyzer needs to use a 100-Hz-RBW (100 s sweeptime).

105 Page 104 / 121 Noise Sidebands Fig. B.14: Noise sidebands The remaining instability appears as noise sidebands (also called phase noise) at the base of the signal response. This noise can mask close-in (to a carrier), low-level signals that we might otherwise be able to see if we were only to consider bandwidth and selectivity. These noise sidebands affect resolution of close-in, low-level signals. Phase noise is specified in terms of dbc or db relative to a carrier and is displayed only when the signal is far enough above the system noise floor. This becomes the ultimate limitation in an analyzer's ability to resolve signals of unequal amplitude. The above figure shows that although we may have determined that we should be able to resolve two signals based on the 3-dB-bandwidth and selectivity, we find that the phase noise indeed covers up the smaller signal. Noise sideband specifications are typically normalized to a 1-Hz-RBW. Therefore, if we need to measure a signal 50 db down from a carrier at a 10 khz offset in a 1-kHz-RBW, we need a phase noise spec of - 80 dbc/1hz RBW at 10 khz offset. Note: 50 dbc in a 1-kHz-RBW can be normalized to a 1-Hz-RBW using the following equation. (-50 dbc - [10*log(1kHz/1Hz)]) = (-50 - [30]) = -80 dbc 2. Sensitivity One of the primary uses of a spectrum analyzer is to search out and measure low-level signals. The sensitivity of any receiver is an indication of how well it can measure small signals. A perfect receiver would add no additional noise to the natural amount of thermal noise present in all electronic systems, represented by ktb (k=boltzman's constant, T=temperature, and B=bandwidth). In practice, all receivers, including spectrum analyzers, add some amount of internally generated noise. Spectrum analyzers usually characterize this by specifying the displayed average noise level (DANL) in dbm, with the smallest RBW setting. DANL is just another term for the noise floor of the instrument given a particular bandwidth. It represents the best-case sensitivity of the spectrum analyzer, and is the ultimate limitation in making measurements on small signals. An input signal below this noise level cannot be detected. Generally, sensitivity is on the order of -90 dbm to -145 dbm. The best sensitivity is achieved at: - narrowest RBW (decreases noise) - minimum RF Input Attenuation (increases signal) - using sufficient Video Filtering (to be able to see and read the small signal): VBW 0.1 to 0.01 RBW However, best sensitivity may conflict with other measurement requirements. For example, smaller RBWs greatly increase measurement time. Also, zero db input attenuation increases mismatch uncertainty therefore decreasing measurement accuracy.

106 Page 105 / Distortion Because mixers are non-linear devices, they will generate internal distortion. This internal distortion can, at worst, completely cover up the external distortion products of the device. But even when the internal distortion is below the distortion we are trying to measure, internal distortion often causes errors in the measurement of the external distortion. The internally generated distortion is a function of the input power; therefore, there is no single distortion specification for a spectrum analyzer. For a particular application, it has to be determined whether or not the distortion caused by the analyzer, will affect the measurement. As for any nonlinear device, the second-order distortion increases as a square of the fundamental, and the third-order distortion increases as a cube. This means that on the log scale of the spectrum analyzer, the level of the second-order distortion will change twice as fast as the fundamental, and the third-order distortion will change three times as fast. 4. Dynamic Range The dynamic range is defined as the maximum ratio of two signal levels simultaneously present at the input which can be measured to a specified accuracy. It determines the amplitude range over which we can reliably make measurements. The dynamic range of a spectrum analyzer is limited by three factors: the broadband noise floor (sensitivity) of the system, the distortion performance of the input mixer, and the phase noise of the local oscillator. The first two factors are used to calculate maximum dynamic range (MDR). Therefore, actual dynamic range is the minimum of 1) the MDR calculation and 2) the noise sidebands. For comparison of methods see Annex C. B.2 Time domain analysis It is the nature of frequency domain methods that they cannot provide any time-related information of a signal, like e.g. a transient and its point on wave. On the other hand it is difficult to obtain any frequency related information, like switching frequency of an inverter by time-domain methods. The first issue is illustrated in Figure B.15. On the left side a signal with constant emission level (red) and a signal with varying emission level (blue) are shown. While both waveforms are obviously different in time-domain, they result in very similar amplitude spectra. The two spectral lines in frequency domain cannot be uniquely traced back to the varying emission in the signal. If the measurement results in frequency domain are aggregated into larger bands (grey bar) both signals cannot even be distinguished anymore. Fig. B.15: Different waveforms in time domain and frequency domain With regard to the characteristics of NCE and MCS, dependent on the parameters of the signal analysis deviations in measurement results in the order of magnitude of several tens of db are possible.

107 Page 106 / 121 As has been recognized in the recent past, signal analysis in frequency domain only may hide very short events. Therefore, as also already mentioned in CISPR [27] -- that discontinuous disturbances. may need additional time domain measurements --, it appears as recommendable to perform also timedomain analyses, in particular for EMI investigations. That for ensuring a comprehensive evaluation of emissions considering also discontinuities with quite short duration and its effect to NCE in general comprehensively evaluating possible disturbing effects of NIE to PLC data transmission over time in general, with respect to the transmission duration for a single PLC frame, for different technologies. In order to illustrate the hiding effect, results from measurements in a LV distribution network in Austria are presented in Fig. B.16. It shows that time domain effects are hidden in spectra because emissions are occurring with very short durations (transients) and are therefore not visible in the frequency domain charts. While the frequency domain charts, having been measured with a time difference of < 0.5 s, appear as being almost identical, referring to peaks of around 1 V in the left lower chart, the left above time-domain chart, referring to frequencies below 160 khz, shows o peaks with a periodicity of 10 ms and o differences in amplitude of up to 1.5 V or with an increase of up to 150 %. Such short-time emissions may be the cause for EMI through saturation effects to input circuits of NCE as well as of MCS, not to be recognized with frequency-domain measurement during related investigations. a) b) Fig: B.16: Time domain (a) vs. Frequency domain (b, calculated): Measurement results in a distribution network, at one and the same point of common coupling (PCC) and with a time difference of measurement < 0.5 s Results from another Austrian measurement having been made at a residential environment at the supply terminal are shown in Fig. B.17, with non-active MCS (no active communication during observation)

108 Page 107 / 121 permanent (noise) emissions > 2 khz on the mains non-periodic but frequently occurring transient (noise) emissions Fig: B.17: Measurement at supply terminal in residential area It can be recognized, that the spectrum indicates a transient but of course does not represent the voltage level. the peak level of the transient is 15 Volt (143 dbµv) frequently appearing (1-2 times per minute) during several minutes The source of these transients is not known, but these results are demonstrating the strongly limited evidence of analysis in the frequency domain (DFT). As transients are occurring not periodically measurements using a spectrum analyzer for peak or quasi peak detection also fail in representing the characteristic of this emission. Such transients can cause malfunctions at electronic devices like touch dimmer lamp but also ICT equipment (Router, modems, network interfaces) could be a victim. The Interference at protocol level seems to be less relevant than the impact via supply voltage finally causing crashes at interface circuits or any small processing devices e.g. router. Regarding MCS it should be highlighted that emissions out of the frequency range being used for communication purposes may have disturbing effects due to saturation of input circuits of MCS, with the possibility of blocking communication.

109 Page 108 / 121 Annex C: Comparison of measurement methods for the frequency range khz Based on investigations at TUD, Germany, beyond what is already described in EN [41], this contribution represents an additional attempt to compare different measurement approaches, namely based on DFT principle and heterodyne principle. Three different test signals based on synthetic waveforms and real measurements in grid environment were used for this purpose (C.2). For comprehensive conclusions towards the development of appropriate measurement methods further tests are required. C.1 Measurement Methods under Test Three signals were measured using four different measurement methods: 1. CISPR Pk: Method according to CISPR [27] with 200 Hz bandwidth and peak detector 2. CISPR QPk: Method according to CISPR with 200 Hz bandwidth and quasi-peak detector 3. IEC : Method according to EN Ed.2 [25] (Informative Annex ) with 200 ms measurement interval and frequency aggregation into 200-Hz-bands 4. IEC : Method according to EN Ed.3 [41] (proposed in a Note of Informative Annex C) with 32 measurement intervals of 0.5 ms within each interval of 200 ms and a bandwith of 2 khz C.2 Signals for comparison purposes C.2.1 Synthetic waveform The first signal is a synthetic voltage signal (Synth). It contains a fundamental component with 230 V rms at 50 Hz and a higher frequency component with 1 V at 20.1 khz. The filtered signal using the filter describe in [112] is shown in Fig. C.1. The filtered waveform still shows a small remaining fundamental component, which is a result of the limited damping of the filter (approximately 63 db at 50 Hz). Figure C.2 shows the related spectrum of a single 200 ms interval in 200-Hz-bands according to method 3 (EN ). Fig. C.1: Waveform of the filtered synthetic signal Fig. C.2: Spectrum of the filtered synthetic signal, in 200-Hz-bands according to EN C.2.2 Battery Electric Vehicle Charger The second signal is the measured voltage at the connection point of a battery electric vehicle charger (Charger) in a public low voltage network. The switching frequency of the charger is approximately

110 Page 109 / khz. The signal shows a significant variation of higher frequency components within the fundamental cycle. The filtered signal using the filter described in [112] is shown in Fig. C.3. Fig. C.4 shows the respective spectrum of a single 200-ms-interval in 200-Hz-bands according to the method 3 (EN ). Fig. C.3: Waveform of the filtered charger signal Fig. C.4: Spectrum of the filtered charger signal, in 200- Hz-bands according to EN C.2.3 Photovoltaic inverter and PLC-signal The third signal is the measured voltage in a public LV network with PV inverters and a narrowband MCS. One PV inverter operates at a switching frequency of approximately 20 khz. The MCS communicates with eight channels between 35 khz and 90 khz. A maximum PLC level occurs at approximately 75 khz. The filtered signal using the filter described in [112] is shown in Fig C.5. A pause in the PLC transmission can be seen between 12.5 ms and 14 ms in Fig. C.5. Fig. C.6 shows the related spectrum of a single 200-ms-interval in 200-Hz-bands according to method 3 (EN [25]). Fig. C.5: Waveform of the filtered PV inverter emission and PLC signal Fig. C.6: Spectrum of the filtered PV inverter emission and PLC signal, in 200-Hz-bands according to EN

111 Page 110 / 121 C.3 Measurement setup Signal generation The signals were synthetized in MATLAB and repetitive outputted with a sampling rate of 1 MS/s using a digital to analog converter from National Instruments. The output signal of the converter was amplified to the desired amplitude (50 Hz, 230 V) using a linear amplifier with very low distortion. Measurement methods CISPR Pk / CISPR QPk The device for the measurements according to CISPR [27] was a Hewlett Packard 8593EM EMC Spectrum Analyzer. All measurements were made with 200 Hz bandwidth. A calibrated voltage divider was used to reduce the signal amplitude by 30 db. The input impedance of the divider is 1.5 kω, while the input impedance of the device itself is 50 Ω. For each signal, only a small frequency range (scan interval) of the spectrum around the specified higher frequency components was evaluated. In this range the maximum value was taken for evaluation. For the peak detector the frequency range was 10 khz. For the quasi-peak detector the frequency range was 1 khz. All results were recalculated into dbµv and V and the attenuation by the divider was compensated numerically. EN / EN For the measurement according to EN and EN respectively, a transient recorder with 16 Bit analog to digital converter resolution was used to sample the waveform at 1 MS/s. The antialiasing filter was set to 300 khz and ± 400 V measurement range was used. The input impedance of the device is 10 MΩ. The measured signals were high-pass-filtered using the digital implementation of the filter described in [112]. Scan time The scan time for method 1 (CISPR [27] with peak detector) for a scan from 9 khz to 150 khz amounts to 10.3 s. The frequency resolution of the scan (not the bandwidth) was automatically adapted by the device. Using method 2 (CISPR with quasi peak detector), the scan interval was reduced to 1.8 khz, because such a scan takes 100 s. A scan over the full range 2 khz to 150 khz at this speed would require about 2 h 17 min, which was not feasible for this study. The methods 3 and 4 (EN [25] and EN [41]) are able to produce the full spectrum from 2 khz to 150 khz every 200 ms or 10 fundamental cycles. While method 3 provides a gapless analysis, method 4 covers only less than 10% of the signal (see [116] for more details). C4. Summary of results Fig. C.7 presents the results at the emission frequencies of the different signals measured according to the four methods. Fig. C.8 provides a comparison of the differences between the methods, selecting method 2 (CISPR [27] with Quasi Pk) as reference.

112 Page 111 / 121 Figure 7: Comparison of the results for the different methods and different emissions/signals Fig. C.8: Comparison of differences of the results between the individual methods and emissions / signals in relation to the method 2 (CISPR QPk) in db For the synthetic signal (Synth) with constant higher frequency component at 20.1 khz all methods result in almost the same level of 120 dbµv or 1 V (RMS) respectively. The variation in the higher frequency content in the signal of the EV charger causes higher deviations between the measurement methods of about ± 3 db. Method 1 (CISPR with peak detector) provides the highest value. Method 4 (EN ) results in the lowest value, which is a result of the limited signal coverage. Method 2 and 3 provide similar results. The emission of the PV inverter at 20 khz results in slight differences between methods 1 to 3, but a significantly higher value for method 4. One reason for the smaller value of method 3 (EN ) compared with the CISPR methods is the location of the emission on the edge of the 19.9-kHz and 20.1-kHz-bands. The measured levels for both bands are dbµv and dbµv and therefore almost identical. Details of the spectrum are shown in Figure 9. This means that the total power of the PV inverter emission at switching frequency is spread into the two neighboring bands and the peak value is reduced by about 3 db. More details about this issue can be found in [116]. For method 4 (EN ), due to the 2-kHz-bands, it is not possible to identify the emission of the PV inverter in the spectrum, which results in the large difference compared with the other methods