CHARACTERISATION OF IN -HOUSE EMC TESTING FACILITIES FOR PRODUCT DESIGNERS. Paul Kay* and Andrew Nafalski**

Transcription

1 CHARACTERISATION OF IN -HOUSE EMC TESTING FACILITIES FOR PRODUCT DESIGNERS Paul Kay* and Andrew Nafalski** *Austest Laboratories, Adelaide **University of South Australia School of Electrical and Information Engineering Abstract A method of characterising the frequency response of a non-ideal radiated emissions test site is proposed in the paper. The method uses a purpose built reference source (10 MHz comb generator) to determine frequency ranges that are adversely affected by resonances in a non-ideal radiated test environment. Reasons for wanting to characterise a screened room are outlined, from the perspective of performing radiated emissions tests in an unlined room across the frequency range MHz. The results of a trial using the proposed method on a particular screened room are presented and discussed. Limitations of the proposed method are considered and the results are interpreted in the context of using the unlined room for radiated emission tests. 1. INTRODUCTION Most electrical products are type tested for radio interference (EMI) in an Electromagnetic Compatibility (EMC) laboratory before they are placed into service. High capacity power and control plant can generate RF emissions that may disrupt communications services or cause process control problems in industrial installations. It is sometimes necessary to perform radiated emissions testing in-situ, in order to identify sources of troublesome interference, or to demonstrate compliance with various regulations for EMI for particular plant. The large physical size and high power requirements of plant employed for industrial motor speed controllers and high capacity power distribution systems prevent them from being tested for electromagnetic interference in a laboratory environment. Many manufacturers perform pre-compliance testing in their product development laboratories, to give confidence that the product will pass the final test at the EMC laboratory. Of necessity, these tests may be performed in environments that are poor approximations to the environment used for final testing. This study proposes a method for characterising nonideal radiated emission test sites, and illustrates the method using an unlined Radio Frequency (RF) screened room as an example. This example was chosen because screened rooms are widely used for this purpose. A screened room is a metal-walled enclosure designed to keep unwanted radio-frequency signals out of a measurement system, or to confine intentionally generated radio-frequency signals to a particular location. Screened rooms are commonly used as EMC emissions and immunity test locations for a large range of electrical products, which range from ozone generators to variable frequency drives. An unlined screened room, commonly available in industry, is such an enclosure with no lining on the internal walls to reduce reflections from incident Radio Frequency (RF) energy. Any receive antenna located in an unlined room with a radiating source will see energy from many paths, and the absolute (that is, equivalent open-field) amplitude of the incident signal due to a single source at a particular frequency cannot be determined. Fig. 1 illustrates the reflection problem in two dimensions [1]. Sizes of the transmitting and receiving antennas are neglected for simplicity. The method uses the received signal strength from a comb generator designed and constructed as part of the project [] to determine frequency ranges that are adversely affected by the test environment. The product designer can use this information to judge the quality of their in-house testing.

2 A common process in EMC emissions testing is to perform a pre-scan of the device under test in an unlined screened room, prior to undertaking final measurements for compliance on an Open Area Test Site (OATS). OATS are universally treated as an ultimate measurement reference. For the research and development engineer, the pre-scan can be done easily, but the compliance measurement probably means a visit to a test-house, so prior knowledge of the performance of the equipment is desirable. commercial products has become mandatory in most developed countries [3], [4]. The idea of using a comb generator to characterise measurement systems was first described by a team of engineers working at Hewlett-Packard which investigated properties of an anechoic chamber [5]. While the design of the Austest comb generator is based on the HP design, it has been changed substantially to accommodate readily available components and to perform additional functions for local requirements. 4. OUTLINE OF METHOD The basic philosophy is to fix an E-field receiving antenna in the room and place a radiating source of known performance at various positions in the region normally occupied by the Equipment Under Test (EUT). Typical antennas used in EMI pre -scans are biconical arrays, log-periodic arrays or a combination of the two. Figure 1 Illustration of multipathing in an unlined screened room The known radiating source is a comb generator that produces vertically polarised emissions at 10 MHz intervals from 10 MHz to around 1000 MHz. The transmitting antenna is a top-loaded monopole, 30cm in length with a 10cm diameter top disc.. SCOPE The method proposed in this paper applies to unlined screened rooms that do not have a time varying aspect to their frequency response. The method may be extended to characterising other non-ideal radiated emissions test sites, where consistent reflections are present, but does not attempt to address timevariations in the test environment (such as paddles in mode-stirred chambers). 3. PURPOSE The purpose of this paper is to propose a method that determines the consistency of the frequency response of an unlined screened room with varying radiating source locations, with a view to highlighting frequency ranges that are adversely affected by reflections in the test environment. The method may also be useful for characterising sites before testing equipment in field installations, although that application is not presented in this paper. International interest in the quality of electromagnetic emissions measurements has increased steadily over the last ten years, because EMI/EMC compliance for Figure Suggested test regions for a screened room used to pre-scan small table -top equipments under test. The main steps in the process are: (i) The region to be investigated is marked out and recorded for repeatability. This region should correspond to the volume of space normally occupied by radiators associated with the EUT in a pre-scan. For small, table-top equipment, this could be a portion of the table top and a vertical plane at the rear edge of the table, corresponding to the area normally occupied by cables attached to the equipment. Figure shows possible test regions, which are divided into a grid with 5-

3 10cm spacing to create regularly spaced test points. The receive antenna is located to the right hand side of the table, and one wall of the screened room is located to the left side of the table. The receive antenna should not be moved during the characterisation process. (ii) The emission profile (frequency response) for each marked location in the room is recorded, for the required polarisation. (iii) The data is downloaded from the measuring equipment to a spreadsheet for processing: the mean, median and standard deviations for each frequency data set are calculated. A frequency data set consists of all apparent field strengths (receiver amplitude readings) for the tested locations. (iv) Frequencies that show a large spread of values as the source is moved across the region occupied by most EUT are easily determined. The large amount of data required to usefully map the test area (almost 100 frequency measurements per point) precludes the use of any manual measurement process. Most modern EMI receivers and spectrum analysers provide some means of capturing data from the maximum level detected in a specified frequency range. The choice of a comb generator for a reference source is a convenient one from this viewpoint: the measurements are made on narrow-band emissions that the automatic measurement receiver can quickly and easily lock onto, and a fairly short dwell time can be used to obtain a stable reading. For the purposes of this study, the region under investigation was confined to a non-metallic table-top, 80cm above the floor of the screened room. The antenna was positioned on a tripod approximately 1m from the table and arranged in vertical polarisation. The internal dimensions of the room under investigation were.4 x.4 x.4 m. 5. RESULTS Eighteen points were investigated, spaced 10cm apart in a square grid pattern near the centre of the nonmetallic table. This corresponds to the region occupied by a physically small EUT during a radiated pre-scan. SD vs freq. with varying source pos SD (db) Frequency (MHz) Figure 3 Graph of standard deviation of received level at each test frequency as the test source is moved through the 18 test points in the screened room. Two receive antennas were used: a biconical array antenna for the frequency range MHz and a log-periodic array antenna for the range MHz. They were positioned (one at a time) on a tripod approximately 1m from the table that held the comb generator. The tripod position was marked to ensure the antennas are placed in the same position during consecutive scans, although the receive

4 antennas were not disturbed once the run of 18 measurements was started. The standard deviations of received field strengths were plotted against frequency (Fig. 3). This highlights the frequency ranges that are most susceptible to EUT positioning in the room. 6. INTERPRETATION When a signal is unintentionally generated (by an EUT in the case of a radiated pre-scan measurement), the source of the electromagnetic wave that impinges on the receive antenna is unknown. It is likely to be a combination of signals from: one or more cables attached to the EUT tracks on the PCB equipment case The standard deviation calculated for each frequency in this characterisation method is used as a measure of the sensitivity of the received level to the position of the source. The source is deliberately constructed to generate signals at only one location (the monopole antenna), and this location is moved in a methodical way through the volume of space normally occupied by the EUT and cables. Thus, for frequencies that show large changes in receive level with position, any correction factors that relate the pre-scan reading to an OATS field strength must be regarded as unreliable. However, the converse statement must be made more carefully: apparently consistent frequencies are only reliable if the radiating source is confined to the characterised region. For most practical EUT, the characterised region needs to be increased from what is presented in this paper to include the region shown at the back of the table in Fig.. This will form some estimate of the stability of the frequency response when cables are the radiating source. A rectangular-walled enclosure has resonant frequencies F [6], given by: k m n F = ( MHz) (1) l h w + with k, m, n Z, and only one of k,m,n equal to zero at a time. l,w,h are the enclosure dimensions in metres. A simple calculation exercise with a spreadsheet, simplified considerably by the equal wall dimensions of the room, gives rise to a chart of resonant frequencies for the different resonance modes in the room (Table 1). Table 1 Resonance behaviour of.4x.4x.4m squarewalled room observed null frequency (MHz) predicted resonant frequency (MHz) resonant mode ,1, ,1, ,, ,3, ,3, ,4, ,4, ,5, ,6, ,7, ,7,7 However, the following frequencies were not within 10 MHz (1 comb emission) of an expected resonance, but nulls in the amplitude response were observed: 40 MHz, 130 MHz, 40 MHz, 350 MHz, 630 MHz and 780 MHz. The nulls at 130 MHz and 40 MHz have been identified as generator artefacts. 7. LIMITATIONS OF THIS TRIAL The emissions reference source used for the investigation is not expected to model EUT with multiple interfaces very well. Equipment that has many cables connected to it may have emissions coming from more than one cable, which may substantially change the interference pattern in the room. The findings presented here are the first results from what will become a considerable exercise. This study compares a multipath measurement environment to a nominally single path environment. It should be noted that OATS measurements usually require the equipment under test to be rotated in the horizontal plane, and the receive antenna raised from 1m to 4m or 6m height to maximise the emission level at each frequency. This is done to maximise the multipath (two-path) receive level over the reflective ground plane used at most sites. This screened room characterisation method does not lose much by not rotating the device under test in the

5 small screened room, because emissions are reflected from the walls near the EUT and reach the receive antenna with minimal path loss. On-site characterisations should include antenna positions at pertinent points around the EUT position. 8. CONCLUSIONS This study does not purport to offer an alternative to OATS measurements. It seeks to quantify the limitations of a commonly used process in developmental and diagnostic EMI investigations the unlined room pre -scan. Further, it provides a method for determining limitations in real-world field test environments such as factories. For the particular screened room installation under investigation, the study has already made it clear that EMI re -design work should not be performed in the room in several bands. The variation in field strength with position in these bands is such that removing a device for modification and replacing it in almost (but not exactly) the same position for a re-test is likely to introduce more field strength changes than most modifications! This experiment has provided a baseline from which further investigation can be performed, and serves to illustrate the benefits of RF absorbing linings in such installations. It also helps clarify the rationale behind mode-stirred chambers, where the introduction of a rotating paddle allows use of time -variant multipath theory that is not applicable to the static case. The rotating paddle in a mode-stirred chamber causes the multiple paths to change, giving the electric field strength a Chi Squared distribution [7]. 9. REFERENCES [1] Keiser, B., Principles of Electromagnetic Compatibility, (3rd Edition), Artech House, Norwood, MA, [] Kay, P. Austest Laboratories Investigation into RF Propagation Variations, Final Year Project Thesis for BEng (Honours) in Electronic Engineering, University of South Australia, 000. [3] 89/336/EEC Council Directive On the Approximation of Laws of Member States Relating to Electromagnetic Compatibility, Official Journal of the European Communities, No.139, 5 May 1989, pp [4] Code of Federal Regulations No.47, Washington DC: Office of the Federal Register, U.S. National Archives and Records Administration, Part-15 Radio Frequency Devices, Sub-Part J-Computing Devices, Regulations Specifying Electromagnetic Emission Limits for Digital Devices. Part-18 Regulations Specifying Electromagnetic Emission Limits for Industrial/Scientific/ Medical Equipment (CISPR 11). [5] Wyatt, K. and Chaney, D., RFI Measurements Using a Harmonic Comb Generator, Hewlett- Packard Company, Colorado Springs, Co. in RF Design, January [6] Williams, T., EMC for Product Designers, (nd Edition), Newnes, Oxford, 1996 [7] Goldsmith, K. Reverberation Chambers What Are They?, IEEE EMC Society Newsletter (Fall 1999), IEEE, NJ, 1999.