57 CHAPTER 5 SWITCH MODE POWER SUPPLIES 5.1 INTRODUCTION The latest advancements in the consumer electronics market resulted in producing high quality classy electronic equipment, but they are compelled to work even in low-voltage power supply systems (e.g. personal computers, laptops, televisions, CD/DVD players, etc.). This equipments are sensitive to voltage variations and requires a regulated dc voltage supply, usually referred to as a switch-mode power supply (SMPS). It is estimated that power electronic loads now constitute for about 30% of the total demand [54] in the Indian residential load sector, and it is predicted that their share would increase tremendously in the times to come. In this chapter the analysis of the SMPS in relation to the residential load sector is emphasised, the general conclusions to a large extent are also applicable to the other load sectors where SMPS category of loads is prevalent, i.e. commercial load sector. The nature of the operation of SMPS load marks the nonlinear current waveform being drawn from the power supply system [54]. This is due to the charging/discharging of the dc link capacitor (Cdc, Fig 5.1), which is used to reduce (i.e. to smooth ) the variations of the bridge rectifier dc voltage output.
58 Fig 5.1 General Block diagram of SMPS load As the SMPS devices are non-linear loads, they are significant sources of harmonics in modern power supply systems. The occurrence of harmonics will have negative effects on the operation of power supply systems, including higher thermal stresses and overloading of system components (e.g. shortening of the lifetimes of transformers and cables), or increased neutral conductor currents. The harmonics may also interact with supply system impedance, leading to distortions of the supply voltage (e.g. flat-top voltage waveforms). This chapter presents some of the results of the continuous research work on providing more specific guidance (than the regular trends) about the harmonic emission characteristics of modern distribution system loads [55]. Even more precisely, this chapter considers the SMPS load category; the direct effects of harmonics on equipment have been studied extensively. The main
59 objective of this chapter is to analyze if harmonics in the voltage supply affect the sensitivity of equipment to the voltage sags. 5.2 CLASSIFICATION AND MODELLING OF SMPS LOADS 5.2.1 Classification of SMPS Loads Harmonic legislation in [55] stipulates that the electronic loads with rated active power less than or equal to 75 W do not need to satisfy any of the prescribed harmonic emission limits. Therefore, SMPS load category is divided/classified in this chapter into lowpower ( 75 W) and high -power (> 75 W) sub-categories (i.e. types) of SMPS load. The general structure (i.e. circuit topology) of low-power and high-power SMPS loads are almost similar, with an exception that low-power SMPS will usually not have the power factor correction (PFC) circuit, Fig 5.1, as they do not have to adhere to prescribed harmonic limits. High-power SMPS loads will utilise one of the two general variants of PFC circuits: passive-pfc (p-pfc), or active-pfc (a-pfc). SMPS with a-pfc use an additional dc-dc converter to shape input current into a sinusoidal waveform [56]. The devices with p-pfc, on the other hand, include a relatively large inductor in the current conduction path. As the inductor opposes the change of current, this will smooth the input current waveform, effectively widening input current pulses and therefore by reducing its harmonic content. The SMPS with a-pfc use more sophisticated circuits, and inject only a (very) low-level of harmonics. This assertion is additionally
60 strengthened with the fact that p-pfc type of SMPS load is more common, as it is considerably cheaper to implement [56]. It should also be noted that as the cost of power electronic circuits falls, the contribution of a-pfc type of SMPS load increased. In this chapter, the term high-power SMPS refers to SMPS devices with p-pfc and rated power > 75 W, while term low-power SMPS refers to SMPS devices with no-pfc and rated power 75 W. 5.2.2 Equivalent Circuit Model of SMPS Load During the steady-state operation, both low and high power types of SMPS load can be represented by the equivalent circuit given in Fig 5.2. Fig 5.2 Equivalent circuit model of SMPS load In Fig 5.2 Lsys and Rsys represent the system impedance (Zsys), while RSMPS and LSMPS represent the sum of all resistances and inductances in the SMPS conduction path. Resistance req is the equivalent load resistance, which represents dc-dc converter and the dc load
61 supplied by the SMPS. This equivalent load resistance is defined as [57]: v dc2 (5.1) Prated where vdc is the instantaneous value of the dc link voltage and Prated req is the rated power of the modelled SMPS device. Further details on the implemented equivalent circuit model are given in [57]. 5.2.3 Parameters of the Equivalent Circuit SMPS Model Table 5.1 lists per-unit (p.u) parameters of equivalent circuit SMPS model (based on the values identified from the actual SMPS devices), which have been shown to represent generic SMPS load in [58]. TABLE 5.1: SMPS GENERIC CIRCUIT/MODEL VALUES, [58] SMPS type Model Parameter Low-power RSMPS [pu] 0.00142 XCdc [pu] 0.036 XLSMPS [pu] - High-power 0.00709 0.036 0.0371 Although system impedance is not part of the model, it is used to represent the power supply system to which SMPS load, together with the other loads, is connected at the point of common coupling (PCC), Fig. 5.3. The impedance of the lines/conductors connecting loads to PCC is assumed to be negligible. The nominal and maximum values of system impedance used in this work are taken from [58], while
62 minimum system impedance value is estimated based on these two specified values, Table 5.2. Fig 5.3. Aggregate load connected to low-voltage network TABLE 5.2 : SYSTEM IMPEDANCE VALUES Value System Impedance [Ω] Min Z = 0.12 + j0.11 Nom Z = 0.25 + j0.23 Max Z = 0.46 + 0.45 Rsys [Ω] 0.12 0.25 0.46 Lsys [mh] 0.35 0.73 1.43 5.3 PARAMETER VARIATION Although the described generic equivalent circuit SMPS model can be used to represent some important characteristics of the aggregate SMPS load (e.g. their aggregate active and reactive power demands, [58]), it does not have the ability to correct the model harmonic cancellation between the individual SMPS loads. 5.3.1 Resistance RSMPS The resistance of the SMPS (RSMPS) is dominated by the resistance of the negative temperature coefficient (NTC) thermistor used for inrush current protection. Although resistors generally have a much smaller tolerance range, typically around ±1 %, the range applied in the analysis is taken as ±20 %, in order to correctly
63 represent different types of components and different operating temperature regions. A uniform distribution is taken for RSMPS to allow for a more random variance in this parameter. The influence of R SMPS on harmonic emission of low power SMPS load is small, Fig 5.4, but still more significant than in case of high-power SMPS, Fig 5.5. This is because the large inductor present in high-power SMPS with p-pfc will dominate the high-power device input impedance. Fig 5.4 Influence of resistance (RSMPS) on low-power SMPS: a) Instantaneous current waveform b) Magnitudes/Amplitudes of current harmonics
64 Fig 5.5 Influence of resistance (RSMPS) on high-power SMPS: a) Instantaneous current waveform b) Magnitudes/Amplitudes of current harmonics 5.3.2 Capacitance Cdc The capacitor (Cdc) in SMPS devices must be large enough to allow SMPS to ride-through a voltage interruption of up to 10ms [59]. However, if the voltage interruption starts just before the C dc is about to charge, then a 10ms interruption corresponds to a 20ms interruption for a fully charged capacitor. A common industrial practice is to select the value of Cdc to satisfy a hold-up time of 23ms, which is 20ms hold-up time plus a safety margin, [59]. In order to calculate the size of Cdc it is necessary to achieve a particular hold-up time; the rated power of the SMPS and the minimum operational input
65 voltage of the dc-dc converter in the SMPS must both be known. Although the exact value of minimum operating voltage will be dependent on the specific dc-dc converter, values of around 80 V are common. Therefore, the selected value of Cdc should be large enough to maintain dc link voltage greater than 80V. This was simulated using full circuit SMPS model in [59], where the value of Cdc was adjusted at each rated power to just satisfy the holdup criteria as shown in Fig 5.6. Fig 5.6 Range of typical values of Cdc found in SMPS load The values of Cdc obtained in simulations with full circuit SMPS model are compared with the values of Cdc found after the inspection
66 of actual SMPS, and with data from manufacturer s specifications in Fig 5.5. This allowed specifying a nominal p.u value of Cdc: X Cdc, pu 1 / wc dc 0.036 pu 2 (V phase / Prated ) (5.2) where Vphase is the rms value of the supply voltage, Prated is the rated power of the modelled SMPS and ω is the angular frequency of the supply voltage. The actual values of the capacitors in SMPS circuits may vary based on their manufacturing tolerance, which is typically ±20% for electrolytic-type capacitors used in low-voltage single-phase SMPS devices [60]. 5.3.3. Inductance LSMPS The inductance (LSMPS) of the SMPS device is dominated by the value of the PFC inductor selected to satisfy harmonic legislation [55],[60]. Accordingly, this model parameter is only present in highpower SMPS load. To determine the value of LSMPS inductor to satisfy harmonic legislation, the inductor size was adjusted until the harmonic limits were just met at each rated power using a detailed full-circuit SMPS model. This was repeated for the three values of Cdc identified in the previous section, which were found to have only a small effect. The minimum value is given by (5.3). wlsmps 0.0315 pu (5.3) (V / Prated ) It was found that this minimum value was approximately 15% lower X LSMPS, pu 2 phase than measured values. This generally agrees with typical inductor tolerances, usually given as ±10% [13] and ±15% [61].
67 5.4. Variations in SMPS Operating Conditions The rated power of an SMPS device is the maximum power that the device can safely provide during the normal operation. For the majority of SMPS applications, however, it is highly unlikely that the device will continuously provide this power. During the normal operation, power demand of the SMPS will alter, on the basis of the actual loading conditions at its dc output(s). It has been found that the power consumption of a typical low-voltage SMPS (e.g. TV, PC, monitor etc.) will depend upon the specific operating mode of the device. Therefore, it is important to consider how loading conditions of an SMPS device will influence harmonic content of the input current. The most dominant type of SMPS load in residential and commercial load sectors are PC s and monitors. To determine the range of their loading conditions (Pload), several of these devices were measured during typical operations [62]. The results of the power drawn, as a percentage of the device rated power, are shown in Fig 5.7. Fig.5.7 Measured variations in SMPS power demand.
68 During active (i.e. normal) operation, the SMPS is usually loaded at around 50% of Prated, so nominal value of Pload is taken as 50%. A normal distribution is taken as Pload is not constant, but is expected to vary close to nominal. For both SMPS types, the current pulses will become wider as power demand of the supplied load increases [62]. As a decrease of loading of an SMPS device reduces its harmonic emission, this indicates that they are designed to satisfy harmonic legislation for the operation at fully rated power. 5.5 DIVERSITY FACTORS AND HARMONIC INJECTIONS The harmonic cancellation occurs due to phase angle dispersion between the same-order harmonics produced by different individual SMPS loads. Diversity factor is the ratio of the vector sum of magnitudes of individual current harmonics in the considered aggregate load to their algebraic sum, (5.4). In this work, aggregate load refers to an aggregation of SMPS load. The values of the DFh will lie between one and zero, where DFh of one indicates no harmonic cancellation, while DFh value less than one indicates harmonic cancellation. DFh N I n 1 N n h I n 1 n h (5.4) where: I hn I hn hn is the harmonic current of order h injected by the nth load, θ is the corresponding harmonic angle and N is the total number of loads in the aggregate [62]. To determine the diversity
69 factors of low and high-power SMPS loads, all model/circuit parameters were simultaneously varied as previously described using the Monte Carlo simulation technique to allow for random variations of selected parameter values. The results in Table 5.3 show that harmonic cancellation between individual low-power SMPS loads is not as strong as the cancellation between individual high-power SMPS. For high-power SMPS, the assumed variations in p-pfc inductor values will have strongest effects among all other circuit parameters. Additional variations in power demands will further increase and determine resulting levels of harmonic cancellation in practical SMPS applications. Calculated diversity factors for a mixed aggregate of low and high-power SMPS s are generally between the diversity factors of same-type aggregates (Table 5.3), but cancellation of lower order harmonics (3rd-9th) is more pronounced. TABLE 5.3 Calculated diversity factor values Diversity Factor Harmonic Number Low-power High-power Mixed Aggregate 3 0.99763 0.99727 0.90869 5 0.99343 0.98846 0.66516 7 0.98714 0.9264 0.10072 9 0.97878 0.78415 0.77145 11 0.96837 0.85021 0.92453
70 Table 5.4 presents the asymptotic values of magnitudes and phase angles of current harmonics for both considered types of SMPS loads. TABLE 5.4: Calculated current harmonics magnitudes and phase angles Harmonic Number Low-power SMPS Mag. [% of fund.] Angle [ ] High-power SMPS Mag. Angle [% of fund.] [ ] 100-81.9 100 72.2 3 98.33 103.3 76.85-144.2 5 95.05-71.4 43.16-4.7 7 90.31 113.8 15.59 118.6 9 84.29-61.2 6.79 11 77.22 123.7 5.91 1-160.9-75.4 Using the Monte Carlo method for simulations of different aggregates of SMPS loads, it was shown that the harmonic cancellation between high-power SMPS is considerably greater than the harmonic cancellation between low-power SMPS devices. For the realistic scenarios of mixed-type aggregation of both low and highpower SMPS devices connected to low-voltage network with typical system impedance values, it was shown that harmonic cancellation effects are stronger in case of mixed-type aggregates, and that system impedance in most of the cases, but not always, will reduce harmonic emissions.
71 5.6 SINGLE PC AS A HARMONIC SOURCE The Power supply unit of a modern PC consists of input AC/DC converter with capacitive filtering on DC side. Simplified electrical scheme is shown in Fig.5.8. Current on AC side is determined with capacitor charging/discharging and therefore is impulse in nature. Such a wave shape is far from sinusoidal, so harmonic distortion is high. Fig 5.8 Simplified representation of a PC power supply unit.
72 The figure 5.9 shows the internal schematic circuit of PC power supply. The source of a PC is a nonlinear load that introduces many harmonics. It is clear that this complicated high frequency source will have a big influence to the power quality. Fig 5.9 Circuit diagram of PC source practice
73 5.6.1 SIMULATION RESULTS The circuit presented in Fig.5.2 was simulated by using MAT LAB. Fig. 5.10 and 5.11 shows an effect of non linear load (PC model) on source voltage and harmonic spectrum respectively. The voltage harmonics produces the total harmonic distortion (THD) i.e. 73.6% which will affect more when a large number of PCs are connected. As the source voltage becomes distorted sinusoidal wave, even harmonics are present, but their magnitudes are negligible because of low magnitude. Fig. 5.12 and 5.13 shows source current and harmonic spectrum respectively. The current harmonics produces the total harmonic distortion (THD) i.e. 108.53%. Fig 5.10 Source voltage of pc model
74 THDv=73.6% Fig 5.11 Harmonic spectrum of source voltage of PC model Fig. 5.12 Source current of PC model THD=108.53%. Fig. 5.13 Harmonic spectrum of source current of PC model