29 CHAPTER 3 A LOW FREQUENCY PULSE WIDTH MODULATION STRATEGY FOR CYCLOCONVERTERS 3.1 INTRODUCTION Power quality problems are of increasing concern in contemporary industry particularly with progress in the application of ac-ac conversion systems involving power electronic converters (Lander 1987). It augurs the need of proper harmonic management methods to acquire a complete understanding of harmonic source behaviors and necessitate the need to explore the interactions between the converter and other system components (Syam et al 2004 a, Wang et al 2000, Syam et al 2004 b). The ac-ac conversion systems and their control strategies are thus under intense research to meet the requirements of the utilities. A control strategy refers to a set of firing signals that trigger the devices in a cycloconverter to achieve a specific target output voltage and frequency (Miyazawa et al 1989, Ishiguro et al 1991). The use of a particular control strategy may help to improve the shape of the input current and/or output voltage (Basirifar et al 2011). However the available approaches do not appear to offer the desired performance. It is in this prelude that modifications in the existing PWM techniques are contemplated to contribute in this perspective. There is a growing interest in the use of cycloconverters for bulk power transfer in a good number of applications (Das et al 1997). 30 Cycloconverters inherit the ability to provide simultaneous voltage and frequency transformation without the help of intermediate stage reactive component by synthesizing a low frequency waveform from appropriate sections of a higher frequency source (Pelly 1971, Lander 1987, Chu et al 1989, Shicheng Zheng et al 2009). However it assuages the dent of harmonics and consequent effects of increased losses, deliberate de-rating and lower efficiency. It therefore continues to urge the development of new measures to build methodologies in order to refine the shape of the specified output voltage (Agrawal et al 1992, Jeevananthan et al 2007 a). 3.2 PROBLEM FORMULATION The main aim is to design a low frequency, programmed PWM (PPWM) strategy in order to minimize the distortion in the output voltage of single phase cycloconverters. The proposed EAC (Grahame Holmes et al 2003) attempts to reduce the lower order harmonic content especially in the low output voltage range. The performance of this scheme is evaluated through MATLAB based simulation and validated using a prototype. The approach involves deriving mathematical relations for a host of techniques and ventures to highlight the merits of this methodology. 3.3 EXISTING TECHNIQUES The cycloconverter, being a primordial converter imbibes a facility to work with a family of control schemes. A number of firing schemes appear to revolve around to suit specific applications. Jeevananthan et al (2007 a) proposed a host of most commonly used methods (a) constant firing angle method (CFAM), (b) cosine wave crossing method (CWCM), (c) HWSVFAM and (d) QWSVFAM. The standard single phase 230V, 50 Hz AC waveform seen in Figure 3.1 (a) is considered as the input and a frequency transformation (Kf) ratio of three is chosen to explain the schemes. 31 3.3.1 Constant Firing Angle Method Figure 3.1(b) shows the output voltage waveform for CFAM, where the delay angle is allowed to vary according to the voltage requirements of the load. This is a simple straightforward method, which gives a crude output waveform with considerable harmonic content resulting in its poor THD. 3.3.2 Cosine Wave Crossing Method Figure 3.1(c) shows the output voltage waveform for CWCM in which the comparison of a sinusoidal reference of output frequency with two cosine carriers of input frequency (cosine and inverted cosine) generates the gating pulses. 3.3.3 Half Wave Symmetry Variable Firing Angle Method The HWSVFAM is a closer approximation to a sine wave that is synthesized by phase delaying the firing of the switches as shown in Figure 3.1(d). It is a modified version of CFAM and operates with a different firing angle for different input half cycles. However, the firing angle of the first and Kf th (third for this case) pulses are same, and hence the name half wave symmetry. The load receives a full input half cycle at its peak period Kf 1 2 th and progressively reduces its conduction for other sections owing to fact that the output fundamental component requires controlled phase area proportional to its instantaneous values. 3.3.4 Quarter Wave Symmetry Variable Firing Angle Method The QWSVFAM is basically a modified phase controlled cycloconversion, which makes use of hybrid commutation (both natural and forced commutation) for enabling quarter wave symmetry in the output. Each 32 input half cycle is phase controlled in such a way that the position of each output pulse (phase controlled half cycle) is as close as possible to the desired sinusoidal output voltage in addition to their proportionality average value as seen in the output voltage waveform in Figure 3.1(e). It is similar to HWSVFAM in the first quarter of any output half cycle while the second quarter differs, which is the exact mirror image of the first quarter. In this method, the delay angles of segments vary in such a way that in addition to average value of each segment, the pulse orientations also lie as closely as possible to the variations of desired sinusoidal output voltage and renders reduced the harmonics. Though the symmetric firing schemes viz., HWSVFAM and QWSVFAM support a marginal improvement, still there is a definite need to introduce refinements in the available strategies to harness a much lower distortion in the output voltage. Figure 3.1 Input/ output voltage waveforms (a) input (b) output-CFAM (c) output-CWCM (d) output-HWSVFAM and (e) output-QWSVFAM 33 3.4 MATHEMATICAL EXTRACTION OF THE PROPOSED EAC STRATEGY The existing cycloconversion strategies introduce lower order harmonics and increase the THD. The mission is to develop a low frequency, PPWM strategy that provides an improved performance in terms of minimal lower order harmonics and reduced THD. The EAC is a refined PWM strategy suitable for single-phase to single-phase cycloconverter. The width of the PWM pulses in this approach is determined by making the area of the PWM signal equal to that under the sampled target waveform, irrespective of the number of samples. It orients to compute the pulse widths for a wide range of input and output conditions. The basic principle of the EAC method is illustrated in Figure 3.2. Figure 3.2 EAC for cycloconverter 34 The area enfolded by the target output waveform in the kth sampling period is represented by, k m i AkT M sin 0t dt k 1 m i AkT M cos 0 k 1 cos mK f k mK f (3.1) The pulse width at each sampling period is calculated by equating the area under the target in every sampling period with the area of the actual output. The area enfolded by the actual output waveform in the corresponding kth sampling duration is given by k 1 m 2m 2 i AkO sin K f 0 tdt k 1 m 2m 2 i AkO 2 Kf sin k 0.5 0 m sin (3.2) 2 where, m is the number of samples per half cycle of the input waveform, M 2 is the modulation depth, o is the output frequency, Kf o is the i frequency transformation ratio and i is the input frequency. The pulse width, ‘ ’ for kth sampling period is calculated by equating the area under the target waveform represented by Equation (3.1) to the area under the actual output waveform given by Equation (3.2). 35 M k 1 cos mK f 0 2 sin cos 1 k mK f 2 Kf sin 0 X Y k 0.5 m sin 2 (3.3) where X k 1 K f M cos Y 2 j 1 1 cos mK f sin k k mK f 0.5 m The Fourier constants and harmonic analysis of output voltage are determined as follows. f (x) a0 2 a n cos n 1 nx 3 bn sin nx 3 T 2 an 4 nx f (x) cos dx T0 3 an 2Vm 3 3 sin x cos 0 nx dx 3 k 1 an 2Vm 3 m 2m 2 sin x cos k 1 m 2m 2 nx dx 3 (3.4) 36 k 1 m Vm 3 an 2m 2 sin( k 1 m 2m 2 1 an ( 1) j 1 Vm n n 1)x sin( 1)x dx 3 3 n 3 1 n 3 cos n 1 3 k 1 cos n 1 3 k 1 m m 2m 2m 2 2 cos n 1 3 k 1 cos n 1 3 k 1 m m 2m 2 2m 2 (3.5) bn 3 2Vm 3 sin x sin 0 nx dx 3 k 1 bn Vm 3 m 2m 2 cos k 1 m 2m 2 1 bn ( 1) j 1 Vm n n 1 x cos 1 x dx 3 3 n 3 1 n 3 sin n 1 3 k 1 sin n 1 3 k 1 m m 2m 2m 2 2 sin n 1 3 k 1 sin n 1 3 k 1 m m 2m 2 2m 2 (3.6) where j=1 for odd chopping and j=2 for even chopping for a particular value of Kf. The third order Fourier constants and harmonic analysis of output voltage are determined as follows. b3 2Vm 3 3 sin x sin xdx 0 37 k 1 b3 m Vm 3 1 cos 2x dx k 1 m b3 1 2m 2 j 1 2m 2 Vm 3 cos 2k 1 sin m (3.7) k 1 a3 a3 2Vm 3 1 j 1 m 2m 2 sin x cos xdx k 1 m 2m 2 Vm sin 3 2k 1 m sin (3.8) The advent of PWM appears to revelutionise the performance of switched converters in the sense they inherit the facility to be programmed and suit precise design requirements. The firing pulse generated using some form of PWM based approaches hold merit and influence the operation of the circuit in a satisfactory manner. It is precisely the fact as to why a pulse width modulation technique is contemplated as a prerogative in the perspective of triggering a power device. 3.5 MULTIPLE PULSE MODULATION METHOD The multiple pulse modulation (MPM) where several equidistant pulses per half cycle are generated at the points of intersection of the carrier and reference signal waves is explored as a firing mechanism suitable for cycloconverters. A triangular carrier and a square reference wave are chosen and the pulse width is allowed to change in accordance with the requirements 38 by varying the amplitude of the square wave. The Figure 3.3 shows the output voltage waveform for MPM. Figure 3.3 Output voltage waveform The value of the actual output voltage (rms) in the cycloconverter is given by, V0 V0 2 1 3 Vm 2 2 Vm sin 2 td t 3 2 Vm sin 2 Vm 2 sin 2 td t td t 2 sin 2 2 (3.9) 2 The Fourier constants and harmonic analysis of output voltage are determined as follows. f (x) a0 2 T 2 an a n cos n 1 nx 3 4 nx f (x) cos dx T0 3 b n sin nx 3 (3.10) 39 an an 3 4Vm 6 Vm 3 0 (2 Vc ) 2m (k 1) m sin( Vc 2m (k 1) m 1 an ( 1) j nx dx 3 sin x cos Vm n 3 1 1 n 3 n n 1)x sin( 1)x dx 3 3 cos n 1 3 (k 1) m (2 Vc ) 2m cos n 1 3 (k 1) m (2 Vc ) 2m cos cos n 1 3 n 1 3 (k 1) m (k 1) m Vc ) 2m Vc ) 2m (3.11) bn bn bn 2Vm 3 sin x sin 0 nx dx 3 (2 Vc ) 2m (k 1) m Vm 3 ( 1) j 3 cos( Vc 2m (k 1) m 1 Vm 1 sin n 3 1 n 3 sin n n 1)x cos( 1)x dx 3 3 n 1 3 (k 1) m (2 Vc ) 2m n 1 3 (k 1) m (2 Vc ) 2m sin sin n 1 3 n 1 3 (k 1) m (k 1) m Vc ) 2m Vc ) 2m (3.12) 40 The third order Fourier constants and harmonic analysis of output voltage are determined as follows. b3 b3 a3 2Vm 3 Vm 3 0 (2 Vc ) 2m 1 cos 2x dx ( 1) j a3 3.6 Vm 3 sin x sin xdx (k 1) m b3 a3 3 2Vm 3 1 (k 1) m Vc 2m Vm 3 (1 Vc ) 2(k 1) cos m m (Vc 1) m (3.13) 3 sin x cos xdx 0 (k 1) m ( 1) j m sin (2 Vc ) 2m sin 2xdx (k 1) m 1 Vc 2m Vm 2(k 1) sin 3 m m sin (Vc 1) m (3.14) SIMULATION The power circuit of a cycloconverter for a single-phase output with a single-phase input is shown in Figure 3.4. Basically it consists of two back-to-back connected full converters (P-converter and N-converter). A series of positive (rectified) input half cycles appears across the resistive load when the P-converter is gated, while series of negative input half cycles result when the N-converter is operated in an open blanking mode of approach. The configuration sets the load frequency to be a fraction of the input frequency. 41 Figure 3.4 Power circuit of single phase cycloconverter The single-phase cycloconverter with different control strategies is simulated using MATLAB. The circuit specifications are intuitively chosen to acquire a target output voltage of close to 100V (rms) fundamental at one third the input frequency across a 110 resistive load. An intuitive investigation reveals that the viable number of samples (m) lies in the range of eight and two for a modulation depth (M) that varies from 0.1 and 0.636. The suitable value for m in the low output voltage range is eight where the M varies from 0.1 to 0.18 while the same is two in the higher output voltage range for which M is between 0.54 and 0.636. Therefore, the study echoes the appropriate choice of the modulation depth and the number of samples to land at the desired output voltage. The trail and error procedure yields a modulation depth of 0.46 and 3 samples to yield the specified target voltage in all the schemes. The output voltage obtained for M=0.46 and m=3and their respective harmonic spectra are depicted in Figures 3.5 to 3.10. It is to be noted that the performance of EAC is obtained for the same number of samples as observed from Figures 3.9 and 3.10. 42 Figure 3.5 Output voltage and spectrum – CFAM Figure 3.6 Output voltage and spectrum – CWCM Figure 3.7 Output voltage and spectrum –HWSVFAM Figure 3.8 Output voltage and spectrum –QWSVFAM 43 Figure 3.9 Output voltage and spectrum – MPM Figure 3.10 Output voltage and spectrum – EAC The entries in Table 3.1 contain the amplitudes of the target fundamental and harmonic components 3, 5 and 7 in terms of percentage of fundamental component for the same modulation depth and number of samples to illustrate the lower value of THD for EAC. It is interesting to note from Table 3.2 that EAC offers an exhilarating performance even when compared with MPM scheme on a similar platform. The results re-incarnate the creation of the new firing strategy so as to accomplish an improved performance for cycloconverters. 44 Table 3.1 Comparison of lower order harmonics and THD using existing phase angle controlled and proposed EAC methods V3 V5 V7 CFAM Vo1 (V) 104.2 46.56 80.37 64.24 THD (%) 124.29 CWCM 104.3 46.6 80.39 64.33 124.16 HWSVFAM 104.5 45.48 71.45 50.18 119.71 QWSVFAM 104.4 6.01 8.58 77.74 99.49 EAC 104.2 1.45 12.18 17.25 75.38 Strategy Table 3.2 Comparison of lower order harmonics and THD using MPM and proposed EAC methods Strategy Vo1 (V) V3 V5 V7 THD (%) MPM 105.0 39.11 44.96 14.88 114.58 EAC 104.2 1.45 12.18 17.25 75.38 The variation of amplitudes of harmonic orders over a specific output voltage range for a modulation depth ranging from 0.44 to 0.52 with three samples for the different control schemes seen in Figure 3.11 go to highlight the supremacy of EAC. 45 Figure 3.11 Variation of harmonic order with THD (%) for m=3 46 Similarly the variation of harmonic orders with respect to THD obtained by varying M from 0.1 to 0.18, with eight samples for the different control schemes in the defined output voltage range is shown in Figure 3.12. Figure 3.12 Variation of harmonic order with THD (%) for m=8 3.7 HARDWARE IMPLEMEMTATION The scheme seen in Figure 3.4 is implemented through a suitable prototype constructed with the specific purpose to operate in the same horizon. A real time windows target using MATLAB is used to generate switching pulses to the constructed prototype converter based on MOSFET (IRF840) switches. The switching angles are calculated in off-line for different operating points of the converters. The virtual reality toolbox seamlessly integrates with real-time workshop targets. It supports simulations that use code generated by real-time workshop and a third-party compiler on the desktop computer. The virtual reality toolbox also supports code executed in real time on external target computers. The schematic pulse generation using the real time target of MATLAB is shown in Figure 3.13. 47 Figure 3.13 Schematic of real time workshop based pulse generation scheme The systematic digital generation of EAC pulses corresponding to the target output voltage of close to 100V(rms) along with the output voltage waveform is shown in Figure 3.14. Figure 3.14 Pulse pattern and output voltage The fabricated hardware with the work bench is illustrated as a photograph in Figure 3.15. It is tested in the same operating states and the output voltage waveforms along with their harmonic spectra are drawn in Figures 3.16 to 3.21 for the different control schemes. It is noteworthy to observe that the experimental arrangement annotates more or less similar performance thus bringing out the efficacy of the developed control algorithm. The tabulated readings in Table 3.3 authenticate the findings and served to validate the simulated response. 48 Figure 3.15 Cycloconverter prototype Figure 3.16 Output voltage and harmonic spectrum –CFAM Figure 3.17 Output voltage and harmonic spectrum -CWCM 49 Figure 3.18 Output voltage and harmonic spectrum – HWSVFAM Figure 3.19 Output voltage and harmonic spectrum - QWSVFAM Figure 3.20 Output voltage and harmonic spectrum – MPM Figure 3.21 Output voltage and harmonic spectrum - EAC 50 Table 3.3 Comparison of existing and proposed control strategies Simulated values Strategies Experimental values CFAM Vo1 THD Vo1 THD V3 V 5 V 7 V3 V5 V7 (%) (%) (v) (v) 104.2 46.56 80.37 64.24 124.29 107.8 50.7 86.1 72.0 122.7 CWCM 104.3 46.60 80.39 64.33 124.16 108.0 50 87 67.9 122.4 HWSVFAM 104.5 45.48 71.45 50.18 119.71 107.7 51.5 76 53.3 119.5 QWSVFAM 104.4 6.01 8.58 77.74 99.49 103.4 7.8 8.5 79.5 102.4 MPM 105.0 39.11 44.96 14.88 114.58 103.6 39.7 46.1 15 115.7 EAC 104.2 1.45 12.18 17.25 75.38 106.2 2.3 14.1 18.9 75.5 The constructed prototype hardware is also tested with the firing pulses generated in the same way shown in Figure 3.15 for an R-L load with R=110 and L= 5mH. The output voltage and output current waveforms shown in Figure 3.22 prove that proposed EAC also works with R-L load. (a) (b) Figure 3.22 (a) Output voltage and (b) output current waveforms for RL load 51 3.8 SUMMARY The concept of EAC has been implemented for a single phase cycloconverter to drive home the benefits of extracting an entire range of output voltage. It has been found to yield appreciable lowering of the THD values almost through the entire operating range of output voltage. A significant contribution has been to illustrate the superiority of the proposed technique even over a PWM based methodology typically in the lower ranges of the output voltage. The performance of the existing firing schemes has been elucidated to highlight the attractive features of the developed algorithm and portray its capability in drive applications where low speed operation is required.