NIETJOURNALoFENGINEERING&TECHNOLOGY

Winter 2011coNcoL(),enNNNZ(j)(j)Performance Comparison of High

Frequency Single Phase PWMInverters Using PSpice

Samidha Garg'

Abstract

Thispaper presents PSpice simulation of sinusoidalPWMinvertersand theirperformance comparison. AnHPWMinverterhas been included with its analysis andsimulation results. Also BPWM, UPWM and HPWMinverters have been shown by using PSpicesimulation package and the performance of all thethree inverters has been computed and compared;their switching losses by calculating their individualpower ratios at different conditions. The conclusiondrawn is thatBPWMinverterhas a poorer performanceas compared to the other two methods and theremaining methods i.e. HPWM and UPWM havealmost an equally good performance. In case of softswitChing, HPWM has a better performance thanUPWM.

ISr. Lecturer.Nlli'I,Gr. Naida

'Director IIMT College,Gr. Naida

S.N Singh 2

Introduction

As power semiconductor converters arevery widelyused nowadays for industrial applications, like industrialdrives, u.p.s. etc., dc to ac converters are calledinverters. The function of an inverter is to convert theinput dc voltage into output ac voltage. The output of anideal inverter is sinusoidal ac but in practical cases wecan't get pure sinusoidal output. The output voltage of aninverter could be fixed or variable at fixed or variablefrequency. A variable output voltage can be obtained byvarying the input dc voltage and maintaining the inverterparameters constant.

On the other hand, the dc input voltage is fixed and it isnot controllable. A variable output voltage can beobtained by varying the gain of the inverter, which isnormally accomplished by pulse width modulationcontrol within the inverter itself. The inverter gain may bedefined as the ratio of ac output voltage to the input dcvoltage.

The output voltage waveforms of an ideal inverter shouldbe sinusoidal. However, the waveforms of a practicalinverter are not sinusoidal and contain certainharmonics. For low and medium power applications,square wave voltage may be acceptable and for highpower application, low distorted sinusoidal waveformsare required. With the availability of high speed powersemiconductor devices, the harmonic contents of outputvoltage can be reduced significantly by switchingtechniques.

Inverters are very widely used in industrial applicationslike variable ac motor drives, induction heating,uninterruptible power supplies etc. The input of theinverter may be a battery, fuel cell, solar cell or wind r-rr~...,power generator.

Sinusoidal PWM Inverters

In these inverters, the input dc voltage isessentially constant in magnitude, such as in thecircuit Fig 1, where diode rectifier is used torectify the line voltage. Therefore, the invertermust control the magnitude and frequency of theoutput ac voltage. This is achieved by PWM oftheinverter switches and hence such inverters arecalled PWM inverters. There are variousschemes to pulse width, modulate the inverterswitches in order to shape the output ac voltageto be as close to sinusoidal as possible. Out ofthese various PWM schemes, a scheme calledthe sinusoidal PWM will be discussed in detail.The sinusoidal PWM inverters may be broadlyclassified as in three phase and single phaseinverters.

FilterL

50 HZ ac

Ac motor

l...- __ -'- __ ----' inverterDiode rectifier

Fig1

lIa- PWM Inverters

In inverter circuit the PWM is a bit more complex.We would like the inverter output to be sinusoidalwith magnitude and frequency controllable. Inorder to produce a sinusoidal output voltagewaveform at the desired frequency, a sinusoidalcontrol signal at the desired frequency iscompared with a triangular waveform, as showninfig2.

v!~I II II : Vao, fundamental = VaoI

~-, \

""--..::::.= ~--

The frequency of the triangular waveformestablishes the inverter switching frequency and

Winter 2011

(Vao)/ (Vd/2)

1.00.80.60.4

0.20.10.0 L..L_---.l...LL __ ..L..I....LLL- __ ....L...I....l....L.l...L__

Ma=0.8, Mf= 15

mf Mf+2 2mf 2mf+1 3 mf 3mf+3

Harmonics h off 1~

Fig 2 Pulse width modulation

is generally kept constant along with theamplitude Vtria. Before discussing the PWMbehavior, it is necessary to define a few terms.The triangular waveform Vtria in fig 2.2a is at aswitching frequency with which the inverterswitches are switched (is also called the carrierfrequency). The control signal vacant is used tomodulate the switch duty ratio and has afrequency f1 which is the desired fundamentalfrequency ofthe inverter voltage output (f1 is alsocalled the modulating frequency), the inverteroutput voltage will not be a perfect sine wave andwill contain voltage components at harmonicfrequency of f1. The amplitude modulation ratioma is defined as

where, Vcont is the peak amplitude of the controlsignal. The amplitude Vtria of the triangularsignal is generally kept constant. The frequencymodulation ratio mf is defined as

mf= fJf,

liB-Selection of switching frequency

Now we discuss the selection of the switchingfrequency and the frequency modulation ratiomf. Because of the relative ease in filteringharmonic voltage at as high a switchingfrequency as possible except for one significantdrawback, switching losses in the inverterswitches increase proportionally with htswitching frequency. Therefore, in mostapplications the switching frequency is selectedto be either less than 6KHZ or greater than20KHZ to be above the audible range. If theoptimum switching frequency (based on theoverall system performance) burns out to besomewhere in the 6-2kHZ range then thedisadvantages of increasing it to 20KHZ are

NIETJOURNALoFENGINEERING&TECHNOLOGY

often outweighed by the advantage of audiblenoise with fs of 20KHz or greater[6].

Therefore, in 50-60HZ type applications, such amotor drives 9 where the fundamental frequencyof the inverter output required to be as high as200HZ) the frequency modulation ratio mf maybe 9 or even less for switching frequencies of lessthan 2KHZ. On the other hand mf will be largerthan 100 for switching frequencies higher than20KHZ.

The desirable relationships between thetriangular waveform signal and the controlvoltage signal are dictated by how large mf is.Here it is assumed that the amplitude modulationratio ma is less than 1.

lie-Over modulation (ma>1)The desirable features of a sinusoidal PWM are inthe liner range. One of the drawbacks is that themaximum available amplitude of thefundamental frequency component is not ashigh as we wish. To further increase theamplitude of the fundamental frequencycomponent in the output voltage, ma isincreased beyond 1.0 resulting in what is calledover modulation. Over modulation causes theoutput voltage to contain many more harmonicsin the side bands as compared with the linearrange (rnas f .G), as shown in figure3.

~r V 60V V V180' 24 300 360

l"ftJ 6~ do ~ 240 300 tl",Fig. 3 Over modulation, = 1.3

VAolVd/2).., --- -,,,,

i Squcfe wave

evermodultion, ,, I, I

1.5 3.24Mf= 15 Fig4a Voltage control by varying Ma

Winter 2011

VAo/(Vd/2)

11.0

0.5Ma=2.5, mf=15

n

Fig4C Sinusoidal PWM

Bipolar PWM

Vcontrol

I I

- - I"~ - Vd...• I

,,~ .,-" -Vd

- _~-=~L- L

Fig 5 PWM with bipolar voltage switching

Unipolar PWM

Vcontrol

~

Vtri

0-\--\ ~

(a)

VAN

0I1oIJ) h/Vd)Vd0.8~0.60.40.2 , t , II I , 3mf t t t t I I titO.0 1';-~-~2m-f-;-- 7

12=-'m-'c

fc'::

2C"-m"""'"f+-"'1::-=-..L.L~.J....4um...J..f.f-'--

Harmonics of f 1Fig 6 PWM with unipolar switching

(d)

HYBRID PULSE WIDTH MODULATIONINVERTERV

(b) 1/foFig 7 (a) Modulation and CarrierWaveform (b) VAB For HPWM

liD-Filtering of harmonics in the inverteroutput voltage

The harmonics in the output of the invertershould be filtered out to get sinusoidal outputvoltage. An LC filter can be used to filter out theharmonics. One of the advantages of PWMinverters is that the harmonics are shifted tohigher frequencies; hence it is easy to filter them.As the frequency of the harmonics is higher, thesize of the filter components can still be reducedsignificantly.

11I-Simulation process

In power electronics, several types of analysisneed to be carried out. For each type of analysisthere is an appropriate degree of simulationdetail in which circuit components and thecontroller should be represented. It is importantto note that at each step as may be desirable toverify simulation results by a hardware prototypein the laboratory.

Winter 2011

SPICE:

The abbreviation SPICE stands for SimulationProgram with Integrated Circuit Emphasis. It wasdeveloped at the University of California, 8erkely.There are several commercial versions of SPICEthat operate on personal computers underseveral popular operating systems. Onecommercial version of SPICE is called PSpice.

In PSpice. many features are added to make it amultilevel simulator where the controllers can berepresented by their behavior models, that is bytheir input-output behavior, without resorting to adevice-level simulation. There is an option forentering the input data by drawing the circuitschematic. In addition to its use in the industry,PSpice has also become very popular inteaching institutions to understand core subjectslike power electronic and communicationengineering.

Square wave inverter: A square wave inverterfor output frequency of 50HZ has been simulatedon PSpice. IRF 840 MOSFET has been used asswitches input dc voltage is 220V. The outputvoltage output current, input current and thefrequency spectrum of the output voltage areshown on the computer prin touts.

Bipolar PWM: An operational amplifier has beenused as PWM generator and IRF 840 MOSFETsare used as the switches.The switchingfrequency of the inverter is 850HZ and the outputfrequency is 50HZ and hence mf= 17. The peakof the triangular waveform is 6.625V, peak of thesinusoidal waveform is 5.3V and hence ma 0.8. Inthe simulation bipolar voltage is given to theswitch because it is easy to simulate and it doesnot need any digital circuit components like NOTGATE and AND GATEIn addition to that, the bipolar nature switchingrequires no blanking time and other gate drivecircuits to drive the switch. The disadvantage ofthis type of switchinq is that magnitude of theoutput voltage will be somewhat less than theunipolar switching signal case due to reversevoltage appearing during the off period of theswitch.

Unipolar PWM: The unipolar PWM is simulatedfor m=0.8 and mf=18 (effective mf) on PSpice.The simulation circuit, output voltage waveform,

NIETJOURNALoFENGINEERING&TECHNOLOGY

input current and frequency spectrum of theoutput voltage are shown in the computerprintouts. Similar components as in the bipolarPWM has used for simulation

HPWM Inverter: The modified PWM inverterhas simulated for ma = 0.8 and mf = 18. Theslow speed switches are replaced with thyristorsand simulated the circuit for resistive load. Theresults of this circuit are also similar to those ofthe earlier MOSFET inverter. As the thyristors areconnected in series with MOSFET for resistiveload the current in the thyristors will be zerowhen the MOSFET in the circuit is off. Theoutput: Input current and frequency spectrum ofthe output voltage are shown on the computerprintouts.

IV- Comparison of harmonics

In the square wave inverter, all odd harmonicsstarting from 3rd order are present. It is verydifficult to filter out the lower order harmonics.Hence this inverter is not recommended fordrives applications. The square wave inverter isunavoidable. For resistive loads like lighting theharmonics will not have any impact, hence it isrecommendable. In bipolar PWM the lowerorder harmonics are shifted to around switchingfrequency and its multiples.

The harmonics around the he switchingfrequency are high magnitude hence they canbe filtered. When compared with the UPWM, thistype of inverter is not recommendable for driveapplications. In UPWM lower order harmonicsare eliminated and shifted to around switchingfrequency and its multiples. As the frequency ofharmonics is higher, they can be filtered outeasily. Hence by putting a filter we can get thesinusoidal output. This type of inverter isrecommendable for drive applications.

But in SPWM inverters it is somewhat high eventhough it is not having any serious impact on theload because these harmonics occur atswitching frequency and its multiples.

Comparison of switching losses

In the square wave inverter the switches areoperated at the output frequency hence theswitching losses are very low. We can't comparethem with the pwm inverters. In bipolar PWMinverter all four switches will be changing states

Winter 2011

at every switching, hence switching losses willbe maximum. It is also not recommendable inthe point of switching losses. In UPWM invertersonly two switches are changing state at allswitching except at the 180 and 360. Hence theswitching losses will be much lower than thebipolar pwm. It is also recommendable in thepoint of switching losses. In the hpwm onlyswitches are operated at high frequency andremaining two switches are operated at lowoutput frequency, hence the switch losses willbe lesser than the upwm the slow speedswitches are replaced with low conduction lossswitches. In that case therefore it is morepreferable than the upwm.

Switches losses of the bpwm, upwm, hpwm arecompared as follow. These three inverters aresimulated for the same switching frequency andma calculated the output fundamentalcomponent's power by using the relation wherePo1 is the fundamental power component of theinverter output Vo, is the fundamentalcomponent of output voltage of inverter (r.m.s.value) io, is the fundamental component ofoutput current of inverter (r.m.s. value) The inputpower is calculated by using the relation P, =Vdld(avg) where Pd is the dc fundamental powerVd is the input dc voltage Id is the average valueof dc current fundamental power ratio of thegiven inverter is calculated as the ratio offundamental output power to input power.

Power ratio calculationsTo compare the power ratios of differentinverters these inverters are simulated for thesame condition like ma, mf, load power factoretc.

ma mf Vrms Irms Ro Id(avg) Powerratio

UPWM 0.8 18 205.448 4.11 50 2.949 95.36%(eff)

BPWM 0.8 17 288.974 5.78 50 5.777 96.36%HPWM 0.8 18 203.56 4.071 50 2.926 945%

For resistive load

ma mf Vo1 io1 10 Ro Id Powerratio

UPWM 0.8 18 230.3 3.171 116 500 1.20 995.6%(e mh

BPWM 0.8 17 227.02 3.668 116 500 1.217 92.4%mh

HPWM 0.8 18 231 116 116 500 1.23 94%mh mh

For inductive load Lo= 116mh and Ro=500

BPWM inverter For calculation of power ratiobpwm inverter has simulated with the followingdetails

Vd=300V, ma= 0.8, mf= 17, Load inductance=50mh, Load resistance = 50n

The following readings are taken from thesimulation results

m. rn, Vo, io, 10 R" I, Powerratio95.6%92.4%94%

UPWM 0.8 18 227.084 4.346 116mh 500 1.209BPWM 0.8 17 227.02 3.668 116mh 500 1.217HPWM 0.8 18 231 116mh116mh 500 1.23

For inductive Load Lo=50mh, R.,=sonwhere V,msis the rms value of output voltage

I,msis the rms values of current

Vdis the input dc voltage

Idisthe average value of input dc current

By comparing the power ratio values in the

above table it is to be noted that the hpwm

inverter is having poor power ratio than othertwo.

The hpwm and upwm have approximately thesame power ratios. Hence bpwm is notrecommended for practical applications.

In the hpwm inverter all switching losses occur inthe two pwm operated switches so the heatdissipation in that switches will increase hence itis also not recommended. In addition to that,operating frequency of the pwm switches inhpwm is double that of the upwm. But in case ofsoft switching, the hpwm inverter hasadvantages over the upwm inverter because theswitching losses in soft switch inverter are nearlyzero hence there is no switching loss dissipationin the switches.

By using the low power loss devices in place ofslow operating switches, we can reduce theconduction losses. Hence in case of softswitching it is recommended and for lowswitching frequencies also it is recommendedbecause the overall switching losses will bedecreased significantly. If we replace the slowspeed switches with the (GTO), the readings inthe above tables are taken at steady stateconditions.

Winter 2011

V-Inverter selection for a specific application

The selection of inverter for specific applicationdepends upon the load requirement. Forexample for restive loads, there will be noproblem due to harmonics and hence squarewave converter can be selected. For inductiveloads like induction motors. The harmonicsshould not present particularity at lowfrequencies. For that application, the harmonicsshould eliminated.

To eliminate the harmonics the output filter isneeded. If the frequency of the harmonics is lowthe filter size will be increased. In the PWMinverters the frequency of the harmonics will beshifted to high frequency i.e. at the switchingfrequency and its multiples. Hence the filter sizerequired to eliminate the harmonics will bereduced at cost of inverter switching losses.Therefore for drive application output filter isessential.

VI- Conclusions

In this paper, different types of sinusoidal PWMinverters are simulated on the PSpice Microsoftpackage. In the square wave inverters we haveall lower order harmonics (odd) i.e. 3'd,51h etc. Fordrive applications these harmonics should notbe there. To avoid these harmonics filters areneeded.

For lower frequency harmonics, the size of thefilter components required is large. Hence it isnot recommended to face lower frequencyharmonics. In addition to that, to control the drivemotors variable frequency and variablemagnitude supply are needed. It is easier to getvariable frequency and variable magnitudevoltage from the sinusoidal pwm inverters thanfrom square wave inverters.

The main disadvantage of sinusoidal pwminverters is the high switching losses. In thesquare wave inverter the switches are operatedat the low ouput frequency hence the switchinglosses are little. But in sinusoidal pwm inverters,the switches are operated at high switchingfrequency i.e. at the carrier frequency hence theswitching losses will be more as compared to asquare wave inverter.

In the sinusoidal pwm inverters the harmonicsoccur around the switching frequency and itsmultiples. As the switching frequency increases

NIETJOURNALoFENGINEERING&TECHNOLOGY

the harmonics frequency will also increase. Soat the cost of switching losses were pushing theharmonics to higher frequency.By thissimulation we observed the harmonicsspectrum of each inverter hpwm and upwmhave same harmonics spectrum and samepower ratio but hpwm has less power ratio andpoor harmonic spectrum.

Hence for any kind of application upwm is moresuitable than the hpwm and hpwm because theoperating frequency of the switches in hpwm isdouble the operating frequency of upwm and allswitch losses occur in high frequency switchesonly. Heat dissipation in the high frequencyswitches will be higher.

In sinusoidal pwm inverters however, theswitches are operated at high switchingfrequency i.e. at the carrier frequency hence theswitching losses will be more as compared tothe square wave inverter. In this way wesimulated the three phase sinusoidal pwminverter single phase square wave inverter,UPWM and HPWM with filters. In case of softswitching hpwm inverter is advantageous overthe upwm. We replace slow speed switches withlow conduction loss switches like GTO, it isbetter to use the hpwm over upwm.

Appendix

A three phase inverter has been simulated form.=0.8 and m,= 15 and results of the simulationare given here. The power ratios of BPWM,UPWM, HPWM with filter was calculated whichare as follows:

m, rn, V" i" 10 Ro I, Powerratio

BPWM 0.8 17 235.85 4.503 50 50 2037 84%UPWM 0.8 18 234.4 4.475 50 50 1.815 92%HPWM 0.8 18 232.25 4.431 50 50 1.8 91%Power rations with filters at Lo = 50mh, R, =500hmCALCULATION OFTHOTHO = .,j (VORMS2-V012)No1BPWM At rn, = 0.8 and m,=17 Lo= 50mh Ro=

500hm L,= 116mh and C,= 80uF

THO = 15.500

References1. R.L. Steigerwald, A.Ferraro and F.G.

Turnbull, ''Application of power transistor t 0

Winter 2011

residential and intermediate ratingphotovoltaic array power conditioners,"IEEE Trans. Ind. Applications. VoI.IA-19, No.2, pp. 254-267 Mar/Apr 1983.

2. F.Z. Peng, II. Akagi, and A. Nabae, ''Apractical approach to harmoniccompensation in power system-seriesconnection of passive and active filters" lEE ETrans. Ind. Applications vol, 27, No.6, p.1020-1025, Nov/Dec-1991.

3. H. Patel and R.G. Hoft "Generalizedtechniques of harmonics elimination andvoltage control in thyristors: Part-1 Harmonicelimination: IEEE Trans. Ind Applications, volIA-9, no.3, p. 310-317, May/June 1974.

4. K.D.T. Nago "Low frequency characterizationof pwm converter" IEEE Trns. Powerelectronics vol PE-1 no.4 pp 223-230. Oct1986.

5. Berrezzek Farid and Omeiri Amar, "Study ofnew techniques of controlled PWM inverters",Annaba European Journal of scientificresearch ISSN 1450-216X Vol.32 No.1 (2009),pp. 77-87http://www.eurojournals.com/ejsr.htm

6. A.Sakharuk, A.M. Stankovic, "Modelling ofPWM Inverters- Supplied AC Drives and LowSwitching Frequencies" IEEE Transactions oncircuits and systems, Fundamental Theoryand Applications, VOL 49, NO.5 May2002,pp-621-631.

7. Haneyoshi,T.,Kawamura "WaveformCompensation Of PWM Inverter With CycleFluctuating Loads"lndustry Applications, IEEETransactions on Ju1/Aug1988, Vo1.24, No.4,pp.582-589.

8. A. Mechi, S. Funabiki, "Step-up voltage PWMAC to DC Converter with one SwitchingDevice", IEEE Proceedings-B, Vol: 140, No.1,January 1993, PP:35- 40.

9. Sujit K. Biswarup Basak, Mahesh M. S w amy,"A Three Phase Half Controlled Rectifier WithPulse Width Modulation", IEEE Transactionson Industrial Electronics, Vol: 38, No.2.Apri11991, PP: 122-125.

10. Sidney, R.Bowes, Paul R. Clark, "NovelMicroprocessors of New Regular-sampled Harmonic - Elimination PWMTechniques for Drives and Static PowerConverters", Intelligent Motion, June 1990Proceedings. PP:175-180.

11. Zhenyu Yu ,Arefeen Mohmmed , Issa Panaphi ," A review of Three PWM Techniques " ,Proceedings of The American Control

Winter 2011

Conference, Albuquerque, New Mexico ,June1997,PP:257-264.

12. Gyu - Ha Choe ,Alan KWallace, Min - Ho Park,

"An improved PWM Technique OnPower Electronics For AC Choppers", IEEETransations, Vol: 4, October 1989.PP: 496-505.

ODD

A neural network-based CAD model is developed for thedesign of a rectangular patch antenna) which is robust both

from the angle of time of computation and accuracy. A distinctadvantage of neuro computing is that) after proper training) a neural

network completely bypasses the repeated use of complex iterative processesfor new cases presented to it. The single network structure can predict the

results for patch dimensions provided that the values of?r, [l, f2 and h are inthe domain of training values.

Vandana Vikas Thakare and Pramod SinghalNIET Journal of Engineering and TechnologyVol.: I (Autumn 2010) , P 15.