as i increse switching frequency of igbt in inverter my output ac voltage decreses

[piyush_bhatt]

hello friends,

i am working on three phase inverter. i use pic18f4331 microcontroller for pulsing irfp450 mosfet, i use ir2112 as mosfet gate driver,

my experimental result is as bellow..

Fpwm ----------- Vdc(i/p)-----------Vac(o/p)----------THD----------sinusoidal---------- no.of pulses/half cycle
====================================================================
5khz ------------ 400v ------------- 240v ---------- 20%---------- NO ------------- 9
10khz ----------- 400v ------------- 240v ----------- 20% ---------- NO ------------- 15
20khz ----------- 400v --------------160v ----------- 14% ---------- YES -------------- 33


my dream output is as below....

Fpwm ----------- Vdc(i/p)-----------Vac(o/p)----------THD----------sinusoidal---------- no.of pulses/half cycle
====================================================================
20khz ----------- 400v -------------- 270v ----------- <5% ----------YES -------------- 33



so what should i do??

 

[mtwieg]

First you should check that the output of the IGBTs looks okay at all frequencies (fast transition times, full amplitude, etc). Assuming that's not the issue, then it sounds like a software issue. Just look at the PWM signals coming out of the PIC and see if their duty cycle is spanning near 0% and 100%. Also you can put an RC filter on the PWM signals and you should get sine waves, like a preview of the inverter outputs.

Also what frequency AC are you trying to produce? Your number of pulses per half cycle don't make sense for 50/60Hz output. You seem to be generating something around 300Hz.

 

[piyush_bhatt]

first igbt is got heated at 20khz frequency, my duty cycle variation is = 10% to 98% at all above switching frequency.

the output frequncy of my inverter is 50hz that i saw on power analyze. the variation on o/p frequency is less then 1%.

my problem is as i increse switching frequency my igbt get hited and o/p voltage also decrese considerably but i can get sinusoidal o/p voltage.
then if i decrese switching freuency to 5 - 10khz my igbt is not get hit up and i get more ac output voltage as i wish but it is square wave.

so i want to get output voltage sinusoidal with 260vac output at 370vdc.

why you tell me to put r-c filter on pwm signal line. is it put before driver ic(ir2112) or after.

 

[mtwieg]

the output frequncy of my inverter is 50hz that i saw on power analyze. the variation on o/p frequency is less then 1%.

Well then there's something wrong about the tables you posted. For example, 33 pulses per half cycle at 20KHz results in 20000/(33*2)=303Hz.

my problem is as i increse switching frequency my igbt get hited and o/p voltage also decrese considerably but i can get sinusoidal o/p voltage.
then if i decrese switching freuency to 5 - 10khz my igbt is not get hit up and i get more ac output voltage as i wish but it is square wave.

The IGBTs getting hotter at a higher frequency is not surprising. Whether the heat is at acceptable levels depends on the details of the inverter design (IGBT types, gate drivers, snubbers, layout, etc). But it should not cause the output voltage to droop, so long as the DC voltage does not droop.

Again, if changing the PWM frequency causes the shape or amplitude of the sine wave to change, then I suspect a software problem. Does the droop only happen when the inverters are loaded? Can you post some waveforms?

why you tell me to put r-c filter on pwm signal line. is it put before driver ic(ir2112) or after.

Filtering the PWM with an RC filter would just be a small experiment to see if those PWM signals are actually representing a good sine wave before being fed to the inverter. So you would disconnect the PWM signals from the real inverter, and instead attach them to filters with similar properties to the AC line filter you're using. If the SPWM signals are good, you should see good looking sine waves on the outputs of the filter, and their shape/frequency should not significantly change with PWM frequency.

Consider that the inverters are really nothing more than very high power digital logic gates. They should perfectly reproduce the PWM signals given from the MCU (at higher voltage). Filtering either the small PWM waveforms or the large inverter outputs should give waveforms with identical shape (but at different amplitudes).