IGBT collector -emitter getting shorted in 3-phase inverter

[indrayudh]

IGBT gets burned making collector -emitter getting shorted in 3-phase inverter

Hi
I am building a 3-phase inverter using IGBT(IRG4BC30KD) and IR2110 driver.

First the SPWM(sinusoidal PWM) signal is activated in microcontroller. PWM switching frequency is 2 KHz. It is then fed to the gate of the IGBTs(IRG4BC30KD IGBT,which has a 600v Vce, I=28A and internal diode ) of the inverter bridge circuit using the IR2110 as gate driver . Then the DC rectified voltage is increased and fed to the three phase inverter. The voltage across each phase is checked with multimeter (BK PRECISION-Model no: Test Bench 391A).A balance 3-phase voltage is shown but the multimeter is making a buzzer sound similar to continuity test, though the function switch of the multimeter is not set at continuity measurement. After the line voltage across each phase has reached 90 volt, suddenly two branch of the inverter bridge burns (the high and low IGBTs, become a short: collector to emitter R = 0). This cant be caused by over current, because no load is connected across the inverter, and the IGBT are rated for 28A @ 25ºC.

Query:
a) Distance between the upper and lower IGBT in a branch of inverter is 16 cm. Details is shown in figure. Will that cause problem?

b) An RC series circuit is added as snubber for each IGBT.The capacitance is 0.33 UF (polyester type) and the resistance is 100 ohm (wire wound type).Detail is in figure. Is that correct?

Pls see the attachment : prob_int.pdf.
I have no idea .pls help me.
Indrayudh

 

[chuckey]

I am unclear about the circuit. The transistors as shown, they provide one phase of 230V three phase made by PWM at 2KHz. i.e. they are 2KHz 230V pulse amplifiers?
The reactance of the .33MF at 2KHz is about 250 ohms that and the 100 ohm resistor then give an impedance of about 270 ohms, so there is a fair amount of AC current at the 2KHz and and at 6Khz Xc = 80, total impedance 120, 3 Xf, 1/3 X V, so the third harmonic current will be about 80% of the fundemental. So it could be marginal if there is no heat sinking of your devices.
It could be that the transistors are oscillating in the VHF region, Try ferrite beads on the gate leads and a .01MF disc capacitor across the Vcc line with short leads to earth.
Are you sure that your drive pulses are square so the transistors transition from on to off quickly?
Frank

 

[mtwieg]

Do you notice any heating in the IGBTs as you approach a failure? You really shouldn't see any significant heat if there's no load...

 

[Orson Cart]

Re: IGBT gets burned making collector -emitter getting shorted in 3-phase inverter

Firstly you have quite a bit of power in your snubbers at full line volts, 2 * 0.5*C*V^2 * f = 70watts in each snubber resistor if the igbt turns on reasonably fast with a 325VDC bus, also wirewound resistors aren't a good choice for snubber resistors if you want to alleviate the higher frequencies that can occur.

Check that you have plenty of dead time in your gate drive, so on each leg both IGBT's are off for 500nS say (or more) before the next one turns on - if the dead time is too little you will get some element of shoot through due to the tail current in the igbt's, and this (the high current impulses from shoot thru) is most likely causing RF interference that is making your volt meter misbehave (not uncommonly observed) and causing the eventual failure of your devices. The shoot thru RFI can get back into your controller and/or gate drive ckts also and give extra problems as you raise the bus volts. THe fact that you get these problems as you wind up the variac points to some type of shoot thru.

Layout is fairly important here even at 2kHz, you should try to have all the devices close together, with bus capacitors close to the bridge, with 1 x 4uF good quality cap (wound polypropylene) close to the bridge.

When you have fixed the above you should be able to reduce your snubber caps to 10 - 47nF, and the R to 470 ohms approx.

Hope this helps, Regards, Orson Cart.

 

[indrayudh]

Thanx orson
pls answer my querry.
u have written wire wound resistance is not a good choice .Then what type of resistance ,U suggest? u have also mentioned to take 470 ohm....what will be the wattage?

U have mentioned about checking dead time.How I will increase the dead time in the driver ? If i increase the gate resistance will it be benficial? I have used a 47 ohm resistance anti-parallel with Schottky diode(SB560) at the gate of each IGBT( refer to the figure in the attachment of my previous post).The 47 ohm value i have got by considering platue voltage of IGBT [Vce(sat) ] as 4 volt for 0.5 amp.The minimum gate charge required for this platue voltage is considered as 30 nc. Refering to Application note (AN9876), i have calculated the turn on delay as 120 nsec.

DC Bus capacitance ,i have used around 1000 UF electrolytic capacitor...........Is it not ok?
Pls reply

 

[Orson Cart]

you need to find some low inductance resistors, use the 470 ohm only when you have addressed the issues in my previous post (5 watts) you need to design some part of your circuit that will let you increase the dead time, as an interim step you could try using 150 ohm gate drive resistor with the back diode to give more dead time than you currently have. (120nS is too short for IGBT's when they get hot.)
Regards, Orson Cart.