[SOLVED] Full bridge PWM transient

[xeratule]

Hi all,
I am designing a 25khz pwm triggered full bridge circuit to drive inductive load. Please see the attached circuit. High power IGBT's are all N channel and high side IGBT's utilize from the bootstrap circuit. The pwm signals are amplified and isolated from the control circuit by the optocouplers (optocoupler outputs are CMOS). Although the circuit operates well, at high power levels one or more of the optocouplers (usually ones from high side) get destroyed.

I can't control the high frequency transients at pwm pulse edges at both positive and negative polarities. When the VCC (high power) supply voltage increases transients also gets higher. In order to protect optocoupler outputs I tried to use both varistors and TVS diodes between optocoupler supply and references.

Do you think the transients destroy the optocouplers? Is there some way to suppress those transients? Maybe my TVS diodes aren't fast enough?

I welcome those all helps. Thanks in advance.

Erhan

 

[mtwieg]

Do the optocouplers get hot before they fail? After they fail, does the isolation barrier still hold?

It may be transients on the outputs of the optocouplers, not their supply rails. Try putting a TVS across the output of each optocoupler (from output to the emitter), as close to the opto as possible. Also put a schottky from the outputs to the positive supplies. Also I've seen many app notes advise a small resistor in series with the bootstrap diodes, so that reverse recovery currents are reduced and don't screw with the supply rails.

Finally, keep in mind that although your low side drivers nominally at the same voltage as the bridge ground, high transients can cause them to jump several volts, potentially destroying those drivers. Very large drives generally usefully isolated drivers on all the switches.

 

[Orson Cart]

Since you are switching above the resonant freq, you need to have diodes across your power devices (if they are not there already), what is the voltage your opto drivers can withstand? Regards, Orson Cart.

 

[xeratule]

Dear mtwieg,

Do the optocouplers get hot before they fail? After they fail, does the isolation barrier still hold?

Optocouplers doesn't heat at all, they suddenly fail. After they fail only their outputs (CMOS) are destroyed so isolation still protects the controller circuit. TVS are also still operational. I'll apply your suggestions and reply the results here soon.

Dear Orson,

Since you are switching above the resonant freq, you need to have diodes across your power devices (if they are not there already), what is the voltage your opto drivers can withstand?

Actually I try to operate at the resonant frequency and where should I exactly use the diodes at? Optocoupler supply is max 25V and outputs have 15kV/us common mode transient immunity.

 

[Orson Cart]

Diodes from C-E on the igbt's are essential if you are near or above resonance, it appears from your drawing that you have none, try changing the TVS to 15V zeners (1W or bigger) to see if this will stop the failure of the opto-couplers, Regards, Orson Cart.

 

[xeratule]

I applied the diodes between C-E of the IGBT's and used TVS diodes both for optocoupler outputs and their supply, as a result that seems to solve my failure problem. Thank you very much for the replies.
My best regards,
Erhan