The Mosfet is burned and current sense resistor sparking

Hi all,
I have to design the 30V SMPS (the SMPS goal is 30V 7A) using On-Semi NCP1937 and NCP4355 to form a flyback SMPS. For the first try with no load, it is ok, voltage is shown 30V, then provide the 2A loading by e-load, it also ok and can see the PFC is working.
However, after on and off the e-load few times, the current sense resistor(R20,R43,R23,R24) is sparking, and MOSFET is burn (since cannot observe from the outlook, just check by the multi-meter ).

I have attach my schematic for reference, I want to ask, is it relate to snubber circuit? I doubt that the back emf break through the MOSFET make it short circuit cause the current sense resistor sparking. Is it correct?

Thanks all!
View attachment 180Wsmps.pdf

Many of the components are marked "NC" and they are placed where alternative components can be fitted, for example synchronous rectifiers or simple diodes. It would help to know which are fitted in the model causing your problem.

Brian.

My hunch is that it's something with the synchronous rectifiers. The NCP4304 is specced to work with a QR flyback, but the way you're using it is very different from the datasheet. I really hope that the arrangement of F4, D8, and D22 is a typo, otherwise that's guaranteed to blow. If you remove F4 entirely, does it still work?

Also 1N4148 is a poor choice for high peak currents in gate drivers.

betwixt said:

Many of the components are marked "NC" and they are placed where alternative components can be fitted, for example synchronous rectifiers or simple diodes. It would help to know which are fitted in the model causing your problem.

Brian.

Click to expand...

Thanks your comment, since I want to step by step to develop the smps, so I based on the evaluation board to start develop, In the beginning, you are correct, to observe much clear, I modify the schematic and delete those NC component.

mtwieg said:

My hunch is that it's something with the synchronous rectifiers. The NCP4304 is specced to work with a QR flyback, but the way you're using it is very different from the datasheet. I really hope that the arrangement of F4, D8, and D22 is a typo, otherwise that's guaranteed to blow. If you remove F4 entirely, does it still work?

Also 1N4148 is a poor choice for high peak currents in gate drivers.

Click to expand...

Yes, the NC marked are no component at there, since I am the newbie for SMPS, to avoid too complicated circuit, I just seperate 3 section to construct the SMPS, 1: flyback, 2: PFC, 3: synchronous rectifiers, at the first try, the flyback is ok, can regulate to my desired voltage and functional by e-load, next step I add the PFC MOSFET and transformer to form PFC, when first and second try, it also ok both voltage range and current consumption. However, I just on and off the e-load two or three times, the current sense is blowed.
I will add back the synchronous rectifier if it is must.
For the diode, did you mean D3 and D18? any suggestion to change? or use 1N4004 instead? thank you!

In most of the cases, a Schottky diode would be a better type to use, 1N5819 or similar. The 1N400x series are not suitable, their storage time is too long, meaning they can't change from blocking to conducting mode quickly enough and they start to behave like resistors.

Thank you for posting the updated schematic, in the original there were some conflicts of components so it was difficult to make a diagnosis.

Brian.

If you're blowing out sense resistors R20, etc., then F6 must not be turning off soon enough. The way these power supplies work, if the MOSFET stays turned on after the transformer has saturated, there's a short circuit. You're always a few microseconds from burnout.

Put scope probes on the gate of F6 and the top end of R20. The gate voltage should look like a square wave, of course.
You should see the voltage across R20 ramping up on each cycle. If the voltage across R20 stops increasing before gate turn-off, you've saturated the transformer and are close to burning out.

You say this will work with no load, so try it with no load, or a very light load, first. Check the maximum voltage across R20, calculate the current through the sense resistors, and calculate how much power they're dissipating at idle. The power should be much lower than their rating. Calculate the maximum safe voltage at R20 before the current sense resistors burn out.

Increase the load a little and watch F6's gate and R20. The duty cycle of F6's gate should increase slightly, but the maximum voltage across R20 should not get close to the maximum safe voltage. Since you're reporting burnout, there's some load level where you exceed that limit.

Once you've seen how the control of the MOSFET is behaving, you have to figure out what all that control circuitry is doing wrong. It's all about controlling the signal at the MOSFET's gate properly.

Thanks Nagle and betwixt explanation and suggestion, I will try to modify it and trace the problem step by step.
Beside, I want to ask how can I know that is it relate to snubber? how can I know the snubber is sufficient /not enough to suffer the back emf?
thanks and sorry I have so much question.