Troubleshooting a cycling SMPS


Joined Jan 30, 2016
measure its resistance.

It might be instructive to measure chip pins when MOSFET not active as well as with it active...

Thread Starter


Joined Jan 18, 2016
Good news, it turned out a resistor on the source side of the MOSFET was open. It was a 0.47ohm resistor. A 2.2ohm was parallel to it which was fine. The power supply came back to life after replacing it. I put a load this time before powering up as I didn't want to take any chances. The P/S installed in the unit and back with its owner.
I have got some further questions though:
1. Why did the UC3842 blow itself up and the MOSFET but the A3842 didn't do that? Is it somehow superior?
2. Could a broken aux winding caused the IC & MOSFET to blow up?
3. Could not having loaded down the outputs caused the blowing up of the above mentioned components? Is it a good idea to always load down the P/S when troubleshooting. Any advice on what to get for a multiple variable dummy load.
4. What is the purpose of the point which is connected to the mains earth via capacitor found on many SMPS?

Thanks again everyone. I have learned so much from this post.


Joined Jan 30, 2016
the 0.47 was the current sense resistor. It probably got taken out with the chip when the MOSFET went s/c and put 300+ volts across it. Fortunately it didn't fry the transformer windings, but I suspect it might have been close...

The broken aux winding may have been the original issue. with no vcc supply from there the controller was unable to regulate effectively. Possibly as time went on things got more stressed until something else gave way...

It shouldnt be necessary to load outputs. An SMPS should run indefinitely with light loadings.

The capacitor to protective ground is called the Y-capacitor. Its there to dump HF switching noise to ground and helps with EMC approvals.


Joined Jan 27, 2015
If it's a circuit like the picture, the lower aux winding, get powered magnetically, from the main upper windings, that are all on the same transformer/coupled inductor core.

So if that wire breaks/burns out, the chip loses main power, and then might get stuck in the startup phase from the statup resistor, as someone else said.

If they lose power completely, so the whole things unplugged, those little chips will just keep working until the power drops so low that stuff starts turning off. I don't think they ever damaged just losing power, it's just glides to a halt as the caps de-charge.

1 of the easiest repairs I did, was on a PSU marked broken, so without powering up, I just started checking w/ohms/diode mode, made my way into the circuit from mains, and found that a wire had broken off 1 side of the auxiliary transformer. Soldered it back on and it was perfect again, sold it and never had a problem since.
Last edited: