Help needed with TopSwitch SMPS

smps with top224

I have designed a SPMS using Top-Switch II family from Power Integartions. Output voltage is 12.5V, and the power is 25W. The feedback I use is opto/Tl431 feedback.
My problem is that the switch enters auto-restart mode all the time, getting as high as 3.8V at the output during the "normal operation". It means that the feedback circuit does not work, of course.

I looked at the drain voltage and it looks almost perfect - no noise, even no up-voltage spikes when the switch turns off.

What I see, and seems to be the source of my problem is, that after a regular switch on-cycle, the superseceding 3-4 on

cycles are quite short, one could say minimal, this I think is the reason that output voltage does not go high enough, since

it very soon enters auto-restart. The same happens if I disconnect the bias winding.

The topology I use is from the reference design 204A, only differences are R4 (40k instead), C1 (22u), D2 (MUR820), and C3 (100u). The switch is Top224.

I am in a really great trouble with this, and any help would be appreciated. I did many tried myself, adding capacitors across control and source pin etc. but in vain still

I am attaching the topology of the SMPS

First you have to check all the connections ,what about C2 1000mF is it the same ,then verify the polarity of windings they are critical and third verify the optocoupler (better change it) and take a look at the topology of the PCB there are some criteria to be applyed. One more thing what kind of transformer core are you using -is it good for 100KHz and did you made the windings with minimum parasite inductance?

From my experience with these devices, make sure that you don't have too much leakage inductance in your windings (normally the result of sloppy windings/construction but that will show up as V spikes on the drain). I also found them to very sensitive to the amount of magnetizing current needed within any particular transformer design. It is possible that this will cause the internal current limit to be exceeded prematurely and having the device reset's itself.

One way to find out is to stick a current probe in the drain connection and observe the slope of the rising current over the time interval that the switch is on, and to check for any sudden increase in the slope of the current (indicating possible saturation of the core or in other terms, where the winding starts to act more like a piece of wire rather than a inductance) or as an alternative test, insert a power resistor in series with the drain to lower the peak current. If the unit behaves (it will be at reduced thru-put of course) you know where to look. My guess is the mag current is too high.

One last thing: the star-point ground connections on the source are very important!

I think that the mag current is too high,
maybe it is possible you use the ungap ferrite for transformer, the problem in any case is on transformer.

thank you all guys for the feedback. I have removed the switch from the PCb and will do some intervention on the PCB regarding source connection, hopefully this will be the solution. If not, then I will check the transformer.

Thanx again, I might come up again with results