Clamp dissipation problem TOPswitch-GX

[dj_maxxro]

Hi all,

I am working on a SMPS prototype using TOP250 in forward configuration and ran into some trouble during testing. I used the recommended zenner clamping schematic (same one they used in Engineering prototype reports 12 and 31) . During testing with a load of 70 Watts for a couple of minutes the zenner diodes started smoking upon inspection they were dead ... although the rest of the supply was cool. I have spare diodes but I feel reluctant to smoking those too before I figure out what went wrong.

The power supply was intended for audio use working off the mains 230V +/-15% to provide +/-30V , the transformer is a EER35C core (3C90 material i guess) wound according to the methodology recommended in AN-30 on power integrations website and turns ratio 4,5:1 . Bmax within common sense limits (160mT). TOP250 is operated at 66khz (pins C and F connected together) , current limit set at maximum with 6k8 resistor at X pin . No power limit implemented , just undervoltage and overvoltage protection. Feedback is simple and a little overcompensated maybe ... just a secondary side zenner , optocoupler and some phase lead. 200V Schottkys on the secondary with coupled output inductor.

I have some experience with push-pull and half-bridge supplies , but this is my first forward design. I saw some forum members mentioned using these switchers so I suppose they would know more about this kind of issues. Any ideas would be greatly appreciated !

[star882]

Have you tried an energy recovery snubber?

[wrenchone]

Are you sure your transformer is phased properly? On the positive output, you should be getting positive voltage at the output diode when the TOPswitch pulls down. Also, do you have the catch diodes installed for both outputs?

In a multiple output supply, it's a good idea to disconnect all the output diodes except for the main output and try to get that running first. It saves headaches and head scratching.

[dj_maxxro]

star882,

I am using the clamp they recommended. It's said to "recover most of the energy from leakage inductance and magnetization inductance and return it to the input or deliver it to the output". I have not investigated other clamp techniques because they clamed it is very efficient ( 0,8W power disipation in a 145w power supply - EP 12).

wrenchone,

The transformer is phased properly , i double checked it many times for sanity. Also the catch diodes are present on all outputs and seemed to be working properly.

[m-tech]

What is your output inductor specification ? For your output voltage it should be quite large inductance ( disadvantage of forward converter ). I think that output inductor and magnetization inductance have a big influence on your smoking zenner.
You should experiment with clamping capacitor ( 2.2nF originally ).

Topswitch devices are designed for flyback, and there are only two engineering reports for forward. Also, PIexpert does not support forward converter design.
There is probably good reason for that.

[dj_maxxro]

m-tech,

You are right , my output inductor is quite large at 423uH per output. Magnetization inductance ... well i did not measure that but at 140 turns with no airgap it should be large enough to keep the magnetizing current to a low value.

Actually I found a reference today on PI forums ... earlier versions of PIExpert (5.0 if I recall corectly) supported TOP248-250 for forward applications. It seems the support has been removed in newer versions because it was tricky to get a good working supply. I saw they also do not recommend using this application anymore ... too bad they didn't mention that in the engineering reports would've saved me a lot of trouble

so there we have it ... the evil beast named marketing

I think I will abort this route and try a more conventional 2 transistor forward converter but that would mean even more work to rewind the transformer for maximum duty cycle of 45% ... and the quest never stops