UCC28740 cold start problem

The UCC28740 has no restrictions for use over tempperature as long as you don't exceed the datasheet specification.  The EVM was designed for room ambient conditions (although most parts on the EVM should be okay over temperature there was no extensive effort to meet spec at extremes). 

There is an NTC on the input (RT1) that most likely is causing the scold start up issue.  If you could short the NTC component and test again at cold you may have a better result.  Keep me posted....

Hi, Lisa

Thanks for your help. I have tried to short the NTC, but it still can’t start up with full load. From output voltage waveform, it seems that something triggers IC’s protection.

  Channel 1 : output voltage(5V); Channel 2: input voltage(85Vac)

 

I had an EVM available here so I put it in the temp chamber on Friday afternoon and let it soak at -25C for an hour before turning it on at full load and no load. It turned on without issue even with the NTC still in circuit.
You can determoine if the part is "seeing" a fault if there are only one or three pulses at startup.
Do you have any other probes on the board? Can you look at the gate drive to see the initial pulses?
Do you have another EVM to test? Again, keep in mind that the EVM was not designed to operate at anyting other than room temp (I did not spec a temp range when requesting the transformers for example).

Hi, Lisa

Thank you for all your assistance. I know that it is unreasonable to test EVM under other temp range except room temperature. I have confirmed that components on EVM operated specification (such as Mosfet, diode…) except the transformer as you mentioned. But it seems to be fine when operated in low temperature.

 

My test conditions: put EVM in the temp chamber (-25C), and after one or two hours, then turn on it with low line(85Vac) and full load(2.1A) at first. The output voltage and gate drive signal waveform as bellowed:

 By the way, I only have one EVM...

hmmm, if you zoom in on the first pulses, a fault response would be shown by 3 runted pulses, then no pulses, followed by the 3 runted pulse attempt...from this far away, I can guess that that isn't happening. but what about looking at CS to see what's going on there. I'll commandeer the temp chamber here again to abuse my board a bit more also...

Hi, Lisa

Thank you for the information
I also think that the brown-out function must be triggered. I probed the voltage after the rectifier, and it is true that the input Alum Elec caps lost too much (down to 28V) in low temperature. These days I will try to triple the bulk caps, but can it shot down the brown-out function? It will be a problem to put the larger bulk caps in my design product( compact size requirement) .

Yes, the larger the input cap, the more robust the bulk rail is and the lower the rail can go before triggering brownout.

It seems capacitor manufacturers focus more on high temp than low temp.  Granted, I for one appreciate high temp data but the low temp info would also help out.

You may look at film capacitors, they have a better reputation at low temp, or even putting a 1uF high voltage cermic (or two) in parallel (these come in surface mount so they may be small enough for your application), and I'm not sure if automotive grade aluminum electrolytic caps would be better at low temp. but they may be worth looking into.

I will try my best to find the caps that can work in low temp...
All in all, thanks again for your help.