TPS23757: starting issue

Michael, 

Darwin is out on business travel so i'll fill in. For the purpose of this testing you should be able to just short the primary and secondary ground. 

-Artem 

Hi Artem,

thanks for filling-in.

Maybe you have a hint or pointer where to look: When the board is cold, the PoE usually starts fine. Once it starts it runs perfectly.

Whene the board is getting warm, it does not start. The output LED comes on shortly and very dim and I hear audible noises.

When I touch parts of the circuit, e.g. primary or the feedback transformer, I am often able to start the PoE.

BR

Michael

Michael, 

Is this with the TPS23757 EVM or your own board?  

-Artem 

Hi Artem,

it is our own board but the schematic is based on the TPS23757 EVM. 

I attach the schematic here.

We had to use different transformers for space reasons:

Flyback: Wuerth 749 119 350

Gate Drive: TDK/EPCOS B82804A0354A110 

BR,
Michael

Michael, 

Comparing your schematic to the EVM, the first thing I notice is that you used a smaller VC cap. It appears that you replaced a 22uF electrolytic cap with a 10uF ceramic cap.  I don't know what voltage rating ceramic cap you are using, but those have a voltage  coefficient, which makes their capacitance go down with voltage. 

My initial thought is that you are on the edge with the Vc cap and it's barely enough to start-up at room temp and it falls apart over temperature. You are likely to have variation in capacitance, gate charge, and IC's current consumption over temp, which pushes you over the edge. 

Could you please take a waveform capturing VDD1, and VC relative to RTN at room temperature and elevated temperature. This should give us some clues.

-Artem 

Artem,

thanks for looking into this issue. We have already increased the VC cap to 22uF and also did the folowing changes:

R15 = Rclass = 90.9 ohms: equivalent to the EVM

R12 = Rdt = 49.9 kohms: the value of 10k is rather small, I have taken the 49.9 from the datasheet. Or shoudl I use 10k?

R14 = 10 ohms: preventing shoot-through as Luis suggested

What changes should be done in order to account for the changed type of transformers?

I have added 10uF on top of 22uF. Now the circuit starts. And here is the waveform at room temperature of the start with VDD1 in blue on top and VC in orange on the bottom:

It seems VC is rather low at a little over 6V?

BR

Michael