TPS54231 fail after powering up from 12V or 24V

That last diode is going to clamp the output lower than the set point until your super cap charges up.  First thing I would try is to set your slow start time to be longer than the cap charge up time.

Can you post some waveforms of your start up behavior?  Be sure to include VSNS, COMP, PH and if you can the inductor current.

Hello John,

Thank you for your reply, I will send some waveforms tomorrow.

I have already set the startuptime to 10ms (27nF) this didn't help. The datasheet advices not to go above 27nF. But can I do that you think?

I know the supercap beheaves like a "short circuit" at startup but the TPS54231 can handle the kind of short-circuit conditions I asume.

Thanks again!

Xander

Yes TPS54231 has over current protection.  I am not sure why you would be damaging the IC, but I am pretty sure that the 10 msec start up time is a lot faster than the charge time of the super cap.  that will cause the TPS54231 to got to large on time.  during the on time the inductor current will ramp up to the current limit.  If I can see some waveforms I may be able to tell what is happening ( I assume it does not damage on every power up, or does it?).  I think if we can get the circuit to start up in a controlled fashion, it will be ok.

Hello John,

Here are 3 waveforms from  Vout (Yellow), VSNS (Green), COMP (Blue), PH (pink)

2nd picture is a startup with a 12V Vin and the 3th with a 24V input. (10ms timebase)

in the first picture you can see how it blew up. (500us timebase)

I hope you can see any problems.

Hello John,

Short update. I have also tried the 3A version TPS54331D. Same problem, at 24V it blew.

I tried another inductor (15uH). did'nt help.

attaching the supercap to your TPS54331EVM demoboard gives no failure so far.

On the demoboard I see that the lower resistor from the VSENS network is connected to the ground at the compensator network, At my design I did not do this, this resistor is just connected to the groundplane (near the chip) could this cause the problem?

Hello John,

I will check, but I don't think so.

I have now actively pulled down the enable with a switch. So the chip is disabled. Even now It kills itself instantly when I apply any  voltage between 21V and 28V

Hi John,

Altought I couldn't imagine it, there appears indeed a big overshoot on the input capacitors type: MC1210F475Z500CT.

This should be an alternative for the earlier used the MCCA000587, And even though I can't also find any major differences between both caps I now get all these problems.

I have also connected your EVM board onto my PCB at the place where my SMPS circuits starts. I have removed my own caps and the chip ofcourse. Measuring the spike on TI's EVM board show a much smaller spike. So I think I am gonna use the same capacitors as you EVM board. Could any difference in used PCB materials also contribute to a bigger overshoot?

I want to thank you very much for your good help!

Do you have any good recommendations about the input capacitor to use and can you please review the COMP circuit for me? When I do calculations now I come to different capacitor and resistor values. In what way can a bad COMP circuit be harmfull?

best regards,

Xander

Those capacitors are YV5 dielectric.  I recommend only X5R or X7R or better.  I am somewhat surprised that the capacitor makes a big difference on turn on.  Are you hot plugging your supply?  Do you have the same problem if you bring the input supply up slowly?

Hello John,

I am also very supprised about that.

This is the overshoot from my old PCB 

this one is from my new PCB:

and the one below is from your EVM board.

So if this diffence is so big I think farnell may not propose this capacitor as a good alternative for my old one. I am gonna call them later today.

Yes, I have always switched my supply with a small switch. bringing the power up slowly I could go up to 28V with no problem indeed.

I will use an X7R type now. Thanks for the advice. Any recommendations regarding my comp circuit?

best regards,

Xander

It will take a while to evaluate the COMP circuit.  It seems ok as the values seem reasonable, but the best way to check it is an actual loop measurement or maybe a simulation.