LM25011: Input capacitors on LM25011 blowing up/shorted

Hi,

Do you have waveforms of the input voltage that you can share?

Switching the input voltage on and off with the large input inductor can be causing higher input voltages than expected.

Thanks, Jason

That was my first suspicion also but I have not been able to detect any voltage spikes on pin 1. We have tried mounting a TVS on the R215 place but one card still broke in the testrig(C200 was shorted).

Here is a picture of the startup sequence. (Measured with 10x probe, working voltage is around 22 VDC)

Hi,

Yes, from that waveform it doesn't look like you are applying too high of a voltage. Check both the turn ON and turn OFF waveform to be sure.

You mentioned earlier that it happens when you power up the card. Is there any type of voltage surge at initial power up? Is the test card being hot swapped or inserted while the power is live? 

Other things that may cause ceramic caps to fail are heating or damage during manufacturing. Make sure the caps are not being stressed or soldered too hot during manufacturing.

Do you have the capacitor part numbers you can send?

Regards, Jason

The card is tested according to the IEC 61000-4-4 (2kV) standard, without problems. It feels like whatever happens to the capacitors is genereated by the card itself.

In the testrig we are currently running all power is cut off before any swapping. When the testrig is running it is basacally just plugging in mains ON and OFF via relays (230 VAC, 50Hz) to the transformer on the card. 
The same way as the cards are used at end customer.

These are the capacitors that get shorted, sometimes just 1 or 2 of them, sometimes all of them. All connected to pin1 on the LM25011
Part numbers:

C200 - C2012X5R1V106K085
C201 - GMK107BJ105KA-T
C207 – CL10B104KB8NNNC
C218 - C3216X5R1V226M160

Power down
Measured on pin1 on LM25011

Hi,

The waveforms that you share look fine. It looks like about 22VDC. Is this the expected voltage after the rectifier bridge? 

Do you monitor the input voltage during testing? Is it possible to monitor the voltage and catch a failure?

Thanks, Jason

Yes, this is the expected voltage. 
We have not been able to catch a fault on the oscilloscope unfortunately. Sometimes we have to switch on the cards 100 000 times before it happens.

Is it possible for the switch to somehow exhaust the capacitors? To the point of burning?

Hi,

The switching circuit should not exhaust the capacitor.

Have you noticed any heating of the capacitors during your testing?

Is there any flexing or stressing of the caps?

Have you requested a failure analysis for the capacitors? I found this capacitor website that discusses capacitor failures. This may be helpful.

Regards, Jason

Hi, 

Just following up to see if any progress has been made with this issue?

Let me know if I can be of any help.

Thanks, Jason

Hi,
We have decided to mount a tvs diode on the R215 footprint for now. Hoping it will help suppress any overvoltage to could possibly break the capacitors. The fault is as mentioned very hard to reproduce so we cannot at this moment verify the solution, only time will tell.