Inrush current limiter NTC failures, what possible causes?

[userx2]

My final statement: NTC are inappropriate as compressor inrush limiting device. In case of refrigerator compressors, they'll even fight with the built-in starter circuit (either PTC or mechanical contact). Other than rectifier circuits with filter capacitor, motors are already limiting the inrush current by their winding resistance.

I am aware of it and I also read it the first time you mentioned it. Thank you.
It is not clear why the NTC was added as that is now historic. However, we believe it was added to stop occasional triac failures in the past.
I have removed the NTC and we are currently cycle testing without it. So far no failures of anything. Not even when the compressor stalls at low voltage.

@Easy peasy
As for incorrect parts fitted, we do not know. The damaged parts are not readable. They look the same as all the others. I have started the process of failure analysis with TDK but so far it is hard to get through to the old EPCOS division.

--- Updated Feb 23, 2022 ---

@Easy peasy, I just tested it with a dead short to see what happens.
Nothing happened. It kicked in, the system reset and restarted but nothing blew up.
The system restarts and a few seconds later, the cycle repeats. But no damage to anything. Not even the fuse.
I think the problem is our variac can't deliver enough energy for this.

 

[Easy peasy]

Who makes the boards ( and therefore buys the NTC's ) ... ?

 

[userx2]

@Easy peasy and group,
I re-tested the dead short. Our variac has a low rande setting that can go to 37A 120V.
It was spectacular. After the smoke settled, I found.... nothing.
The shorting wire had burned away but the circuit and NTC stll worked.
I increased the shorti g wire gauge a lot and tried one more time.
This time, the unit reset again. But, again, everything is operational and currently cycling again.
If it wern't for these weird field failures, I would say that this is one of the most robust designs I have come accross.

 

[userx2]

UPDATE:
There is still no explanation of why the component fails out there in the wild.

The components have now been traced and verified as original Epcos/TDK.
I have done energy calculations on the actual voltage and current wave forms measured and the component is not overloaded at all in this circuit.
I have done several thousand additional cycles with it in place and nothing happens.

Furthermore, I am also testing without the NTC in the circuit (bypassed) and this is also very uneventful in that the test is at 4000 cycles currently and nothing happens either.

Regards
X

 

[FvM]

I'd add for consideration that the datasheet doesn't guarantee more than 1000 cycles. There may be a large ratio between typical and minimal endurance.

 

[userx2]

I'd add for consideration that the datasheet doesn't guarantee more than 1000 cycles. There may be a large ratio between typical and minimal endurance.

I am aware of this but I was told that this is only a test criteria for the testing and not a guarantee of any kind.

 

[FvM]

Yes, it's only a test specification. But it gives an impression of the intended number of operation cyles. It indicates that each cycle involves considerable component wear.