Odd relay coil failures?

[PwrElectronics]

A former colleague contacted me recently about some failures with relays his current company was having.  Years ago, I had worked on some projects with relays so he thought I may have some ideas.

So, the failure seems to be the coil going open circuit inside someplace.  Very fine wire I was told.  The mfg of the relays has not been able to come up with a cause (my thoughts on that later).

These relays are installed in controllers in outdoor locations on utility installations in numerous states in the US and some overseas.  They are expected to be in service for 15yrs but apparently they are seeing a relative large number of failures.  Not early mortality either, some have been in service for several years.

Relays are through hole on a PCB and supposedly sealed.  I get the idea these are common low cost items.  12VDC coils and they are switching mains AC voltage to a inductive load.  I got the impression the duty cycle of the coils being on/off is fairly low in the sense of how many minutes out of a day, etc.

Things I thought of:

1. Vibration breaking the wire.  But, the breaks seem to be inside the coil, not on the terminations and also the coil is dipped in varnish so unlikely the wire is moving.  Also, there is not likely much vibration in a box on a utility pole.
2.  Poor varnish leading to arcing between coils that eventually opens it.
3.  Insufficient flyback suppression again leading to #2 from too much volts/turn.
4.  strain hardening of the wires if wound too tight and enough thermal cycling.  They get brittle and break
5.  High temp and humidity breaks down varnish again leading to #2.  We've had that happen on products my company makes (with inductors and transformers) but those applications are often 105 or 125C and these relays are not subjected to that.  Probably under 70C given the use.

From the conversation, I wonder if the mfg is putting much effort into this.  I got that the coils are impossible to unwind to find the break and I suggested possibly xray or a ultrasound scan could be used to find it.  Where I work, we could xray something like this as we have a FA lab.

One other thought brought up was concerns the firmware on the controller could get erratic and "chatter" the relay.  Speculation is that somehow over stresses the coils but how?

Best I could advise was to somehow find the breaks if location is a clue and perhaps some sort of accelerated life to test to duplicate the problem but what parameters?

Any ideas in the EE community here?

[schmitt trigger]

To validate or discard the wire overheating theory, you could measure the coil’s resistance while cold and after being powered up a few hours. There will be a resistance increase due to the wire’s temperature increase.

Then, utilizing copper’s TEMPCO of about 0.39% per degree Celsius, you should be able to calculate the AVERAGE temperature rise across the coil. Before anyone complains, yes, I know the hotspot temperature will be hotter than the average. But copper is an excellent heat conductor, and in my experience the hotspot temperature won’t be significantly warmer than the average value. Specifically if you allow some stabilization time.

[floobydust]

There's 50 reasons the coil could fail open-circuit. No part #?
Take it apart and do an autopsy. Otherwise we'll just spin around the possible causes.

[PwrElectronics]

"Take it apart and do an autopsy. Otherwise we'll just spin around the possible causes."

I think my old colleague frustrated that the supplier has not determined the cause when sent failed parts to inspect and looking for other ideas to investigate off the side.   I did not get any details on the mfg part numbers, etc.

After providing them with some thoughts and ideas, wondering if there is anything else one could try checking out.

[floobydust]

You have to use this thing called "science" which requires some tangible data. No failed relays or P/N is beyond my crystal ball.
A list of the 50 causes of coil failure won't fix anything.

[wraper]

The most probable fault location is not inside the coil but where wires are soldered to terminals. It could be lose terminal, vibration, too much wire tension, corrosive flux used during relay production, corrosive atmosphere. Without taking apart and inspecting a faulty part it's a useless talk about nothing.

I got that the coils are impossible to unwind

They can be unwound just fine. Will need quite a bit of time and care due to very thin wire and amount of turns but there is nothing special about unwinding.

[PCB.Wiz]

So, the failure seems to be the coil going open circuit inside someplace.  Very fine wire I was told.  The mfg of the relays has not been able to come up with a cause (my thoughts on that later).

These relays are installed in controllers in outdoor locations on utility installations in numerous states in the US and some overseas.  They are expected to be in service for 15yrs but apparently they are seeing a relative large number of failures.  Not early mortality either, some have been in service for several years.
...
Any ideas in the EE community here?

They need to start plotting age with failure, and an accurate inspection of the exact failure location is needed.
You can carefully probe the side of a coil with a needle, to determine a region of failure. A microscope is also useful.

Most common is stress and corrosion failures, where atmosphere can reach.   

[coppercone2]

of course its possible to unwind it did they pot it in unobtanium?

You just need to dissolve everything thats not copper.

if someone is telling you something is impossible to unpot their trying to bury some shit.

Now it might get messy, but the fact is, you will have 2 strands of copper, making the original length when you put it together . The ends should also show elongation or whatever under a microscope.