Relay post-mortem - am I drawing the right conclusions?

[Ranayna]

Hi everyone,

I have a small question about a suspect relay.

First of all, the device the relay comes from is likely beyond economical repair. But I was tasked to take a look at the device and look for a possible failure reason.
The device in question is an about 4 years old APC Automatic Transfer Switch. This device has two AC Inputs, and can switch between then without interrupting the output. So it is used as fail-over PDU for devices that only have one power supply themselves.

Reported failure mode: Does not switch over anymore, or does not turn on anymore. More details are not known. Could not test myself, since my boss did not want to risk blowing a fuse. (Don't ask   )

After visual inspection there was nothing showing any visible defects. No bulged caps, no burned resistors, no PCB discoloration at all. Except for a bit of dust, looks like fresh out of the factory. Wiring looks high quality, with crimped and isolated spade terminals. The "smarts" of the device are delegated to two sub-modules, one connected with a flatcable, the other module interestingly plugged into an old style 72pin EDO-RAM socked 

Measured the input MOVs, no issues. Two fuses I found are both intact. But those are not inline the power switching anyway.

The input relays are two HASCO HATF902ASAC230 (http://www.hascorelays.com/electromechanical-relays/hat902-series-spdt-30-40-amp.html) for each channel. The coils of those two are directly connected to their respective mains input.
I desoldered one of them and measured it. Coil resistance is a bit above 10k (should be about 15k according to the datasheet). Measuring the NO contact, I got something about 3 Megaohms... so yeah 
After I cracked the relay open, I noticed three things: a typical "burned electronics " smell, black marks on the contact area and a slight sticking of the contacts when I manually close and release them.

My conclusion: The power supplied to at least one of the inputs is unstable, having caused many switching cycles, therefore wearing out the relay. As a consequence, in addition to replacing the ATS (already happened), we should also look into the stability of the circuit itself, otherwise more equipment damage may (or will?) happen sooner or later.

Any other opinions?

[mzacharias]

15K coil resistance seems awfully high for most relays. Should that maybe be 1.5K?

That would be more common for, say, a 24V relay.

[Ranayna]

Well, 15K coil resistance is what it should be, according to the datasheet. I actually measured around 10K on the relay that I dismantled, which is consistent with the other ones. There are two relays in parallel on each input, and I measured 5K on the inputs before I removed the relay. Could the lower resistance be degradation of the relay coil?
These relays are directly connected to mains, so I think they had to choose relays with higher coil resistance, to reduce heating.

I was not able to find a european distributor for those HASCO relays, and was not able to find suitable replacements at Digikey et. al.(mind you, these are PCB mount 40 Amps relays), so this seems to be really beyond economical repair.

[Seekonk]

I don't exactly know what the load is, but it likely operated as long as it was supposed to in that application. It got out of warranty didn't it? Contacts burn for a reason.  I have a hot tub with a 2HP rated relay. Curiously enough it is connected to a 4.5HP motor. Never underestimate the power of cost savings.

[hli]

15K coil resistance seems awfully high for most relays. Should that maybe be 1.5K?
That would be more common for, say, a 24V relay.

According to the linked datasheet, the coil voltage is 230V AC.

[Gregg]

It is very hard to make a conclusion on something that you are not allowed to properly test.  I assume you are troubleshooting one of the small rack mount ATS units that are considered throw away items if they stop working.  Maybe you can rescue it from the dumpster and take it home to properly test it without management looking over your shoulder.

Most of these small ATS units are designed to switch very quickly even though the two inputs may not be in phase.  This requires the control board to monitor both inputs and the output simultaneously while keeping them separate.  While switching, they usually use a fast solid state switch to initialize the transfer faster than a relay can close and then time off after a delay set for the expected time for full relay closure.  It is usually just a time function and not a feedback sense to shut off the solid state switch to prevent smoke release.

There also has to be a fail-safe circuit to prevent the two inputs from becoming parallel in case one relay contact doesn’t break as desired.  My guess is that the designers of the control module use something like fusible resistors that default the whole device to open circuit.   

From my experience with APC, they are big on “features” they consider marketable but not terribly robust in their basic function.  Like most products from mega-corporations they cut as many corners as possible especially if doing so aids planned obsolescence.  Most of these little ATS units are purchased in bulk via a bid process.  Big users of thousands of these probably pay less than half the list price. 

Another thing to troubleshoot would be the power supplied to the ATS.  An intermittent connection could cause it to transfer so many times it simply wore out the relays.  Environmental issues like moisture and high heat may shorten the life considerably.