Appliance failures - relays

Interesting failure mode, though I'd be really surprised if your Acura didn't use nice brittle lead-free solder.

Best regards, Spehro Pefhany

I don't know for sure--did they use RoHS solder in 1986? I fixed it by reflowing the connections; they melted easily, like good old-school 63/37.

The appliance repairs were likely Pb-free, so I was surprised to see the same failure.

The failure looks like this:

Another:

In my Acura the joint would go open, until a lucky vibration made the contact. That closure would spark and temporarily weld the connection, until more temp cycling broke it loose again. Rinse, lather, repeat.

Cheers, James Arthur

It's an '86? I thought I got one of the first ones at 1990. Usually the salt eats them in about 10-15 years around here.

I doubt that would happen with a double-sided board and perhaps even if the holes were a bit smaller.

I've seen this exact thing happen with PCB-mount transformers on single-sided boards.

Best regards, Spehro Pefhany

'87, built in '86. Still runs fine, 41 mpg in the summer (it likes the heat).

It was a royal pain to troubleshoot. I had the board on the bench and knew by elimination where the open *had* to be, but it measured perfect, milliohms.

Energized, the relay worked flawlessly. Then I happened to leave it on and the connection magically went open after a couple minute delay. Thermal. The relay had a 2w coil, IIRC, heating things up. Gently blowing on the joint made it toggle.

An alternate theory is vibration. That could be, but it doesn't explain the identical-looking appliance board failure. The appliance board was SMD. Double-sided IIRC (but I'm only 65% sure).

Cheers, James Arthur

Do you think that it's because the relay is putting an axial load on the lead, or because the board and solder is expanding radially around the lead?

In the former case, you may be able to mitigate the problem with cleverly bent relay leads. Perhaps.

these pics show the typical failed solder joint you'd see around relays, switches, connectors - anything that can move or has thicker leads. Unless the damage is severe, retouch it with an iron and a little fresh solder and you're good. I always hit the other joints on the same device or anything else that looks like it's next. It's just a matter of time until more joints fail, and some may already have cracks you can't even see yet.

I completely unsolder any joints that shows signs or arcing, clean the area and resolder if the component lead or connector block isn't damaged. Board mounted Molex connectors are always suspect.

My guess is it's something with stress in the joint from assembly plus stress from thermal cycling. This sort of cracking seems common with thick leaded devices where assembly was sort of skimpy on solder to start with. also, relays, and connector blocks can be pretty stiff as a whole and may not expand and contract at the same rate as the PCB itself, and lets face it, solder isn't exactly the strongest stuff out there and is known to fatigue and break all the time.

I have also seen it on line-output transformers, multi-pole high-current connectors and turret tags. My car had an intermittent connection of this type in the main ignition relay. Luckily it failed in a quiet area and I realised I hadn't heard the familiar 'click' when I turned the key. I thumped it and temporarily restored the connection, then dismantled it and re-soldered it when I got home.

The shape of the joint seems to be the main factor. Large disc-shaped solder joints on single-sided printed circuit boards seem to be the most prone to this type of failure. Spherical joints, such as those on solder tags and hollow solder buckets ("Mitcham tagstrips"), don't seem nearly so badly affected.

I think "yes." The Acura's failure (opening and closing of the connection with hot/cold, respectively) was visibly from axial motion of the pin under thermal expansion. Since the pin was demonstrated pretty well beyond any doubt to be moving vertically, it seems likely that's also the cyclic motion / stress fatiguing the solder joint.

Good idea.

I wish I'd taken a few photos of the Acura failure--I think you're right, and I think the relay pins were straight, no bends.

That would create an effective axial TCE mismatch with the solder, since a given ppm over a long pin creates a longer displacement than within the joint itself.

If all pins heated at the same rate the relay would simply rise off the board under heating, but that's not what happens. The Acura relays' pins are the opposite of isothermal. The coils ran pretty warm, causing some pins to expand and punch through their solder joints first, while the other relays' pins held each relay down.

That axial motion is what made and broke my failed unit's connections on the bench.

A strain-relieving bend would absorb that cyclic force; a cooler-running coil would help too.

Cheers, James Arthur

True, but the effort here was to spread ideas for better design practice. We've noticed the high failure rate of appliance electronics; maybe we designers can take note and do better.

I just reflowed my unit, adding good ol' 63/37. I removed one of the two relays for inspection, checked & cleaned the contacts, found nothing wrong, then re-soldered with 63/37. The original assembly lasted 15 or 20 years, so the repair should outlast the car.

Cheers, James Arthur

A very important, and often overlooked, point concerns the orientation of the relay. It must be mounted so that the gap between the points is vertical if it is to give the longest, trouble-free, life.

This was well known to telephone exchange designers in the 1920, but seems to have been forgotten. Despite what the relay manufacturers might tell you, it still applies.

I don't have enough data to know, but your idea is consistent with a few other observations. A flat, thin joint would be axially weaker, and more susceptible to axial failure.

I observed axial displacement in my sample.

And Spehro suspects a double-sided board with smaller holes would've been tougher. I suspect he's right.

All those relate to axial strength--a double-sided joint with nice fillets on both sides and a PTH hole be a lot stronger against axial forces.

Cheers, James Arthur

Something like this:

Regards,

Boris Mohar

Got Knock? - see: Viatrack Printed Circuit Designs (among other things)

void _-void-_ in the obvious place

Interesting--I'd not heard that. Any rationale for it, or just empirical evidence?

On the axial-punch-thru, a soft pad under the relay would also allow some vertical movement without strain on the pins. Might be worse for vibration, though.

Tim's bent-leads idea is better, a more general solution.

Cheers, James Arthur

A cooler-running coil translates directly into more copper expense, so that's probably out for most automotive and appliance applications.

A bend in the lead someplace, or, if you're ordering relays 100,000 at a time, some magic in the plastic relay frame that allows the pin to float a bit, would do the trick.

The double sided boards with plated though holes or eyelets for stuff like relays are bulletproof from what I've seen. Appliances are just junk these days with no attempt to create any sort of reliablity. There's no way the designers are not aware of these problems.

It's definitely not out for us, though.

I bet auto designers would take it into account too if they knew. This assembly (main relay) stops the car dead when it flakes.

Yep, that's the best.

Cheers, James Arthur

The dirt and oxides, displaced by the wiping action of the contacts, work their way out of the gap instead of staying there as resistive particles which grind away the contact surfaces and promote arcing. Even in sealed relays, oxides can form if there isn't an inert gas filling.