Tin whiskers in Grayhill switches

Last time we sold something into India, it took us two years to get (mostly) paid. The customer was the government power utility.

Guess what happens if anything comes back for repair or recal.

We had a repair from India, (we were partly to blame) it was a nightmare. (Though they are now ordering more.) The "office" hates international orders, India might be the worst.

George H.

Umm, repair takes two years and you over-bill them to compensate for under-payment?

Oh, so that's why you need to unsolder 200 switches.

I had a similar problem with a rotary switch way back in 1980. I think it was made by CTS, but I'm not sure. Nobody knew much about tin whiskers at the time. The company experts, including me, attributed the mysterious shorts to zinc plating falling off the insides of the switches. That was wrong, but nobody had a better guess. We also didn't want to replace the rotary switch so I devised a quick fix. We first removed the rear plate from the rotary switch and Rotex punched two small holes in it. The holes were then covered with sticky paper dots. While the rear plate was removed, we hit the insides with dehumidified compressed air. That blew all the existing tin whiskers out of the switch. We then replaced the rear plate. If it failed in the field, users were instructed to pump compressed air into the switch through one of the holes using a basketball fill needle[1]. That broke loose the tin whiskers again. I later checked with a microscope and found that they were so fragile, that a little air would cause them to crumble into tiny pieces, too small to bridge the contact gaps inside the switch. I didn't stick around long enough to be sure that it worked, but 2nd hand info indicated that it did.

Increase the current. Burn the whiskers away!

That's hard to do when the switch is in India.

And some of the switches are in semi-precision circuits. Trying to get the gain good to better than 1%. How much of the whisker burns away?

Anyway a day spent pulling switches, (my neck hurts.) The redesigned smoke stack (input air centered) worked like a charm. About 30 seconds to get a switch loose. It felt like the center pin was the last to let go.. (I've 2-56 screw across the middle of the air flow.) And I'm also getting real good at clearing solder holes, with my big blue solder sucker.

We are having similar problems w/ Grayhill rotary switches. We produce an i nstrument w/ 7 switches, purchase about 1000 switches a year. Units come ba ck because of failures, 2-5 years old. Internally, a switch contact gets sh orted to the body, so the signal on that contact gets shorted out to the pa nel the switch is mounted on. How is you can disassemble the body w/out trashing it, to find/see the whis kers? Our attempts to open the switch just ends up w/ a mess.

instrument w/ 7 switches, purchase about 1000 switches a year. Units come back because of failures, 2-5 years old. Internally, a switch contact gets shorted to the body, so the signal on that contact gets shorted out to the panel the switch is mounted on.

iskers? Our attempts to open the switch just ends up w/ a mess.

Sorry, I feel your pain. The good news is that this (seems to be) is an ol d problem. Grayhill came out with an announcement a few years ago that they have changed the body shell material. (You can blame Rohs for the original c*ck-up, if that helps.) We ended up replacing all switches with the 'wrong' date codes. (maybe 200 total.) We got replacement switches from our supplier (Digikey) and they worked out the arrangement with grayhill. AFAICT grayhill never admitted to any problem, and we did not seek any reimbursement for all the time spent taking apart and replacing.

I made desoldering tool. I attached a brass cylinder to the air nozzle of re-work hot air station. The diameter of the cylinder just touched the outer ring of the solder terminals of the switch. About 30 seconds of heat was enough to melt joints and allow for switch to be pulled. I can post a pic of tool if that would help.

George H.

I wonder how much current will blow away a tin whisker.

an instrument w/ 7 switches, purchase about 1000 switches a year. Units co me back because of failures, 2-5 years old. Internally, a switch contact ge ts shorted to the body, so the signal on that contact gets shorted out to t he panel the switch is mounted on.

whiskers? Our attempts to open the switch just ends up w/ a mess.

old

ey

nal

to

Right, They are microscopically thin, so not much. I had one user who tired this and claimed it worked. My problem with the technique was that some switches had a huge number of whiskers... you don't want to have to 'clear' the shorts every week, month...

Some also measured as a rather high resistance ~100's of ohms.

George H.

John Larkin wrote on 7/19/2017 12:06 PM:

NASA says an ohm meter can do it.

Sounds to me like a lower impedance circuit won't have a problem with tin whiskers, they just get blown away.

ce an instrument w/ 7 switches, purchase about 1000 switches a year. Units come back because of failures, 2-5 years old. Internally, a switch contact gets shorted to the body, so the signal on that contact gets shorted out to the panel the switch is mounted on.

he whiskers? Our attempts to open the switch just ends up w/ a mess.

an old

they

iginal

ch to

Nasty prickly things... I'm glad you are such an expert. :^)

George H.

Oh? You've encountered tin whiskers? My understanding is you can't even see them without a microscope. How could they even be felt?

ROTFLMAO!

(I used to clear "whisker" shorts on Alumina substrates with a

I have stash of failed switches I was going to send to Grayhill if they ever contacted me in response to my comments on their website. They never responded. Here are the date codes, interesting that there are pairs w/ the same code. We have had no instrument that had 2 switches replaced, each switch below is from a different unit. Failed switch date codes: 1103, 1103, 1424, 1424, 0752, 0752, 1001, 0709, 0709

Paul

te:

duce an instrument w/ 7 switches, purchase about 1000 switches a year. Unit s come back because of failures, 2-5 years old. Internally, a switch contac t gets shorted to the body, so the signal on that contact gets shorted out to the panel the switch is mounted on.

the whiskers? Our attempts to open the switch just ends up w/ a mess.

is an old

at they

original

st

itch to

tin

Right our stereo microscope was enough to see them... you had to get the light just right.

George H.

Ouch! Our bad switches were all from one big order, so we had only one or two failed date codes. We pulled all the switches with those date codes. (1331 and 1306) I sent emails and called Grayhill. I was assured that their switches could not grow whiskers. There are most likely some liability issues and grayhill is not going to give anything away. If you ordered them from a distributor, then I'd suggest going through them. I told digikey about my problem and got ~200 new switches. No one wanted any of the failed switches sent back, which is telling.

Of course ~$10 for a switch does not pay for having an instrument shipped back from overseas. (I was able to just send new good switches to our overseas customers with instruction on how to replace.)

George H.