Gold-Plated PCB-mount Sockets?

We do solder or tin coated pins in gold sockets all the time. No problems in the last 20 years.

Not a problem either. The gold on most pins is so thin, microinches, that it totally dissolves in the solder.

John

Thanks, John.

It's reassuring to hear the results of your experiences. From everything else I had read, mating tin and gold contacts sounded like a very bad idea. But I couldn't really tell HOW bad. Maybe it's not as much of a problem as I was beginning to think it was.

I have just read some threads about mating tin and gold connector contacts, in the Google Groups archive, that imply that the "leaf- spring"-contact types of gold-plated IC sockets are much worse to use with tin or tin/lead IC pins than are the "turned" or machined types of gold-plated sockets, at least as far as galvanic corrosion of the tin- or solder-plated IC pins is concerned.

There's also the thermocouple effect of the two dissimilar metals in contact, creating a small series voltage which might be a problem in high-precision circuits (but probably only something like a 1 uV offset).

And, FWIW, I just found a short Kester article about soldering gold- plated metals, which mentions the sometimes-major effects of different thicknesses of the gold plating, which can be downloaded at:

I also read somewhere that using silver-bearing 62/36/2 solder helps to alleviate some of the problems associated with soldering to gold- plated metals, by preventing most of the leaching of the gold during soldering.

All in all, if possible, it seems best to try to mate contacts of the same metal. But in the case of ICs and sockets, since almost all plastic DIP through-hole ICs have tin or solder-plated pins, is there any advantage in using sockets with gold-plated contacts, to make it worth facing the possibility of eventual degradation from galvanic corrosion (assuming the use of sockets, at all, is a given)?

- Tom Gootee

The main issue mixing sockets is the contact pressure. Gold is soft and requires (withstands) little pressure. Tin is harder and needs some contact pressure to penetrate the surface. A connector designed to mate with gold may not have enough pressure to mate[*] with a tin contact. A contact designed to mate with tin may scratch through gold.

We generally use tin/solder plated cheap, spring-type ic sockets, and the gold ones are the round socket machined types.

I suppose in a very humid environment you could see some corrosion effects form mixing metals, but I haven't personally. For bad environments, we tend to avoid sockets and conformally cost.

That can be corrected by proper layout, making sure all such connections are isothermal. See the pcb pic I recently posted to a.b.s.e, on the subject of slit ground planes. The lower-right region is thermocouple stuff.

John

As had been evidenced by the old HP4191 here in the lab. One sunny day I couldn't calibrate it anymore. Turned out they had pushed fancy gold-plated mil-type PROMs into el-cheapo spring type sockets. Like Porsche Design leather seats in a Yugo. It did work for 20 years but then quit.

Still, if it isn't absolutely necessary to use sockets, don't.

That is yet-another problem with it.

Thanks.

- Tom Gootee

Twenty years? Not too bad! Better than what I'd been led to believe from what I've read about the problem. (But probably not good-enough as an actual lifespan design goal, in my case.)

That is, no doubt, the best solution.

1. For almost all receptacle-to-pin-mating situations, both socket and pin surfaces should be made of the same metal. One exception seems to be that acceptable performance can be gotten when putting a tin or tin/ lead DIP IC pin into a (round) machined or turned gold-plated socket (but NOT a gold-plated "flat contact" or "leaf-spring" type of DIP socket), although, if practical, no IC socket at all would be best.
2. Soldering a gold-plated pin into a PCB, using tin/lead solder, "should" be OK. But using 62/36/2 tin/lead/silver solder should be better.

- Tom Gootee

When I found out what the problem had been I was also surprised that it hung on for 20 years. But from such professional equipment I'd expect the same life span as people expect from aircraft. Something along the lines of 30+ years. IMHO this socket/PROM combination was a design flaw. Oh well, we all make mistakes, nobody is perfect.

Doing that right now. But we have something HP didn't have back in those days: EEPROMs. It's amazing, I just spec'd in a 256k EEPROM for under $2. Nicely SPI bussed. Heck, it even has pretty good over-write protection where you can lock up a part of the address range and leave the rest open for writes.