On Tue, 18 Dec 2012 20:37:04 -0800 (PST), whit3rd
wrote:

On Sunday, December 16, 2012 6:24:58 AM UTC-8, Jeff Liebermann wrote:

...
contact preservative, which is just a coating of oil or grease. You
can do as well with almost any thin grease. However, note that all of
them are non-conductive, so adding grease increases the contact
resistance slightly.

Not entirely true; some conductive greases ARE available, including
transparent ones that aren't easy to tell from 'normal' grease

http://store.caig.com/s.nl/it.A/id.1589/.f?sc=2&category=185

Bad example. There are carbon doped greases that are used mostly to
dissipate static electricity between sliding surfaces and to lubricate
mechanical switches. It doesn't take much resistivity to be
considered "conductive".
http://www.mgchemicals.com/products/...e-grease-846/?
117 ohm-cm bulk resistivity. There are also some silver doped
greases. I have no clue what those are for, but they should have
better conductivity than carbon doping.

...this applies only to tin and
silver contacts, but not gold, which doesn't tarnish, oxidize, or
require lubrication.

If gold really DIDN'T oxidize, it'd weld to itself on contact.

As in cold weld? That does happen, but the adhesion forces to the
nickel base metal is stronger than the wiping forces, so the gold
stays in place on the contacts (unless the plating is really soft and
thick). Here's the rules of the game for gold:
http://www.te.com/documentation/whitepapers/pdf/aurulrep.pdf

There's
no crust of oxide, not even a micron thick layer. But, there's a nanometer
of oxide, all over any gold surface.

Gold oxide is an insulator. I did some googling and couldn't find any
references mentioning such an oxide coating on gold. Certainly
aluminum is protected by an oxide coating, but methinks not gold. To
the best of my knowledge (at this late hour) gold's primary attribute
is its resistance to oxidation.

At elevated temperature and humidity,
even a clean gold/gold connection will fail, because something grows
on that gold surface.

Nobody plates contacts with pure gold. Nickel and Cobalt are added to
make "hard gold". There's also an under-plating of nickel, on which
surface the gold is plated. If one gets the contacts hot enough, the
nickel will diffuse through the gold plating, get exposed to air, and
oxidize or tarnish forming nickel sulfate. If the gold has a greenish
tint, you have nickel sulfate. That might be what's happening.

We lowered the storage humidity spec and our PC-based
product stopped getting memory and video and PCIe errors at the environment
test lab.

When I was building marine radios, we used big 0.156 gold edge
connectors for everything. Dry loads, high power, DC, RF, whatever,
it all used the same 50 micro inch gold plated connectors. We had a
few problems, but never any env test failures. I even built a machine
that would insert and retract the cards repetitively until the
connection showed some "noise". I gave up after about 10,000 cycles
and no noise appeared. Inspection under a microscope showed that none
of the hard gold had migrated.

Lubricant might not be irrelevant, after all, on gold contacts.

Maybe, but I would want to know the failure mode of your PCIe
connectors before I passed judgment.

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558