Connectors: Both contact surfaces must also be the same material?

Sun Jul 10 14:40:05 CDT 2016

actually  some  DEC  backplane  had  gold dos  inside   on finders of 
and in one instance we had a 8i that has all gold plated everything on the  
backplance and heavy  gold  too.
back in the  days - -  Ed Sharpe retired  CEO Computer  Exchange Inc  Phx
In a message dated 7/10/2016 12:30:30 P.M. US Mountain Standard Time,  
tothwolf at writes:

On Sun,  10 Jul 2016, Jon Elson wrote:
> On 07/10/2016 12:07 PM, Paul Koning  wrote:
>> On Jul 10, 2016, at 9:07 AM, Tothwolf  <tothwolf at> wrote:
>>> On Sun, 10 Jul 2016,  Paul Birkel wrote:
>>>> Stated Tothwolf  tothwolf at
>>>>> "Both  contact surfaces must also be the same material or tin oxide  
>>>>> will form on the surface of the gold plating and  cause a major 
>>>>> headache. This was a serious problem  with 486 and earlier Pentium 
>>>>> PCs with 30 and 72 pin  SIMMs and it led to a number of lawsuits."
>>>> Almost every  DEC System Unit ("backplane") that I've ever seen uses 
>>>>  tinned-contacts, yet the Modules all use gold-plated fingers.
>>>  I'm not familiar with them used in DEC systems in that way, but the  
>>> problems with mixing tin and gold plated connectors is well  
>>> documented. Even the connector manufacturers warn against  mixing 
>>> different platings.
>> While "don't mix  contact surfaces" is sufficient, it isn't necessary. 
>> What matters  is the "anodic index" of the metal, or rather, the 
>> difference  between those two values for the two metals in contact.  If 
>>  that difference is large, you have a problem; if it's small enough, you 
>> do not.  "Small enough" depends on the environment; aboard  an 
>> oceangoing ship the number has to be smaller than in an office  setting. 
>> I remember looking into this topic for an investigation  of what types 
>> of contact platings are acceptable for lithium coin  cell battery 
>> holders in IT equipment.
> This  applies to bolted contact for structural things.  Gold connectors  
> usually have light contact pressure to preserve the soft gold  plating. 
> Tin contacts usually have higher contact force to scrape the  oxide off 
> the tin surface.  When they are mixed, the tin can  wipe onto the gold 
> and then allow oxides to form due to the lower  contact force.  Tin 
> contacts are supposed to provide enough  pressure to form gas-tight 
> contact areas.

Another thing to  keep in mind here is that electrical current is being 
passed through the  junction. Mixed metals greatly increases the potential 
for  electromigration.

> And, of course, when exposed to salty air, then  everything goes downhill 
> REAL fast, corrosion galore. In a salt  environment, I'd use 
> semi-hermetically sealed connectors, and still  expect lots of problems. 
> The Navy probably knows a LOT about these  things.

Even in a reasonably good atmospheric environment weird issues  can crop 
up. I once evaluated an air handler controller which had worked  perfectly 
in product testing, but once field deployed, had a very high  failure rate. 
It was made up of two pc boards with a pair of .100" pin and  socket board 
to board interconnects. The two boards were physically held  together with 
4 nylon snap-in standoffs. The lower board contained  terminal blocks, 
modular connectors, and the power supply circuitry and  the upper board 
contained the microcontroller, network circuitry,  etc.

The cause of the failures turned out to be fretting corrosion of  the board 
to board connectors caused by vibration. Another contributing  factor was 
that many installers were not installing all 4 mounting screws  when 
mounting the controller inside the unit (these were field retrofitted  
controllers) but were instead only installing 2 screws in opposite  

The fix was to replace all of the existing board to board  interconnects, 
both the header and socket with parts that had 30  microinches of hard gold 
over nickel (the original parts had 15  microinches of gold) and to use a 
contact lubricant during assembly.  Repaired boards were also to be 
installed using all 4 mounting screws. The  vendor later redesigned the 
controller so it was all on a single board  (while still admitting no 
fault, of  course).

More information about the cctalk mailing list