IBM 1620

[Tothwolf]

On Sun, 30 Aug 2015, Paul Koning wrote:
>> On Aug 30, 2015, at 1:55 PM, Chuck Guzis <cclist at sydex.com> wrote:
>> On 08/30/2015 09:47 AM, Tothwolf wrote:
>>
>>> If repair of the core memory in CHM's IBM 1620 is ever attempted, I 
>>> think either low-tin solder alloy with 1-2% added copper or possibly a 
>>> more modern indium-lead solder alloy would probably be the best choice 
>>> for the enamel wire to terminal connections since that would greatly 
>>> reduce the likelihood of any future wire breakage.
>>
>> Wouldn't solderless bonding/welding be a better alternative?
>
> That's what the IBM IEEE article mentions.
>
> I'm still a bit puzzled by the dissolving of copper wire by conventional 
> solder.  The wire used in core memories is thin, but not outrageously so 
> by the standards of, say, Litz wire, and that is soldered routinely.

See the other paper I mentioned earlier in the thread: 
https://app.aws.org/wj/supplement/WJ_1975_10_s370.pdf

On page 5, they describe 0.0031" 40 AWG magnet wire being reduced to 0.002" 
after soldering with 60/40 tin-lead solder. That 0.0011" reduction is 
basically 1/3 of the diameter of the wire. This explains the failures of 
the enamel wires in CHM's IBM 1620.

Copper erosion from tin is something I'd never really given a lot of 
thought to, but it will certainly be something I keep in mind from now on.