[time-nuts] HP 5370B vs SR620

[Chuck Harris]

Hi John,

I know that socket problems in the 1970's could be a real
bear, but most of them started very early in the life of
the product.  For instance, I had heard about smacking
Apple I memory on the table to fix problems by the time
they had been out for only a couple of years.

Maybe my 5370A already has replacement sockets?  I'd have
to look, but since it is working reliably, I am not feeling
really inclined to de-rack it to check.

I have used gold plated machined-pin sockets for many years,
and never saw a failure due to the gold/solder interface.
I am certain that to some degree the problem is real, but
it is also not real most of the time.  Stressed joints are
certain to be a contributing factor.

If you look carefully at the augat machine-tool sockets, you
will likely notice that there is a tunnel that runs under the
socket so that you can strap the IC into the socket.  A good
idea much of the time.  I have never had a plastic packaged
IC fall out under any circumstances, but I can see where the
more massive ceramic packages (eg. 5400 TTL) might be a real
problem.

It always bothered me a little bit how the fingers in a machined
pin socket are not in any way aligned with the faces of the IC
pin.  They are clearly designed for a round pin!  ... and are
being used on a square pin with random orientation.  Probably
not an ideal situation.

When I bought my 5370A, I also bought an 8082A so that I could
service it.  The 8082A is about the only generator that is up
to the task.

It was a major disappointment when the 5370A turned out to be
working well, and in specification.  Hopefully it will drift
out so I can put in the differential linearity modification,
and calibrate the beast... in the mean time I will just use
it.

-Chuck Harris

John Miles wrote:
>> As to the socket issue, my 5370A has been dead reliable.  No problems at
>> anytime.
>>
>> I think your socket issue is unique to your unit, or perhaps the
>> series your
>> unit came from.
> 
> I've heard of socket failures happening on at least one 5370A besides
> Didier's, but I don't think it's reached pandemic proportions yet.  Seems
> that the socket manufacturers didn't really understand their own reliability
> figures back then.  I've been bitten by them myself -- not by my 5370, but
> by the Apple II+ I had as a high-school kid, which used similar DIP sockets
> on every chip.
> 
> My guess is that the gold-plated machined-pin sockets will be fine for the
> duration.  The pins probably saw enough tinning during soldering, and if
> not, you can just reheat them later if needed.  So far, I've seen problems
> with gold-to-copper solder joints only in cases where the rule against
> making a physically-stressed solder connection has been violated.  The 8662A
> is prone to those problems because HP failed to use pigtails to connect
> their SMC center pins to the PC boards.  A bigger problem with the
> machined-pin DIP sockets is that they can let go of their chips if subjected
> to vibration in some orientations, another fun phenomenon that the
> manufacturers don't seem to talk about.
> 
> 5370s are nice boxes but they can be a challenge to work on.  There are a
> couple of editions of the service manual, both apocryphal at best.  One
> concern is that the interpolators are hard to calibrate properly without an
> 8082A pulse generator (read: I'm not sure how you'd even attempt it.)
> 
> -- john, KE5FX