Saturday, October 10, 2020

Chip Replaced, GSC 6000 Counter Fixed

 

This thing has been half-broken for a long time.  I needed to get the input for 40 MHz - 650 MHz working    I got the a replacement SP8630B Plessey divide-by-ten counter chip on e-bay, and yesterday I extracted the old chip and put in the replacement.  I took great care NOT to solder this one in upside down (as I had done with another chip replacement in this counter). I used solder flux and solder wick to gradually get the pins free of the board. (You can see the old chip in the picture above.)

As to what happened to the original SP8630B chip,  John over on the Vintage Test Gear Facebook page wrote: 

The Plessey SP8630A/B is an ECL divide by 10 prescaler, with a upper working frequency of 600MHz. That generation went out of production in the late 1980s. Plessey was bought by a Canadian company now called Micrel. You may be able to find one from one of the specialist obsolete component companies, but it may be dead on arrival. Those ECL ICs had a fairly high mortality rate if they are very old.

It is the old story of "metal migration". In early semiconductors very small impurities in the silicon structure cause minute bits of the metallisation to leach out into the essentially non-conducting silicon insulation. Many old devices, although they have never been used, were found to be very leaky and this degrades the gain of the active devices. The worst types are the very old Germanium transistors.

As the semiconductor scientist learnt more about the super cleanliness required and the better purification of the metals the problem tended to improve. The Marconi company I worked for back in the 1980s had a real problem with comms satellites failing after a few years of service. Of course you can't go up there and swap out the faulty devices. Accelerated ageing of a backup satellite showed that some devices just stopped working after being subjected to high and low temperature cycling, which is a common problem with satellites in orbit!

I am liking this little machine more and more.  It is very simple -- no microcontroller, just a collection of gates.   I discovered that the main main crystal oscillator is actually built inside a little oven to keep the temperature stable -- oscillator and the oven stay on as long as the counter is plugged in, even when the device is switched off.  I calibrated the counter with WWV and with my HP8640B and with my little Feeltech sig gen counter.   I wish I knew how to calibrate the counter in the Rigol DS1102E oscilloscope.  


5 comments:

  1. Morning Bill, the migration effect was actually very cool, while working for a major defense contractor in the early 80's I wrote a report in the subject. The branches of metal migration was called "dendrites" and under observation they looked like tiny trees! It's was the reason manufacturing moved away from silver to platinum palladium. You could actually see the migration if you temperature cycled it under bias for five days, but you could measure it after only two. Was exciting times. Cheers, Keith N6ORS

    ReplyDelete
  2. This comment has been removed by the author.

    ReplyDelete
  3. Very interesting Keith. So how long before I will have to do ANOTHER chip-ectomy? 73 Bill

    ReplyDelete
  4. As I've just read it, it looks like that chip should be in an oven for longevity, not necessarily but possibly the same one? Get it up to temp and keep it there? 40C would take care of most habitable zones on the planet. :)

    ReplyDelete
  5. The company that I used to work for built hybrid circuits. After a few years they started failing. Dendrites. I forget how we corrected the problem, besides replacing the hybrid.

    ReplyDelete