As I noted in a previous post, clock stability on my reproduction SCELBI was pretty bad. The system clock slowed down 10 or 15 percent as the system warmed up. This would have played havoc with my bit banged serial communications, but by the time I was able to toggle in the loader, it would be warmed up and fairly stable.

SCELBI cpu clocks

The timing circuit is made up of 4 interconnected 74121s one shots that control rising and falling edges of 2 interleaved clocks. The timing of each 74121 is controlled by a 5K trimmer resistor and a 300 pF cap. Figuring that the cap was part of the problem, I looked for some NP0 300 pf caps to replace the ones that i was using.

What I found was a modern 330 pF cap, Kemet part # C322C331J2G5TA. While I was at it, I decided to replace the 5K trimmers that I had been using since first bringing up the SCELBI. The original trimmers I bought had the wrong foot print and turned out to be a little jumpy while adjusting. I replaced them with Bournes # 3006P-1-502LF.

Though I don’t know which change made the difference, several trials have shown the new trimmers and NP0 caps seem to have stabilized the clock to the point, where I think I can forget about the “problem”.

The only thing I might do at some point, is try to find some old style NP0 caps to replace the obviously modern Kemet ones.