[time-nuts] Digital Pots, wiper noise and fine tweaking oscillators

Charles Steinmetz csteinmetz at yandex.com
Mon Jan 27 18:32:04 UTC 2020


Bob wrote:

> Digital pots have *lots* of issues. A high quality wire wound pot likely will be significantly
> more stable and lower noise than your typical digital unit. In addition the 10 or 20 turn wire
> wound will have far more “steps” than a digital pot.

Digipots come in two flavors -- resistive ladders, and multiplying DACs 
("MDACs").  Each has lots of issues, some in common and some different.

Even with the MDAC variety (which can have as many as 16 bits worth of 
steps), I can't imagine ending up with sufficient resolution to give 
satisfactory step sizes for time nuts purposes, unless you cascade at 
least two of them in a "coarse and fine" arrangement.  Look back through 
the archives at the many discussions of suitable DACs for homebrew 
GPSDOs, for discussions of how many bits of resolution you need and what 
the tradeoffs are [e.g., lack of range].  Resistive ladder digipots are 
just hopeless, at around 10 bits of resolution maximum.  By the time you 
had sufficiently small increments to be useful to time nuts, you would 
no longer have enough range to compensate for oscillator drift over a 
usefully long period.

Also, look at the temperature coefficients on the datasheets.  You see 
values in the high hundreds to thousands of ppm per degree C.  Not ppb, 
mind you, ppm.  This, by itself, is very likely a fatal flaw when 
trimming measured in ppb is at issue.

> If you apply a 1 ns rise time step to the EFC of an oscillator it will not change frequency in
> a nanosecond. The tune port has a bandwidth. On an OCXO that bandwidth might be in
> the 10’s of Hz range.
>
> If you have bypass caps all over the place (and some large resistance here and there) on your
> EFC then indeed the caps can have various issues (leakage changes, dielectric absorption …)
> that can take a while to settle out. Is that minutes or days? It very much depends on just what
> you have wired up.

And fundamentally, the quartz takes time to "relax into" its new 
frequency.  This can be hundreds of ppb or more, and can take anywhere 
from a few days to a few months to settle within ppb.  Just one more 
reminder that there is no hurrying precision oscillators.

Best regards,

Charles






More information about the Time-nuts_lists.febo.com mailing list