[time-nuts] Re: Trimble Thunderbolt temp coefficient
Bob Camp
kb8tq at n1k.org
Mon Feb 19 16:55:08 UTC 2024
Hi
Unless you have a *very* stable temperature environment, 1000 seconds is long for the typical TBolt.
Any time you tweak this stuff you have a compromise involved:
Longer TC gives better ADEV (up to some point â¦)
Shorter TC gives better time accuracy
Thatâs just how itâs going to work. Long control loop response is not great for staying exactly on target. Fast control loop response lets a lot of noise through.
Bob
> On Feb 18, 2024, at 6:58â¯PM, Ben Gelb via time-nuts <time-nuts at lists.febo.com> wrote:
>
> Hi all -
>
> First time poster, hopefully this is on point for this list. I have
> recently come into an early 2000s era Trimble Thunderbolt and have
> been enjoying fooling around with it and trying to learn more about
> how to understand its behavior and assess its performance. But still
> quite green. Probably not qualified as a "nut" yet by any stretch, but
> maybe on the way...
>
> Though I don't currently possess the instruments to measure it myself,
> the internet (and this list) has brought me to understand the default
> TC and Damping parameters are rather conservative, and the best freq
> stability behavior is observed w/ a longer time constant (1000 sec
> appears to be a common choice). Intuitively I understand this to be
> effectively that the control loop is being too heavy handed in
> steering the OCXO and that gps jitter starts to dominate the short
> term stability (where the raw OCXO is probably better than GPS).
>
> So I have been experimenting w/ TC=1000sec, Damping=0.707, and
> watching the behavior.
>
> I notice the vco tuning voltage moves much more slowly and less
> noisily (as expected), but also notice large excursions in average PPS
> error (up to several 10s of ns) sometimes lasting hours at a time.
> These seem to be highly correlated with the diurnal temperature
> rise/fall in my garage, as reported by the thunderbolt.
>
> For example, today between 8am and 2pm, the temp (as reported by
> tbolt) rose by 2.125C before ~stabilizing. The average PPS error began
> around 0 at 8am, rose steadily to a peak of around 40ns at 12:30pm,
> then finally came under the control of the control loop and was
> brought back to near 0 again by 2pm. The DAC offset from start to
> finish was about -800uV, which is about 3e-3Hz (using the LH-measured
> -3.733Hz/V VCO gain) or 3e-10 parts/C.
>
> I think my main question is just whether this is ~normal temp
> stability for the tbolt OCXO? I don't have a good intuition for what
> "normal" is for these devices, so I guess that's my first question.
> And I guess the second is whether the rough analysis above passes
> muster (or is a completely wrong way to think about it?).
>
> W/o really fully being able to intuit how this temp variation would
> show up on an ADEV plot (I'm really just getting my feet wet here)
> this temp effect seems like it is going to show up as a high order
> bit?
>
> I did find this old thread on this list:
> https://www.febo.com/pipermail/time-nuts/2012-December/072823.html
>
> Which seemed sort of related but did not see any firm conclusion.
>
> Thanks in advance for any feedback/pointers!
>
> Ben
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe send an email to time-nuts-leave at lists.febo.com
More information about the Time-nuts_lists.febo.com
mailing list