[time-nuts] Tuning a Trimble Thunderbolt
pete at heypete.com
Mon Apr 20 05:39:50 EDT 2015
I recently acquired a 2004-era Trimble Thunderbolt with firmware 3.00
from eBay. It looks essentially identical to the one sold in the
TAPR/Time Nuts 2009 group buy. A sticker says it's the "Rev E"
(it's not a Thunderbolt E, just revision E of the original
I don't use it for anything critical, just local timekeeping and hobby
stuff (I'm beginning to get into ham radio, so a good frequency
reference will be handy). I've been really impressed with it, but I'm
interested in tuning it for even better performance. The list archives
here have been useful, as have other outside resources, but I have
a few questions for the gurus here if that's not too much trouble.
I don't have a time-interval counter or local reference clock; all my
data is from Lady Heather. The Thunderbolt is resting in the shade on
a foam block on a table in my living room, which is not actively
temperature-controlled but is well-insulated and typically within a
few degrees of 20C. It's been running for about a week with
uninterrupted GPS signal, though I typically enable manual holdover
for a few hours a day for testing.
The default tuning parameters keep the phase and frequency error
within the published specifications of +/- 20 ns (1 sigma) and
~10^-12 over the course of a day, respectively, so long as the
receiver is locked to the GPS signal. However, when using the default
parameters it doesn't meet the holdover specifications of +/- 1 us
over 2 hours with a maximum of +/- 15C temperature change: it will
drift at least 20 us over 2 hours in holdover.
1. Is there some preferred, step-by-step method for manual tuning?
I'm familiar with the Ziegler–Nichols method for tuning PID
controllers and that method works reasonably well for adjusting PID
controllers used for temperature control in the lab at my workplace.
Is there some method that's comparable? As a general example, would
"Adjust time constant until the phase error starts to oscillate but
frequency error is stable and low. Reduce damping value until phase
error stabilizes." be sensible?
I ask because although Lady Heather's "autotune" function works well
at setting the gain and DAC values, the time constant (500 seconds)
and damping constant (1.00) is hard-coded into the source and want to
know how to adjust those parameters for my particular Thunderbolt.
2. Is it typical for an oscillator in holdover to drift in a
non-linear way? For example, with a bit of tuning my Thunderbolt
drifted to a PPS offset of 1 us after 126 minutes. However, at 160
minutes the offset was 2 us, at 190 minutes it was 3 us, and so on.
After about five and a half hours (330 minutes) the offset was 11 us.
Can this non-linearity be corrected through the judicious choice of
tuning parameters or some other means?
3. Although my attempts at tuning have improved the holdover
performance over the default parameters, I'm nowhere near the
performance reported by . That document says the units under test
were standard Thunderbolts (not Thunderbolt E's) and were on for three
days and had a "training period" of two hours prior to the test. My
Thunderbolt has been on for a week and had been locked to the GPS
signal for at least 12 hours prior to the test.
Lady Heather shows "N/A" for the Kalman filter (the PV, static, and
altitude filters are on). This appears to be normal, as screenshots
from others  show the same thing. Is this expected? Is there still
some internal Kalman filter?
3. Is it normal for there to be "spikes" in the phase and frequency
error when the number of satellites being tracked changes? I observe
changes of ~100ns and 100-200ppt whenever there's a change in the
number of satellites. Can this be smoothed out?
Many thanks. I apologize if I'm duplicating an earlier discussion, but
my search-fu didn't turn up answers to these questions in the list
archives. If this has been discussed before, pointers to the earlier
discussion would be most appreciated.
More information about the time-nuts