[time-nuts] Tuning a Trimble Thunderbolt
pete at heypete.com
Tue Apr 21 04:30:08 EDT 2015
On Tue, Apr 21, 2015 at 4:25 AM, Charles Steinmetz
<csteinmetz at yandex.com> wrote:
> Pete wrote:
>> On a related note, is it possible to extract any data regarding the
>> training from the unit?
> Not as far as the time-nuts community knows, no (other than looking at the
> DAC voltage and temperature reporting during holdover and attempting to
> reverse engineer the prediction algorithm by correlating those with the
> long-term DAC voltage -- good luck).
Finishing my PhD is enough work already. I don't think I'll try
reverse-engineering the prediction algorithm quite yet. Perhaps later
in my Copious Free Time(tm)?
>> Are the training parameters saved periodically to non-volatile memory,
>> or are they purely stored in RAM and so will be lost if powered down?
>> If the latter, does the RAM have any provisions for backup power
> I doubt it -- mine always act as if they are training from zero if they have
> been powered down. Because of the lack of precise retrace of quartz
> crystals, I don't think you'd want old (pre-power-down) data, anyway. Some
> crystals will even come up drifting in the opposite direction after being
> powered down, and they all take some time (days, at least) to settle down
> after any disturbance (including power interruptions, however brief).
Ok. It'd be nice if there was some way to keep the crystal going
through power interruptions, even if the oven itself cooled off. I
suspect Trimble (correctly) assumed that the vast majority of these
units were to be installed in cell sites with reliable power so that
wouldn't be an issue.
>> Alas, the location for the antenna is suboptimal: in the best location
>> available to me (an outdoor balcony) I have a clear view of the
>> southern sky from 150-300 degrees (az) and from horizon to zenith with
>> only a few low-elevation obstructions. However, this is only
>> accessible in warm months
>> * * *
>> The surveyed position is within about 10 meters of the actual location
>> according to Google Maps and local building information.
> That's a problem. Every meter is approximately 3.3nS, so 10m introduces a
> +/- 33nS error in the raw data (as much as 33nS closer to some satellites
> and 33nS farther from others). Add in the uncertainty due to noise, and you
> get easily hundreds on nS of error in the computed solution.
Indeed. I'm running a 48-hour survey with Lady Heather now to see if
that can improve things a bit more.
> Unfortunately, you are unlikely to do any better than this with the antenna
> location you described. Time to buy a house, with no tall trees nearby.
> (You may already have heard that time-nuttiness can be expensive.... ;-)
I won't be looking for a house for at least a few years, but when I do
the skyview is definitely one of the criteria, as is the friendliness
of the community to radio masts.
More information about the time-nuts