[time-nuts] Lady Heather's Tbolt oscillator auto-tune function

Charles Steinmetz csteinmetz at yandex.com
Wed Sep 14 00:06:05 EDT 2016

Mark wrote:

> Okee dokeee...  here it is.   Not much difference.   The initial step is smaller,  but it still spikes.  After that things are pretty much the same.   After it cool down,  I'm doing another run with the initial voltage set to the peak of the spike.

Excellent!  Many thanks, Mark.  I'll be interested to see the next run, 
with INIT set to the current peak DAC voltage.  It's looking like some 
glitch may be an inherent feature of the Trimble software.

Back when I was still playing with mine, I noticed that there was often 
an ugly, long spike with an oscillatory tail coming out of holdover. 
Even if the held DAC voltage is very close to the new required DAC 
voltage, jam synch is set very tight, the allowable frequency error in 
recovery is also set very tight, and the PPS phase coming out of 
recovery is small (< 50nS), the control software sends the DAC way off 
into the wilderness for no apparent reason.  That does not seem to be 
driven by the normal discipline loop -- rather, it appears to me to be 
an error in the recovery routine.

> One slight difference was with the new initial voltage setting (closer to the current 10.00000000 MHz operating point),  it appeared to acquire satellites a few seconds faster (but that could just be luck of the draw).

Yes, that's an expected outcome -- the closer you start to 10.000000000, 
the faster it should be able to acquire satellites.  Of course, 
acquisition has lots of other variables, including the position of the 
constellation at start-up time, weather, etc., so we expect it to be a 
statistical trend, not a firm rule.

> Setting the initial voltage to something far off of optimum would be interesting... I seem to remember it taking several minutes to acquire.

Both of mine had way-off INIT settings when I received them (years ago, 
now), and yes, they took a few minutes longer to acquire the first two 
satellites than after I set INIT to the value that produced the 
smoothest transition.

Thanks again, Mark!


More information about the time-nuts mailing list