[time-nuts] Tbolt issues
csteinmetz at yandex.com
Sat Sep 10 17:32:12 EDT 2016
> The one thing that autotune seems to do well is to come up with the actual sensitivity of the OCXO you have. It depends on a few things to do this so it might go wrong. I’ve never seen it come up with the wrong number. It then appears to drop in a gain and damping that make more sense than the original numbers. Since it is a pre-defined pair of numbers, it is indeed a “one size fits all” solution.
Warren, who I understand provided the algorithms for the autotune
routine, has advocated in a number of list postings damping factors much
lower than what I consider optimal (and in some cases below the factory
default of 1.2). Also, in the case of my units, the autotune function
adjusted the tuning rate parameter (OCXO sensitivity) for substantially
increased loop gain, which effectively decreased the damping even
further. So, one of the results [long ago, and with my very small
sample of Tbolts] was to adjust the loop toward and even into instability.
It also seemed to tinker with parameters I didn't expect it to change,
which is why I had to do a hard reset rather than just re-program the
settings that I had changed as a result of my prior experimentation.
Again, I have no idea why it did this, and it is very possible that the
autotune routine on current versions of LH works perfectly and gives an
optimum tuning very painlessly. But back when I tried it, with my two
Tbolts, it made them pretty much unusable.
No big deal -- just evaluate the operation of your Tbolt after using the
autotune routine, and if you find that it did not produce the results
you hoped for, be prepared to do a hard reset and tune it manually. If
it works well, then great!
More information about the time-nuts