[GPS_Standard] Proportional and/or integral adjustment

[Bob Bownes]

Alternately, if there isn't room in the controller, do it in a PC
based program that makes the analysis & adjustments externally.

My system unlocked about an hour ago and went into holdover mode. The
slew was so great on the drift (I suspect the AC kicked on at home,
I'm monitoring it from work) that I had to manually adjust the DAC by
-300 to get it to lock again. @ a sample size of 16 (where it was when
it unlocked) that would have been a long time until it locked back up
again.



On Tue, May 25, 2010 at 5:26 PM, Dave Platt <dplatt at radagast.org> wrote:
> Bob Bownes wrote:
>> Dave,
>>
>> That would certainly help avoid the dithering problem.
>>
>> I _think_, if I read you right, it would also help my drift after
>> unlock problem, in which, unless I crank the Sample Size (S) down, it
>> won't lock again because the DAC delta is less than the drift delta.
>
> Yes, I think it would.  It could help reduce the risk of unlocking in
> the first place (by tracking more accurately) as well as speeding up
> the re-lock if an oscillator drifts rapidly or jumps frequencies.
>
>> An alternate might be an 'Initial Lock mode' in which the value of
>> correction is proportional to the freq difference, much as you say.
>
> Yes, a two-mode-of-operation feature (with some logic for automatically
> switching between modes) could be useful, as well.  It would speed the
> process of establishing an initial lock at turn-on time, while
> allowing long-term stable operation with different parameters.
>
> Implementing dual-mode, without requiring lots of nasty code
> changes (e.g. duplicating lots of logic, or adding if/then/else
> on every parameter lookup) might be a bit tricky to do.
> Perhaps this could be done by "shadowing" the various sets of
> EEPROM parameters into two different banks of the microcontroller
> RAM, accessing the variables from RAM rather than EEPROM, and
> doing the mode-switch by just switching RAM banks?
>
>