[time-nuts] First success with very simple GPSDO

[Chris Albertson]

I think the solution is easy.

1) count all the cycles using an overflow interrupt until you see the
result is stable to about 1ppm.  We all know that even a simple $8
controller can do this.

2) then switch to a mode where we look only at the last few bits of the
counter.   I think this will actually perform better than mode #1 above
because there is zero chance of the two interrupts happening at the same
time causing your PPS sample to be delayed because you had interrupts
disabled while counting an overflow.

The next step is to implement a "quite time".  This is common in some real
time systems.  If you KNOW a little in advance you are about to sample some
sensitive signal then you turn off anything that is electrically "noisy"
like motors or even updates to a serial device.  In a GPSDO you can predict
when the next interrupt will happen and insure there is not serial data or
whatever happening near the "tick" time.

A dual mode system means no one needs a counter or WWV or any other
external reference. and you also, after syncing get very deterministic
behavior.




On Fri, Apr 11, 2014 at 8:23 AM, Charles Steinmetz <csteinmetz at yandex.com>wrote:

> I wrote:
>
>  > (Note that a 256 cycles per second error is 51 PPM at 5 MHz.)
>>
>
>  > In the real world, you should  be able to trust that any oscillator
>> that is
>> > chosen for a GPSDO will  free-run within 1 PPM when it is warmed up, and
>> > that your PPS will  always be within 1 PPM, as well.
>>
>
> Hal wrote:
>
>  Why?  The original goal was low cost.  I could easily believe that an
>> inexpensive oscillator would be off by more than 1 PPM.
>>
>
> Decent OCXOs are falling off the ebay tree for $20 every day (less, if you
> look and wait -- I've bought several for <$5).  I have bought dozens of
> them, and have yet to receive one that isn't within 2e-7.  So I'm inclined
> to think any oscillator that is worth using for a GPSDO has a very good
> chance of being within 1ppm without any adjustment or selection.  Further,
> there are good ways to get an oscillator within 1e-6 without a counter --
> for example, beating with WWV.
>
> Indeed, if cost is the limiting factor, I submit that most people would be
> better served by spending their entire budget on the best OCXO they can buy
> and a box to put it in to reduce temperature fluctuations, then manually
> calibrating it periodically, than by buying a cheap oscillator and spending
> money on a GPS, uC, etc. in an attempt to make a silk purse out of it.  GPS
> can only help at long tau (or said differently, it can only help at short
> tau if the oscillator is too unstable to be worth using in the first place).
>
> Best regards,
>
> Charles
>
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/
> mailman/listinfo/time-nuts
> and follow the instructions there.
>



-- 

Chris Albertson
Redondo Beach, California