[time-nuts] Returnee, interested in Lars GPSDO

[Jim Harman]

(I originally sent this just to Bob, repeating for the group)
The "something else" in Lars's design is provided by 5 MHz derived from the
oscillator, which feeds a timer in the processor.  The 1 pps triggers a
"timer capture" which records the number of 5 MHz cycles since the last pps
with a resolution of 200 nsec.  (The Arduino processor clock is 16 MHz and
the maximum timer frequency is Fclk/2, so you can't drive the timer with 10
MHz.)

This coarse detector is used to extend the lock-in range and achieve the
initial lock. It would be possible to make a GPSDO with just this detector,
but the phase sensitivity would be low. The 4046 followed by the pulse
width to voltage converter and the processor's 10 bit  A/D converter reads
the phase with a resolution of about 1 nsec.


On Thu, Dec 22, 2022 at 8:40 PM Bob Camp <kb8tq at n1k.org> wrote:

> Hi
>
> Since you only get one sample per second, no matter what you do, why
> bother with
> the 4046 at all? 1 us will roll over at 1 ppm off frequency ( which is
> very possible with
> an OCXO ). You still need “something else” to get things close.
>
> Bob
>
>
--Jim Harman