[time-nuts] 10MHz to 32MHz?

Dr Bruce Griffiths bruce.griffiths at xtra.co.nz
Sat Apr 21 18:59:15 EDT 2007


John Ackermann N8UR wrote:
> There is a firmware image for the Reflock II that will lock a 10 MHz (or
> other relatively-low-frequency) oscillator to PPS.  The last time I
> looked at it, it worked OK but not up to a "real" GPSDO -- which makes
> sense since it's a pretty basic PLL.  Luis Cupido has updated the PPS
> code since then, though, and for all I know it might be much better now;
> I haven't had a chance to test the new code.
>
> John
> ----
>
>   
John

Unless the hardware has been radically redesigned, changing the software 
doesn't get around the fundamental limitation that, since the VCXO being 
disciplined is used to measure the phase error, the phase error 
resolution is limited by the OCXO period. For a 10MHz VCXO the phase 
resolution will at best be 100ns (it may be even worse depending on 
actual implementation details).
There is also the likelihood (in the absence of insufficient noise) that 
averaging the phase values so obtained will not improve the precision 
over the one shot values since the VCXO is being disciplined to be a 
multiple of the PPS frequency.

Thus while the REFLOCK II is better than nothing its performance will 
fall far below a well designed real GPSDO in this application.
It will work much better for locking to higher frequencies where an XOR 
or phase detector with similar resolution is employed.

The obvious solution of multiplying the 10MHz VCXO frequency up to say 
100MHz or so to improve the phase error resolution to a multiple of 
10ns, is unlikely to be particularly satisfactory as the designer hasn't 
used a synchroniser in the design.

Bruce



More information about the time-nuts mailing list