At 08:12 PM 11/22/2005, Robert McGwier wrote:
We do not want to damage the phase characteristics the 200 Mhz
oscillator provides but we want more accuracy and we are willing to pay
a few bucks to get it.


1) Pick off the 200 Mhz oscillator, buffer it, whatever and bring it
outside the box.
2) Hook into RF oscillator input port on error reporting box.
3) In the error reporting box, we divide the 200 Mhz oscillator by a
decent number.
4) This divided signal will be used as the CPU clock for the PIC.  Pick
the largest number that will allow the PIC to function.
5) The counter pin on the PIC is hooked to your GPS tamed reference,
let's say it is 5 or 10 Mhz.
6) The counter does a large counting job on the 10 Mhz reference and you
do some small coding magic inside that keeps things added up and behaving.
7)  If the 200 Mhz clock were dead on accurate, you should be able to
compute exactly how many ticks of the 10 Mhz reference you should see in
an interval.  You need
     several intervals to cut down on the quantization error..
8) The 200 Mhz clock error will cause the tamed, accurate 10 Mhz
reference to seem to be different from 10 Mhz in these counts.  This
error DIRECTLY translates to the error in the 200 Mhz clock and
    can be reported to the PowerSDR through the USB , Ethernet, or a
serial port.

This will not take care of the soundcard error but that is also easily
calculated by listening to WWV.

Or, by dividing that stable 10 MHz down to something your sound card can see directly. That way you don't have that ionospheric path... And, you've already got the PIC there to do the dividing...



I and others have shown that (for
example) the Delta 44 clock is stable to a very tiny number of parts per
million.  Its frequency can be off, but it is very stable.

These two ideas put together will make your frequency read out very
accurate.  Both Jim Lux and I have commented on this before.  Tom Clark
and Rick Hambly have implemented this procedure
for frequency locking in applications they have worked on.  It works.  I
think this can be done for under $100 in the U.S. if you are homebrewing.


Reply via email to