[time-nuts] WWVB d-psk-r down conversion method

Charles Steinmetz csteinmetz at yandex.com
Mon Sep 22 23:02:19 EDT 2014

Paul wrote:

>The goal is to remove the psk so the old phase tracking receivers can work.

Yes, I understand that.  But you want them to work like they 
originally did, with the disciplined oscillator in the phase tracking 
receiver phase-locked to the WWVB carrier (or else you may as well 
just ignore WWVB entirely and generate your own 60kHz carrier).  Once 
you add an LO/BFO, the signal you end up with is NOT locked to the 
WWVB carrier -- it is locked to some frequency that is determined by 
both WWVB and the LO/BFO (whch means, it is only as accurate and 
stable as the LO/BFO).  So the whole benefit of receiving WWVB in the 
first place is lost.  [In the special case of a TRF receiver, with no 
LO/BFO, the signal will remain locked to WWVB.]

>Whats good about this as I just typed to Bob the signal is slow and easy to
>work on.
> From what I have seen the phase tracking receivers have a fairly long time
>constant. So the fact that the phase detect and flip occurs 1/10 of a
>second later should not have any effect on these radios.

It's got nothing to do with how fast or slow the signal you end up 
with is, or how easy it is to work on.  If the frequency and phase of 
that signal are not uniquely dependent on the WWVB carrier frequency 
and phase, then the oscillator you discipline will not be disciplined 
to the precision of WWVB -- it will be disciplined to no better than 
your own LO/BFO.  [Also note that the phase flips at one second 
intervals no matter what frequency you translate it to -- that is not 
a unique feature of the 100Hz recovered carrier.]

As Alex pointed out, you could in theory use a LO/BFO that is, 
itself, derived from the disciplined oscillator, and in which the 
loops will not lock unless the IF and LO have the correct 
values.  But, as Alex also points out, such a scheme will have about 
the same complexity as a Costas loop.  The Tracor itself uses a crude 
variant of this strategy, in which the LO is guided "huff-n-puff" 
style in steps of 1/100 of a cycle, some steps above and some below 
the correct frequency.  But when you are starting with a signal that 
is already orders of magnitude less stable than a GPS signal, it is 
just rude to throw away even more stability with that sort of 
approximation.  Furthermore, all of this would need to happen outside 
of the old-school phase tracking receiver, so you'd end up building 
your own external phase tracking receiver just to run the old phase 
tracking receiver.

Best regards,


More information about the time-nuts mailing list