On Mar 9, 2012, at 4:28 PM, Peter Monta wrote:

I took a look at this. Pictures and video here: http://jks.com

I believe what they were doing is applying a simple 180 deg phase shift to the carrier (BPSK) during certain one second timecode intervals.
Whether there was anything fancier going on or not (like what DCF77 does) I couldn't tell immediately, although I did make a 192 ksps sound card recording for future analysis.

Apologies for the crude measurement and presentation. I had to throw this together in a hurry.

On Mar 11, 2012, at 12:53 PM, John Seamons wrote:

Okay, so I did a little analysis of the data. See the website for details. For the few minutes I recorded there was a fixed pattern of phase changes that repeated each minute. They did not seem related to the changing timecode. I tried interpreting the bits as various things, including ascii characters, but no luck. Maybe one of you will figure it out, or I suppose we could ask.

I suppose this will help coherent receivers somewhat, but I was
expecting something DCF77-like.  A pity.

Thanks for posting your data.  You certainly have good WWVB reception.

Attached is a plot of the real part of the demodulated signal after
carrier estimation.  They change the phase 100 ms after the second
marker, halfway between the second marker and a potential change in
amplitude back to full power.  So the phase transition always occurs
in the middle of a low-power interval.

Cheers,
Peter

I did ask, using the contact email on the NIST page (John Lowe).  I'll
forward any replies to the list.

Perhaps we could encourage a more ambitious DCF77-like scheme, if
that's really not in the current plans for WWVB.  Especially given the
recent demise of Loran-C, a WWVB service with greater timing precision
would certainly be welcome as a backup for GPS.

Cheers,
Peter

Has anybody looked at the impact of the periodic phase reversals of BPSK
on the loop of phase-tracking receivers, like the Fluke or the HP 117A?

NIST does claim backward compatability for time. But what about time
interval?

I know you can extract the carries from BPSK with a Costas Loop (which
essentially squares the signal and uses the second harmonic) but the
existing, installed hardware does not do this.

If I'm right, that's another broken egg in the frequency reference basket.

Best,

-John

================

Attached are three images of phase data that I captured during and after
the test.  These are the 1PPS output from a Spectracom 8170 time code
receiver compared to the PPS from an Rb standard (with linear frequency
offset removed).

The Spectracom PPS is derived from a 10 MHz crystal that is fairly
tightly phase locked to the WWVB carrier.  It's not quite the same as a
tracking receiver, but you can see from the "after" that it tracks the
phase ID quite closely.

First image is the whole 72 hours showing the last day or so of the test
period and the return to normal operation.

Second image is a zoom in of the new modulation as tracked by the receiver.

Third image is a zoom in of the old modulation over the hourly 45 degree
phase shift ID.

John

J. Forster said the following on 03/14/2012 05:21 PM: