[Discuss-gnuradio] Decoding TDF's Phase Modulated Time Signal

[Iain Young, G7III]

Hi Folks,

I am continuing with developing my SDR radio clocks. TDF uses Phase 
Modulation (Details Here[1] and Here [2]), and I'm coming up against
my limits of understanding, so am after some pointers or at least
confirmation that I'm doing it wrong or am on the right path from
those better versed in the art.

I did some googling for how to do phase change detection in gnuradio.
I came across a post from Eric some years ago[3], and also a URL[4]
where someone had developed a PSK-31 receiver.

Not having much experience with decoding PM, I decided to try both
approaches. [5] and [6] are the flowgraphs in .png format. [7] and
[8] are the screenshots of them running.

The two are laid out the same. At the bottom is the output of the
multiply block. The middle graph is from post the LPF/Complex-to-Arg/
Complex to Real as appropriate. The top graph is the output from the
Binary slicer.

[Note while very tight low pass filters of 10Hz have been shown, wider
ones have been tried as well - 100Hz, 1kHz, and 10kHz, with similar
results]

According to [1] there will be 100 milliseconds of no phase modulation
before the start of second and data bit. While I do see some periods
(from looking at the binary slicer output) of around 100ms (usually
a bit longer than 100ms), I also see them more than once a second
(Well with devel 3 anyway...)

Additionally, there is supposed to be a full second before the start
of the minute, with no phase modulation (in this case that should be
around 48000 samples). I never see it, hence me thinking something may
not be quite right in my approach.

So, am I going about this the "right way" for what I am trying to do ?
Or is there a better method ? It may just be how weak TDF is with me,
but while the station itself is pretty weak audio wise:

a) 15dB above the noise floor
b) I can see a "pip pii pip" on a waterfall, even when I reduce the
LPF down to 10 Hz (See [9])

[10] and [11] are the GRC files themselves if anyone in Europe wants to
try them. Source is my Funcube dongle, tuned to 180kHz, and then the
FIR filter drops down 18kHz to make the 162kHz of TDF.


Thanks in Advance for any suggestions

Iain

[1]http://en.wikipedia.org/wiki/TDF_time_signal
[2]http://www.eecis.udel.edu/~mills/ntp/tdf.html
[3]http://lists.gnu.org/archive/html/discuss-gnuradio/2008-04/msg00027.html
[4]http://sdradventure.wordpress.com/2011/10/15/gnuradio-psk31-decoder-part-1/
[5]http://hal.g7iii.net/GRC/TDF/TDF_Devel_1.grc.png
[6]http://hal.g7iii.net/GRC/TDF/TDF_Devel_3.grc.png
[7]http://hal.g7iii.net/GRC/TDF/TDF_Devel_1.run.png
[8]http://hal.g7iii.net/GRC/TDF/TDF_Devel_3.run.png
[9]http://hal.g7iii.net/GRC/TDF/TDF_FCD_Test_Harness.png
[10]http://hal.g7iii.net/GRC/TDF/TDF_Devel_1.grc
[11]http://hal.g7iii.net/GRC/TDF/TDF_Devel_3.grc