Discussion:
[time-nuts] WWV/H Doppler Shift
Scud West
2018-11-28 01:40:51 UTC
Permalink
I’ve been monitoring the Doppler shift of WWV/H and CHU for a while now from my location 100 km / 60 mi NW of Seattle.

The receiver is a QS1R, using an LTC2208 16 bit ADC, and the 125 MHz clock is provided by a Bodnar GPSDO. A Wellbrook ALA1530S+ loop antenna provides good coverage for the entire range.

I use a Python script to tune the QS1R to each frequency for 11 seconds while in CW mode. The 800 Hz beat frequency is measured using the open source program Fldigi.

Fldigi is mainly intended for HF digital modes like RTTY and PSK31, but the “Frequency Analysis” function is effectively an audio frequency counter with .001 Hz resolution, updated every 1.024 seconds. The first 5 or 6 samples are discarded after a frequency change. The signal must be within about ± 5 Hz.

During many minutes, WWV transmits a 500 Hz tone, and WWVH a 600 Hz tone (or the reverse). By treating these as separate carriers (e.g. tuning to 5.0006 MHz) I'm able to separate the Doppler shift and signal strength of the two stations. It's remarkable that Fldigi is able to make an accurate measurement of the 500/600 Hz "carriers", since there is a 40 ms gap surrounding the tick sound each second.

I've just recently got enough of a handle on Python, Matplotlib, Pandas, JupyterLab, and other excellent tools to even begin to analyze and display the data.


From my location:

WWV: Ft Collins, CO 1,675 km, 1,040 mi, 113°
WWVH: Kauai, HI 4,335 km, 2,694 mi, 241°




Each datapoint for the 500/600 Hz signal is plotted, to give a sense of the signal variation. Only the smoothed carrier signal is plotted, because otherwise it made the graph "too" busy (ha). The 500/600 Hz plots are moved up by 15 dB to correspond with the carrier level. But this means their displayed noise floor is raised as well. For instance, the WWVH signal from 18:00 until past 00:00 is mostly in the noise.




Each observation is for 5 or 6 seconds, and taken a few minutes apart. If the standard deviation of the observation exceeds 0.150 Hz it is rejected. The reject is plotted in light gray, but otherwise ignored. That's the only filter being applied to the data. WWV is usually stronger at my location, and the carrier data correlates with it pretty closely. Based on looking at a few days data, usually WWV gives a more accurate and higher confidence reading than the carrier alone, and considerably better than WWVH. Last week the largest daily 5 MHz WWV median was .010 Hz (2.0e-09); one other day was .003, but most were .001 (2.0e-10).


I'm just now getting presentable results, and expect to find errors: cosmetic, conceptual, and fundamental. Currently the Python script is as shaky as my understanding of basic statistics.

Oh well, back to the data


Rob

_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts
Tom Van Baak
2018-11-28 03:39:26 UTC
Permalink
FYI: The wonderful plots that Rob included in his posting an hour ago did not make it to the list. This is because they were inline or embedded within html rather than plain file attachments.

Anyway, I recovered his images and they are attached to this posting:

plot_5.0_dBm_2018-11-25.jpg
plot_5.0_Hz_2018-11-25.jpg

/tvb
Moderator, www.leapsecond.com/time-nuts.htm
djl
2018-11-28 03:55:47 UTC
Permalink
Tom: Thanks! Rob: your toolchain (to borrow a phrase) produces great
results. If I read the graphics right, you might be able to obtain some
outstanding results from the HAARP Research Campaign 29 Nov - 03 Dec
2018. U of Alaska Fairbanks Geophysical research Institute. A thought
from another QS1R user. (btw, I'm told the QS1R is at the heart of the
Icom IC-7300 SDR transceiver.)
Good stuff
Don
Post by Tom Van Baak
FYI: The wonderful plots that Rob included in his posting an hour ago
did not make it to the list. This is because they were inline or
embedded within html rather than plain file attachments.
plot_5.0_dBm_2018-11-25.jpg
plot_5.0_Hz_2018-11-25.jpg
/tvb
Moderator, www.leapsecond.com/time-nuts.htm
_______________________________________________
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
--
Dr. Don Latham AJ7LL
PO Box 404, Frenchtown, MT, 59834
VOX: 406-626-4304


_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Bill Dailey
2018-11-28 09:15:43 UTC
Permalink
I used to measure WWV carrier only (a much more pedestrian approach than yours) with a qs1r locked to a Valon synthesizer driven by a a very nice (for my budget) 1000A oscillator that Corby provided me.

I often saw (now I could be reading your data wrong) a minimal shift when the signal strength was very low. I researched this phenomenon for quite some time and my hypothesis was that this represents a point where you receive only non-Doppler shifted signal via pure reflection at the ionospher. From the papers is read this component is always received but it is buried within the usually stronger shifted components. If I remember right there is a name for that direct (non-shifted) component but it escapes me.

I then moved, got busy and haven’t gotten an antenna back up. Your results make me want to get less busy again.

Thanks,

Bill Dailey
Post by Scud West
I’ve been monitoring the Doppler shift of WWV/H and CHU for a while now from my location 100 km / 60 mi NW of Seattle.
The receiver is a QS1R, using an LTC2208 16 bit ADC, and the 125 MHz clock is provided by a Bodnar GPSDO. A Wellbrook ALA1530S+ loop antenna provides good coverage for the entire range.
I use a Python script to tune the QS1R to each frequency for 11 seconds while in CW mode. The 800 Hz beat frequency is measured using the open source program Fldigi.
Fldigi is mainly intended for HF digital modes like RTTY and PSK31, but the “Frequency Analysis” function is effectively an audio frequency counter with .001 Hz resolution, updated every 1.024 seconds. The first 5 or 6 samples are discarded after a frequency change. The signal must be within about ± 5 Hz.
During many minutes, WWV transmits a 500 Hz tone, and WWVH a 600 Hz tone (or the reverse). By treating these as separate carriers (e.g. tuning to 5.0006 MHz) I'm able to separate the Doppler shift and signal strength of the two stations. It's remarkable that Fldigi is able to make an accurate measurement of the 500/600 Hz "carriers", since there is a 40 ms gap surrounding the tick sound each second.
I've just recently got enough of a handle on Python, Matplotlib, Pandas, JupyterLab, and other excellent tools to even begin to analyze and display the data.
WWV: Ft Collins, CO 1,675 km, 1,040 mi, 113°
WWVH: Kauai, HI 4,335 km, 2,694 mi, 241°
Each datapoint for the 500/600 Hz signal is plotted, to give a sense of the signal variation. Only the smoothed carrier signal is plotted, because otherwise it made the graph "too" busy (ha). The 500/600 Hz plots are moved up by 15 dB to correspond with the carrier level. But this means their displayed noise floor is raised as well. For instance, the WWVH signal from 18:00 until past 00:00 is mostly in the noise.
Each observation is for 5 or 6 seconds, and taken a few minutes apart. If the standard deviation of the observation exceeds 0.150 Hz it is rejected. The reject is plotted in light gray, but otherwise ignored. That's the only filter being applied to the data. WWV is usually stronger at my location, and the carrier data correlates with it pretty closely. Based on looking at a few days data, usually WWV gives a more accurate and higher confidence reading than the carrier alone, and considerably better than WWVH. Last week the largest daily 5 MHz WWV median was .010 Hz (2.0e-09); one other day was .003, but most were .001 (2.0e-10).
I'm just now getting presentable results, and expect to find errors: cosmetic, conceptual, and fundamental. Currently the Python script is as shaky as my understanding of basic statistics.
Oh well, back to the data
Rob
_______________________________________________
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and f
Scud West
2018-11-28 21:12:44 UTC
Permalink
I think I've seen something similar during and after geomagnetic storms - weaker than usual signals, but very low and steady Doppler shifts. It will be interesting to see how things change when solar activity increases.

The QS1R is a great receiver. Even with the original TCXO I could see WWV on 10 and 15 MHz track within 2e-09 while CHU in eastern Canada would have a shift of nearly 1 Hz. As long as you're not dealing with voice signals a pretty modest antenna will probably do. I use a bandwidth of 20 Hz for carrier level measurements, but the Fldigi would be fine with a wide bandwidth.

Who knew unmodulated carriers could be fun?

Rob
Post by Bill Dailey
I used to measure WWV carrier only (a much more pedestrian approach than yours) with a qs1r locked to a Valon synthesizer driven by a a very nice (for my budget) 1000A oscillator that Corby provided me.
I often saw (now I could be reading your data wrong) a minimal shift when the signal strength was very low. I researched this phenomenon for quite some time and my hypothesis was that this represents a point where you receive only non-Doppler shifted signal via pure reflection at the ionospher. From the papers is read this component is always received but it is buried within the usually stronger shifted components. If I remember right there is a name for that direct (non-shifted) component but it escapes me.
I then moved, got busy and haven’t gotten an antenna back up. Your results make me want to get less busy again.
Thanks,
Bill Dailey
_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Loading...