LF: Slow PSK on Feb. 19/10

[MarkusVester@aol.com]

Dear John and LF,

WD2XES was received again here last night. On Argo 120 at 5.4 mHz resolution, the signal looked like this (22:15 to 8:15):



A few hints of the PSK blips can be guessed, but nowhere as clear as in the UK recordings from Jim and Dave. No other signal was seen within a couple of Hz.

Processing the narrowband recording produced the following panel. It shows XES on the E-field antenna at the top, and a GPS 1pps harmonic (coupled via 33 pF to the H-field antenna) on 137577.0 at the bottom. On the left side, there is a spectrogram having 64.5 microHertz FFT resolution, 10 minutes per pixel from 19:49 to 9:07 UT. On the right, the signal was despread by multiplying with a properly timed 2x15 minute PSK sequence, as done before:



The lower trace shows that the OCXO, though still free running, provided much better stability than the simple homebrew heated crystal. Total frequency variation improved from 3.5 mHz last time to better than one pixel (0.065 mHz), including the effect of the soundcard on the 952 Hz audio.

The stability of the traces encouraged me to try to make an absolute phase measurement and look for propagational variations. To this end, the GPS carrier on the right stereo channel was shifted by up by 0.309506014 Hz in software, and compared against the left channel containing the unspread signal from XES, using SpecLab's colour-DF setting to display phase difference as colour hue. This screenshot has 0.26 mHz FFT resolution and was flipped to have HF up:



With this technique, any LO drift is eliminated, leaving only soundcard samplerate deviations, but these affect only the small frequency difference to the nearest Hz. For 0.31 Hz offset, the estimated calibration error of 0.05 Hz would accumulate a phase error of no more than +-25 degrees in 13.3 hours. The samplerate could also be accounted for if the recording had been wide enough to contain more than a single GPS line. This was not the case last night as I used a decimated rate of 11027.15Hz/10368.

Other sources of phase errors could lie in the transmit and receive hardware (especially the narrowband antennas), or in the common GPS timing, but presumebly these are small compared to the skywave propagation effects.

In case anyone is interested in the decimated audio data, I uploadeded it to http://members.aol.com/df6nm2/SlowPSK/ . The dat files generated by DL4YHF's SndInput tool are 4 bytes per sample (8 bits IQ stereo, LI LQ RI RQ), without a header.

For tonight, I have switched back to 4096-fold decimation since 22:34. At the moment things look rather noisy - but we'll see.

73 and best wishes

Markus, DF6NM