LF: Phase meter for propagation experiment

[DL4YHF@aol.com]

Dear LF group,

I am planning an experiment based on an idea from Alan G3NYK. The idea is to monitor the phase variation of strong signal on LF over day and nighttime. It may be interesting to find out how much phase variation is there; how much the propagation path length varies etc. 

At least two receiving sites are required: One close to the transmitter, and another far away from it. On both ends very accurate phase meters must be used. Of course, the phase of the transmitted signal must connected to an atomic clock like MSF or DCF77.

I think I found a possibility now how the required accuracy can be achieved, here is the basic principle. Most is done by software, only a decent hardware is required:

- The RX stations have clock sources locked to GPS, MSF, DCF77 or TV sync (15625 Hz) of certain broadcasters.

- The reference clock (from the locked source) is divided down to a frequency which can be handled with a standard soundcard. For reasons explained below, an audio tone of 1...3kHz is required. Assume 60kHz/24=2.5 kHz,  or 77.5kHz/31=2.5kHz, or 15.625kHz/6=2.604166666kHz. This audio frequency must be entered in the software's "sample rate calibrator".

- The divided reference frequency (or the 15625 Hz signal) is used in the software to PERMANENTLY monitor the soundcard's sample rate. This is important because the sample rate may drift by a few millihertz which is unacceptable here.
The software can already detect the sample rate from a very weak reference signal, so it is enough to add a small fraction of the reference frequency to the receiver's audio output because it is in another audio frequency band (longwave RX: 100...2000 Hz, reference: 2.5kHz or 15625 Hz). So there is no need for a stereo soundcard !

- The 2.5kHz reference is formed into a square wave like a 'frequency marker generator'. Odd harmonics are the result. One harmonic must be in the longwave receiver's passband, for example 55*2.5kHz = 137.5 kHz, or 53*2.604166666kHz = 138.020833333kHz. A small fraction of this harmonic is added to the antenna signal which goes into the receiver. We need this to compensate the VFO drift of a "normal" shortwave- or longwave receiver via software as explained below.

- Assume your SW receiver runs in USB, the VFO tuned to 136 kHz. For the VFO drift compensation (which is completely done in software), the received audio should contain a weak 'audio peak' at 137.5-136= 1.50kHz, or 138.02083333-136 = 2.02083333kHz. This audio frequency must be entered in the software's "frequency offset calibrator". 

With this system of two "calibrators" (one for the PC's audio sample rate, the other for the longwave receiver's slightly drifting VFO) it is possible to make very accurate long-term phase measurements. 
I have such a system running now, but not perfect yet, because my DCF77-locked source sometimes unlocks for a few seconds which spoils everything. I tried to convince my pocked GPS receiver to produce a 1-pps-signal which could drive G4JNT's GPS locked source but no success. At the present time I use the german ZDF TV broadcaster which has a precise 15625 kHz signal.

If someone likes to participate in this experiment, he may try to get the last version of SpecLab running. The two 'calibrator' routines are implemented but not explained in the manual yet, if there is interest in this experiment I will continue development and tell you how to use it. Or offer the calibration routines (written in C) to anyone who can program nice and clean user interfaces...  ;-)

Regards,
  Wolf DL4YHF.