Dear Wolf, Markus, LF Group,
I have been doing an experiment over the last couple of days to see if it is
practical to do this kind of reception with just the Spectrum Lab "frequency
compensation" facility, using an off-air VLF signal as a reference.
As I noted during DK7FC's last VLF test, the internal sound card in my shack
laptop PC suffers from a short-term cyclic frequency drift, over a range of
a few ppm with a period of about 100s, which appears to be due to the
cooling fan going on and off. This results in a spectral line in a
millihertz-resolution spectrogram being spread out over several millihertz.
I tried altering the "power management" parameters of the laptop, but
unfortunately none of these seem to result in the fan staying permanently
on, which would probably cure the problem. The SpecLab frequency drift
compensation facility had no problem locking itself to GBZ on 19.58kHz, and
gives practically perfect long-term frequency accuracy, but actually makes
the short-term drift problem "spreading" worse. I guess this is due to the
cyclic nature of the drift - the drift compensator shows corrections that
vary in a more or less random way between measurements over a range of about
+/-2ppm, although the long term drift is quite small compared to that.
However, I had a Maplin "budget" external USB soundcard, and tried that
instead. This gives a great reduction in short-term drift, with no visible
sidebands on the millihertz-resolution spectrogram above about -40dB on the
central peak. I then altered the FFT parameters to give 29.8 micro-hertz
resolution (centre frequency 8.97kHz, fs 96kHz, decimate by 6144, FFT length
524288 samples), and left it for a day or so to monitor an 8.97kHz signal
from a Halcyon PFS-1 frequency reference, which is locked to the BBC R4
signal on 198kHz. The resulting spectrogram is attached - the time markers
are 5 hours, the amplitude divisions 10dB. after apparently settling for
several hours, the signal is concentrated into a single peak at least some
of the time, although it looks like there are some "abrupt" phase changes
(at least compared to the 9 hour duration of the FFT samples ;-)). I am not
sure which part of the system these occur in - it could be due to
propagation or other changes, either in the 198kHz signal , or in GBZ.
As it is, the receiving system has frequency accuracy and resolution in the
several parts-per-billion range, which isn't bad considering the cheap and
unstabilised nature of the components. I think this would at least stand a
chance of seeing Markus' low ERP signal - will have to try next time!
Cheers, Jim Moritz
73 de M0BMU
uhz_spectrum.jpg
Description: JPEG image
|
