So far I've been looking at changes in phase (and hence frequency) on the millisecond to minute time scales, but of course the ionosphere varies over longer time scales also. Today's exercise is to look at frequency variations over the course of a few hours. The tool we'll use is the SpectrumLab program written by DL4YHF.

This program is a very powerful analyzer of audio signals, in the case the demodulated audio signal from my Flex5000 receiver. I will only be using its basic FFT functionality to do very fine frequency measurements. In particular I will take 524288 point spectra at the 48 KHz sampling rate which gives a frequency bin size of a bit over 91 mHz (that's a small "m" for milli- ) and a time bin size of a little under 11 seconds, but with a 50% overlap with the previous spectrum. The display window size then allows about 80 minutes of spectra to be displayed at once.

Here is a trace of the Canadian time station CHU at 3.33 MHz. The frequency scale is +/– 10 Hz from the center frequency of 1000 Hz (the amount I detuned the carrier frequency) and the dotted time lines are at 5 minute intervals.

Waterfall plot of CHU time station carrier at 3.33 MHz.

So this is a rather unexciting signal, but it is pretty stable - good to within the resolution of the FFT and no obvious propagation effects.

Same story for the CHU signal at 7.850 MHz, though there is a hint of something at the very beginning of the plot at the bottom.

Waterfall plot of CHU time station carrier at 7.850 MHz.

If we go up in frequency to the 20 MHz WWV signal, where we are more sensitive to Doppler shifts (which scale directly with the frequency for a given velocity) we begin to see some structure to the signal. The weak signal off to the left is not related - its probably one of the multitude of little wall wart power supplies in the neighborhood.

Waterfall plot of WWV time station at 20.0 MHz.

The structure branching off to the left of the main signal is probably some sort of multi-path propagation through the ionosphere, where the patch that the signal is refracting off of is moving at a velocity of about 15 m/sec (which gives a Doppler shift of about 1 Hz). Or more precisely it is moving at a 15 m/sec more than that of the main signal path along the propagation direction. Nevertheless this appears to be a reasonably stable signal, and the phase methods that we were using before would probably yield reasonably stable values.

Earlier in the day, however, the 10 MHz WWV signal was anything but stable. Here are four hours of that signal, with the same scales as above. Time flows from bottom to top, left to right.

WWV signal at 10 MHz over four hours. Time flows bottom to top, left to right.

During the first couple of hours there is a reasonably clear stable base signal which shows up as a vertical bright line. However there is a lot of smear to the signal which indicates that the frequency is varying about 2 Hz over the 10 second sampling time. In addition there is another path, which shows up as the separate trace on the left, which has a frequency shift of 6 to 8 Hz, also with a lot of smearing. By the final hour the signal is really crazy - measurements here would be pretty variable. On the other hand it shows a very dynamic ionosphere, which is kind of cool. It would be nice to be able to say more about the details of the ionosphere behavior, but with only a single receiving station that is pretty much impossible. It would be fun to have a dozen or so stations, geographically distributed, to sample the signal - then much more information could be gleaned.

Sometimes there are signals that I don't know how to interpret. Here is the CHU station at 7.85 MHz. I have no idea what the symmetric multiple trace signal near the middle represents.

There is one other phenomenon which needs to be kept in mind when looking at Doppler signals on this scale: reflections from things other than the ionosphere. In particular airplanes can reflect the signal to the receiver. I'm more familiar with these reflected signals when working at higher frequencies where they are more common and easier to interpret. When I used to monitor meteor scatter using over the horizon tv station carriers at around 60 to 70 MHz, Doppler shifted reflections from aircraft were common and you could even deduce aspects of their flight paths from their time course. But I haven't been able to come up with a plausible scenario in either case that would yield this display. So I leave this post with a puzzle to be solved.

NOTE Added - 28 Dec 2016:

I'm pretty sure the symmetric "sidebands" were due to an AGC artifact - some sort of AGC "pumping" at 1 Hz or so that would give a periodic modulation of the signal which in turn would show up as sidebands on the actual signal. Note to self: turn off AGC when doing these types of measurements.

Tangled Spaces