In the last post, I used a rule of thumb to estimate that my receiver system was stable enough so that it didn't contribute significantly to the slightly varying frequency signal from WWV due to propagation. It would be nice to verify this and maybe measure the actual stability of my receiver in the process.

Since propagation is apparently the source of the variability in the WWV signal, I need to measure a signal not affected - at least as much - by the ionosphere. One excellent way to do so would be to use a local frequency standard fed directly into the receiver. One of these days I may borrow an atomic clock from work and do just that, but let's try another approach today.

Besides WWV signals on the HF spectrum, NIST also broadcasts an RF signal at a very low frequency of 60 KHz from the station WWVB. This station is also located near Boulder, Colorado but because the frequency is so low it can propagate directly to Boston via ground wave instead of through the highly variable ionosphere. How strong a signal finally arrives here compared to my urban background noise is an issue, but let's see what I can do.

The Flex 5000 receiver isn't really ideal for this, as it is optimized for ham bands where it has some nice filters. Flex says that, at the least, you need a good low-pass filter to get rid of signals at higher frequencies which are translated into the LF band by the Tayloe mixer that the Flex uses. The strongest signals are from the broadcast band just above the LF band I'm interested in. And indeed when I tune to 60 KHz with my flex, its swamped with BC signals.

I happen to have a nice BC band reject filter from Jack Smith, K8ZOA, Clifton Laboratories. I plugged this in and those BC signals are nicely attenuated so they are not a noticeable problem. Unfortunately there is still some tuning wonkiness (signals that go the wrong direction when I tune) and there's no WWVB signal to be seen. I think the mixed down noise from frequencies above the BC band are still present since the band reject filter does nothing to them.

I also have an up converter from HEROS technology in the UK, which converts signals in the band from 10 KHz through 500 KHz up to 4.0 to 4.5 MHz where the Flex is much better behaved. And with it I can see the signal from WWVB at 60 KHz, so I'm in business. I recorded about a minute's worth of data and plugged it into my Mathematica analysis to get the following plot.

Oops - this doesn't looks so good. The problem is that an up converter like the HEROS works by mixing a local oscillator with the incoming signal, and if the local oscillator isn't perfectly stable, the converted signal will drift with the oscillator. If the frequency is drifting linearly for this short time period, then since the frequency is the time derivative of the phase, the phase will vary quadratically. Which is evidently what is happening here. At least the amplitude is nice and stable, unlike the WWV signal from the last post.

Its easy enough in Mathematica to do a least squares fit to a quadratic and remove the parabolic part of the phase change, but of course there's no way to distinguish a drift in the up-converter local oscillator from that in the Flex local oscillator. I think the evidence from the last blog post is that in this case almost all the drift will be due to the HEROS, but then there is no way to distinguish any phase drift in the propagation component, so its a lost cause. I did do the curve fit and got a residual phase plot that only varied over about 40 radians instead of the 3500 in the above plot, but I still think its a lost cause in the effort to measure my receiver stability.

At least the steady amplitude indicates that the signal is pretty constant and that bodes well if I am ever able to receive WWVB directly on the Flex. I've ordered a low pass filter from Jack, and I can always put up a better antenna so there is still hope.

Tangled Spaces