Friday, August 15, 2014

Radio Warmup Issues

The WSJT-X software has various uses.  One neat one is to measure frequencies and drifts accurately.


This is the warmup curve for the TenTec Orion 575AT.  Ambient temperature is 75F, and we're starting from that temperature and watching the drift in the apparent frequency of the 15 MHz WWV signal.  The receiver is tuned to 14.999700 USB, so the "correct" response would be a WWV carrier appearing at 300 Hz on the spectrogram.  The WWV carrier (strong red trace) had a starting point (ambient) at about 280 Hz, but this plot begins a little after warmup has started. (The other, blue/yellow traces are modulation sidebands that can be ignored.)

You can see that the initial warmup (fast drift) is over after about 6-8 minutes.  A slow drift continues for another 10-15 minutes, with the apparent frequency stabilizing around 302 Hz.  But after 30-50 minutes, an even slower downward drift comes in that brings the reading down to ~298 Hz. (This is getting into the weeds of ambient temperature stability, etc.  For real statistics, you'd have to repeat all this under various conditions.)

Is this result good or bad?  The initial 20 Hz offset is about 1.3 ppm, while 2 Hz is 0.13 ppm, relative to the 15 MHz signal. The TenTec specification for this radio is +/- 3 ppm over the operating temperature range. The Orion uses a temperature compensated crystal oscillator - TCXO. The oscillator has a screwdriver adjustment for frequency, which I carefully tweaked in 2005.

Clearly, we are in spec at this particular operating temperature, but in fact the specification does not really tell us what to expect in terms of accuracy or stability at any particular ambient temperature. 

I may be in the minority, but I like to be able to use my gear for precise frequency measurement.  (And not as an expensive thermometer!) Some newer rigs are better.  For example, the Flex 6300 and 6500 offer a "0.5ppm TCXO", while the 6700 has a "0.02ppm OCXO" (oven stabilized crystal oscillator.) Flex offers an expensive add-in GPSTCXO option (GPS stabilized TCXO) that claims 5 x 10**-12 stability over 24 hours.

These numbers are helpful, but if you really want to compare time and frequency standards, you need to consider the Allan Variance, which is a measurement of stability on different timescales.  (See informative PowerPoint by H. Fruehauf, or comprehensive Wikipedia article.) What we usually need for precise amateur communications (say using advanced WSJT-style modulation at microwave frequencies) is stability over tens of minutes.  The vendors' specs really don't tell you that.