3

In a typical urban (USA etc) setting, we seem to be awash in 'powerline hum'. Has it been done, or could it be practical, to recover this to get a wireless reference for a battery-powered system, like a clock, for example?

How rural would a site have to be where this would be completely impractical? How far does 'the hum' extend from high-tension lines, etc?

Jim Mack
  • 213
  • 4
  • 9
  • 2
    Analog hum is used to geolocate video/audio recording in forensics, so analysis is certainly possible. I think timekeeping would be harder because errors accumulate if you miss cycles, so SNR has to be relatively high and consistently so or the clock will drift rapidly. – user1850479 Apr 06 '22 at 14:48
  • 2
    @user1850479 Typically you'd use a relatively stable local frequency reference, and only use mains as a "better" reference to derive instantaneous frequency and phase errors and cross-calibrate the local reference to improve its stability. With mains frequency absent and the temperature stable, it'll take a while before even a single cycle slips. Even with $3 TCXOs you can do better than mains for a day or a few. Mains as a good frequency reference for clocks is "common knowledge" that became outdated in the last 30 years. – Kuba hasn't forgotten Monica Apr 06 '22 at 15:11
  • @Kubahasn'tforgottenMonica The problem with that approach is that mains frequency is not completely stable (which is why you can use it to geolocate recordings). Time keeping works because the utility will intentionally insert additional cycles to keep the long term average exactly right. If you miss blocks of data, even if you know the exact duration you missed, you will have some uncertainty about how many cycles were in that missed period because you don't know if they were stuffing extra cycles while you weren't recording. – user1850479 Apr 06 '22 at 15:16
  • As well as insufficient signal there may be close-but-incorrect signals, such as hum from an backup UPS or generator, if that matters. – Spehro Pefhany Apr 06 '22 at 15:27
  • @user1850479: The power companies aren't "inserting extra cycles." The vary the length of the cycles (change the frequency) by a tiny bit over the course of the day so that the average is 50Hz. The don't skip or insert whole cycles. – JRE Apr 06 '22 at 15:43
  • They insert additional cycles by deliberately increasing the frequency for periods of time such that an integer number of additional cycles is added. – user1850479 Apr 06 '22 at 16:48

1 Answers1

5

Thanks to the ubiquity of wireless communications networks, we have excellent frequency references available for relatively nothing.

Mains as a frequency reference is rather poor. That there's no guarantee anymore that the long-term drift should average to zero in reference to standard time.

A $100-class TCXO (temperature-compensated crystal oscillator) would have frequency within 0.1ppm of nominal or better. That's single seconds worth of error accumulated over a year. A $10-class TCXO will have frequency within 0.3ppm of nominal, for well under half a minute of yearly accumulated error. If you keep these oscillators under a constant load, with a well regulated supply voltage, they are about as solid as it gets without going for an atomic clock of some sort.

After soldering the part on the board, assembling the product, powering it up and aging the device for a few months - and rejecting those that drift the worst - you can calibrate out the off-nominal frequency error and get frequency stability within 1ppm for years and years, even for devices that go for $10 in qty 100.

There's no way to beat that with a mains line receiver in most circumstances. Plus, those TCXOs typically have very good jitter and short-term stability as well, so they work very well as frequency references for sensitive radio receivers etc.

You can get a cesium atomic clock on a chip for about $2.5k qty 250. Those hold phase down to a few microseconds per day, or a few milliseconds per year worst-case. If you were making a gadget that cost the equivalent of a Juicero, and wanted a superb real-time clock, you could easily run it off a "low grade" atomic standard. At least you'd get something genuinely useful for the price paid.

Recovering 60Hz line frequency is not really problematic - there's just so much of it, even in rural settings, as long as there's mains in the building or nearby (e.g. on the same lot).

You'd probably want a low impedance coil receiver sensitive to magnetic fields, with some selectivity, and then a preamp, further analog bandpass, more gain, then a high-resolution ADC to find the line frequency peak, isolate it, and drive a software-defined oscillator + PLL to keep in sync with it.

In terms of ubiquitous frequency references that are easy to receive, the cell towers are much better than mains, and I'm talking just about carrier frequency stability, without any demodulation. They typically phase-lock TX to a GPSDO rubidium standard. If you have even an extremely weak signal from a cellular base station, you've got a reference better than most already, better than free-running rubidium clocks. The base station doesn't transmit continuously so very short-term holdover is needed, but low phase noise TCXOs are already good at that.

I'm far from a time nut, so you could get information way better than what I got off the top of my head if you head on over to the Time Nuts mailing list.

Kuba hasn't forgotten Monica
  • 32,734
  • 1
  • 38
  • 103
  • 1
    All good info, and of course there's a society of Time Nuts, because why not. I'm surprised to hear rumors that the long-term accuracy of the power grid isn't very good. Short term, yeah. – Jim Mack Apr 06 '22 at 18:27
  • @JimMack from what I remember, there was a paper out of the interconnection governance people that the scheduled adjustments to catch up/slow down to the correct time were no longer worth the risk/cost to benefit, so they recommended not doing them. In the US at least, I think the Eastern/Western grids stopped doing them, but TX still does. – mbrig Apr 06 '22 at 23:17
  • As long as the grid is supplied mainly by rotating generators rather than inverters, the frequency control is at odds with dynamic stability and load flexibility. If ever most of the oomph will come from inverters - solar, storage batteries, and high-frequency geared wind generators - then the 60Hz can be locked to the SI second standard and left there. – Kuba hasn't forgotten Monica Apr 07 '22 at 21:14