"Toonfrequent" / Ripple control, May-June 2021

Pieter-Tjerk de Boer, PA3FWM pa3fwm@amsat.org

This page documents some measurements of the "toonfrequent" signals on the power grid. These signals, also known as "ripple control" in English or "Rundsteuertechnik" in German, are tones of several hundred Hertz, which are used to remotely toggle electricity meters between high (day) and low (night) tariff, switch street lights on and off, etc. In the Netherlands, these signals are about to be switched off on July 1, with the exception of streetlight control. So, high-time for a first look at them.

The following plot shows a spectrogram of the grid voltage. The horizontal axis shows time in seconds (from some arbitrary starting point), and the vertical axis frequency in Hz. The brightness shows the strength of the signal at that time and frequency.

The continuous lines at 50 Hz and its multiples are simply the 50 Hz 230 V power as delivered by the grid here. In theory it's a sine wave (so purely 50 Hz), but all kinds of equipment nowadays distort the sinewave, producing harmonics. The toonfrequent data signals are the interrupted lines, looking a bit like morse code signals with dots and dashes.

Analysis of these signals, recorded between May 9 and May 24, 2021, reveals the following:

• 228 Hz: these are by far the strongest, matching the fact that several sources say this is the frequency for my area (Overijssel). The signals seem to follow the "Semagyr 52" specification (see here for a list of data formats): a start pulse of 1320 ms, followed by datapulses at 720 ms intervals. Most messages repeat on a daily basis, with variations such as omitting weekend days or national holidays; see here for a list. Some have a varying time, presumably to control streetlights depending on sunset/sunrise and weather. Interestingly, there's one message that is tranmitted some 6 times per day, at 01:00, 03:00, 04:00, 05:00, 14:30 and 19:30.
  Note that the same signal also shows up at other frequencies that are multiples of 100 or 50 Hz away; presumably this is caused by the rectifying diodes in electronic equipment mixing the 228 Hz with 50 or 100 Hz harmonics.
• 183.3 Hz: this has a startpulse of 740 ms and datapulses at 640 ms intervals, which is not a listed format. Otherwise, it seems to transmit similar things as the 228 Hz data, with most messages repeating daily at the same time, or a somewhat varying time.
• 183.3 Hz again: on this same frequency, there's a weaker signal which shows up mostly around 3 o'clock at night, and also around 05:00, 08:40 and 22:50 (and perhaps more, I haven't searched exhaustively). It seems to follow the "Versacom über Ricontic b" format. Versacom is a more advanced system, see here; messages are not transmitted to actually switch something on or off, but to program a clock which will then do the switching at the appropriate time. This matches well with the fact that I only see these at night hours, and that the messages are much longer. Unfortunately the full Versacom specification costs about 100 euros...
  Both 183.3 Hz signals sometimes are on at the same time, and thus interfere with each other, as shown in this picture. So we may assume they are coming from different places, far away from each other, and that I'm somewhere in the middle.
• 183.3 Hz yet again: Twice a day, at 07:00 and 23:00, a third signal is seen on this frequency, with yet another format: a startpulse of about 0.5 s, and pulses in 0.578 s intervals. This could be the "Landis&Gyr" format or one of the "Semagyr 50" variations from the table.
  Also this signal interferes with the main 183.3 Hz signal, see the picture. Possibly, both the second and third 183.3 Hz signal originate from the same place: the third signal controlling some old-style equipment (possibly dual-tariff meters, given the timing), while all other equipment in their network has been upgraded to Versacom.
• 216.6 Hz: here we find a signal that looks like the "RWE" or "RWE(mod)" format from the list: start pulse of 1.56 s, data pulses at 577 ms intervals. Due to the close proximity to 228 Hz, decoding these is somewhat error-prone. Compared to 228 and 183 Hz, there seem to be more non-daily messages here, but that may well be due to bit errors caused by 228 Hz; this needs further investigation.
• 210 Hz: here we find a different format, consisting of a startbit and 10 databits without pauses between them; each bit is 600 ms long. This format is characteristic of Decabit. Analyzing this for daily patterns is harder, as presumably multiple of these 10-bit transmissions must be combined to form a message.

As noted above, some messages recur on a daily basis but not at exactly the same time. Here's a plot of the timing of many such messages in the evening, roughly around sunset time: Red, green and blue lines are for messages on 228, 183.3 and 226.6 Hz, respectively. The orange line indicates the time of sunset in the middle of the Netherlands, and the yellow line the time the streetlights in my street switched on.

Clearly, most but not all of the lines follow the sunset time, either smoothly or which lots of variations. The latter presumably are controlled by light sensors and thus depend on the weather. The streetlights in my street don't seem to be controlled by any of these signals though.

Besides their timing, another interesting aspect of these signals is their strength (amplitude). The following set of plots shows the amplitude (in volt) of the signals on the four frequencies as a function of the time of day. Some observations:

• The signals are quite weak; even the "local" 228 Hz signal has an amplitude of just a few volts, so about 1% of the main 230 V power itself. The other signals are about half a volt or less.
• There's a clear daily pattern; the signals are weaker during daytime, presumably because during daytime the power grid is more heavily loaded while the toonfrequent generators don't change their power.
• The daily variation is not equally large on all frequencies. Perhaps this is a hint at different daily electricity consumption patterns in different parts of the electricity network (e.g., more industry or more households), though I'm surprised this doesn't even out on this scale.
• The timing of the daily variation doesn't seem to be the same for all frequencies. Particularly the amplitude drop in early morning on 228 Hz seems to start earlier and be less steep than on 183.3 Hz. Again, this may reflect differ electricity consumption patterns in different parts of the network.

There are still many questions open, such as:

• What parts of the country do these signals come from, in particular the non-228 Hz signals?
• What are they all used for? This page lists a number of applications, but e.g. I don't see anything there that would need a signal six times a day, or only a signal in the second half of May.
• Can we decode the signals further, especially the Versacom and Decabit ?
• What causes the differences in the daily amplitude pattern?

...To be continued...