RTL 234 kHz power, April 2018 / August 2020Pieter-Tjerk de Boer, PA3FWM firstname.lastname@example.org
Below is a plot of the signal strength of RTL's 234 kHz longwave transmitter as
received in Enschede, the Netherlands over the month of April 2018.
Vertical axis is relative power in dB; carrier power has been averaged over
1-second intervals and plotted as dots. Horizontal axis is UTC time.
At night, the signal strength is highly variable: fading due to skywave and groundwave.
At daytime, there's only the groundwave. The remaining fluctuation must be due to the transmitter power varying. We see that it varies quickly by about 4 dB, presumably due to dynamic carrier control techniques. But more surprisingly, we see that it is 1.8 dB lower on weekends!
Many LW/MW radio stations change their power in a diurnal pattern, typically reducing power at night when propagation is more favourable due to the skywave. But for a weekly pattern, the reason can only be commercial; perhaps fewer listeners in the weekend?
In fact, if we zoom in a bit, we see that RTL also has a daily power reduction:
every weekday evening at 18:00 UTC, 20:00 local time, power is also reduced by 1.8 dB,
just before the skywave influence starts:
We can't see at what time next morning the power is increased again, since that happens during the skywave period.
Presumably, all of this is done to save energy and thus expenses. Nominally, the RTL transmitter runs at 1500 kW, according to MWLIST. Subtracting 1.8 dB from that gives 1000 kW.
Looking back at earlier data, the weekend power reduction has been going on since at least November 2017. It is only tied to weekends, not to other holiday days, like May 1st (last day in the above plots) or the days around Christmas and New Year.
Update July 2020A message on Ydun's Medium Wave Info, mentioning daily transmission breaks for about 5 seconds, prompted me to have another look at my data. Indeed, there are such breaks, but they don't seem to occur at the same moment as the power changes.
By "folding" and averaging the signal-strength data, the power variations
can be measured more precisely. Thus, each pixel in the following graphs represents the
signal strength received at that moment (with 1 second resolution), averaged over
several months, during the summer of 2019 or of 2020, and separate for weekdays and weekends.
(In winter, similar behaviour is observed, but one hour earlier in UTC.)
First, look at the entire day:
We clearly see the power is higher during daytime hours on weekdays. The power goes up around 03:00 UTC (05:00 local time), down at 18:00 UTC (20:00 local time), and down further at 22:00 UTC (midnight local time).
Next, zoom in around 18:00 UTC:
Power goes down by about 2 dB at 18:00. At 18:20, the transmitter goes completely off. After about 7 seconds, it returns, but some 2 dB weaker than just before; it finally jumps back to its previous level at 18:21. Also in weekends, when there is no power decrease at 18:20, this brief switch-off happens. The ensuing 1 minute of low power seems to have been abolished for summer 2020 weekends.
In the morning, regular power is restored in a similar way:
At 02:55 the transmitter is off for a few seconds, then is back with 2 dB more power, and on weekdays goes back to full power at 03:00.
Summary/interpretation: It seems there are two different ways of power reduction being employed. One between 18:00 and 03:00 UTC (and all day long in weekend), and another more drastic one between 22:00 and 02:55. Which role the brief interruptions play, is hard to say. Perhaps part of the equipment is switched off and disconnected?
Finally, also during day-time hours, with stable ground-wave propagation, something
else interesting is visible:
The power seems to vary by about 1 dB on timescales of several minutes. What can this be?
- Deliberate scheduled power changes seems unlikely, as the change is smooth.
- Propagation seems unlikely, as this graph is average data over 2 months (2020) or 6 months (2019), and propagation effects can't be expected to repeat themselves so precisely over such a long time.
- Transmit power variation due to dynamic carrier control? Could be, but recall that the graphs shows the many-month average. So if this is due to dynamic carrier control, then what we're seeing here is the repetitiveness of the programs: with perhaps the same kind of modulation (news bulletin, music, etc.) at the same time every day.
If so, then it's quite surprising that that is visible!
Update August 2020On the Facebook page "Radios du monde", it is claimed that the brief interruptions are due to switching back and forth between their main transmitter at Beidweiler and their backup transmitter at Junglinster.
I did some phase measurements, from which it follows that with 99.7% certainty, the signal before and after the interruption comes from the same antenna; i.e., there's no switching between main and backup transmitter.
Here's a plot of the RTL 234 kHz signal's phase w.r.t. a locally generated 233.9999 kHz signal (which in turn is derived, in software, from the BBC's 198 kHz atomic-clock-stabilized carrier).
We see a couple of things:
- The phase is not constant, meaning Luxembourg's transmitter is not exactly at 233.9999 kHz. The slope indicates the actual frequency.
- The slope changes, meaning the frequency drifts, from about 233999.898 Hz to 233999.899 Hz in the 5 minutes covered by this graph. Nothing worrying, but proof that the transmitter is not locked to an atomic clock.
- Just after 18:20, the phase measurements are random during about 10 seconds: this is the brief period in which the transmitter is off.
- Before and after the interruption, the slope, and thus the frequency, is the same. It is extremely unlikely that this would happen if there were a switch from one transmitter to the other, unless both transmitters share the same frequency source.
- After the interruption, the phase continues where it would be expected, taking into account the slope.
If there really were a switch between main and backup site, the transmitter, the antenna, and the distance to
my receiver would all be different, each contributing some arbitrary phase shift. It would be very
unlikely for all these phase changes to "conspire" to give exactly the same phase at my location
before and after the switch.
(The phase measurement uncertainty is about 1 degree, so the probability of a random phase shift looking right is 1/360 × 100% = 0.3%, hence the 99.7% certainty mentioned earlier.)
- Repeating the experiment for the morning interruption and on other days shows essentially the same result.