Radio signals during a leap second

This web page documents what many radio stations (mostly broadcast) did during the leap second at the end of 2008. Many broadcasters transmit time pips (e.g., 6 pips at 1 second intervals, or 2 short pips followed by 1 long beep) at the top of the hour. These may sound very precise, with the hour starting precisely at the beginning of the last pip, but it turns out that many of them are actually quite inaccurate.

(For a similar set of measurements of the 2005 leap second, but limited to the longwave range, see here.)

Time/frequency reference stations: DCF77 and RWM

Like in the previous leap second, the German longwave transmitter on 77.5 kHz got it right:
[waterfall showing DCF77]
Like all other diagrams on this page, this is a so-called waterfall diagram, which shows the signal strength (by color, from black via blue and pink to white) as a function of frequency (horizontal axis) and time (vertical axis).

The Russian time signal on 4996 kHz also got it right:
[waterfall showing RWM]

My recordings also covered MSF, HBG, WWV and CHU's frequencies, but those signals were too weak to be discerned.

Longwave broadcast stations

In Europe, there is a longwave broadcast band between 148.5 to 283.5 kHz. These transmitters use amplitude modulation to transmit audio to the general public, and thus are not necessarily precion time and frequency reference stations. However, many do transmit time signals in the form of 1 kHz beeps at the top of the hour. These can easily be seen in the waterfall diagram.

Here's the waterfall diagram of this band during the 2008 leap second:
[waterfall showing the longwave broadcast band]
(A similar picture of the situation 1 hour earlier is here.)


Frequency Station Observation
153 kHzDeutschlandfunk, Germany Leapsecond ok, although it seems a fraction of a second early, but in the previous hour the time signal was 1 second early!
177 kHzDeutschlandradio Kultur, Germany Broadcast two long beeps, 1 and 2 seconds too late. In the previous hour, the only long beep was 2 seconds late. So apparently their clock is 2 seconds late (or there is 2 seconds of delay in their transmitting system), but did capture the leap second.
198 kHzBBC No time beeps, but the bell sound starts exactly on time after the leap second.
207 kHzDeutschlandfunk, Germany Leapsecond ok.
225 kHzPoland Missed the leap second.
252 kHzIreland & Algeria Time signal is more than 5 seconds late! An hour earlier it was 6 seconds late, so they didn't insert the leap second.
An Irish reader of this page (see here for his leapsecond observations) pointed out that the Irish station did not send any time pips, so the visible late time signal those must have come from the Algerian station.
Listening to the demodulated recording, one can hear both stations, with a voice count-down from Ireland and pips plus a French announcement from Algeria.
261 kHz? Missed the leap second.
270 kHzCzech Was 3 seconds early.

So out of 8 stations transmitting a time signal, only 3 got it right.

Mediumwave broadcast stations

Here's a picture of the center 600 kHz of the mediumwave broadcast band:
[waterfall showing the longwave broadcast band]

Correct leap second:

Time signals without leap second: Other errors: Not all of the above observations can be derived from the graph; some are too weak to be visible in this graph, but were observed in a more zoomed-in graph and/or by listening.

Conclusion: about half of the stations got the leap second right.

Note, both here and in the longwave plot, that the time signals are not always precise to a fraction of a second, presumably due to delays in the distribution networks (e.g., satellite links). An extreme case is the (Dutch) station on 747 kHz.

Shortwave broadcast stations

Waterfall diagrams of the 49 m and 41 m shortwave broadcast bands are here and here. Again, some got it right, and some didn't.

Test setup

All of the above pictures were made using the latest version of my home-built Software-Defined Radio system, see here. I used this to record six parts of the radio spectrum, with a total bandwidth of 1.8 MHz. (Actually, the system could have recorded an even larger bandwidth, if the power supply had not been insufficient, which forced me to run the ethernet connection to the PC on 100Mbit/s instead of 1 Gbit/s.) The antenna was a dipole for the 40 m band, aided (for the lower frequencies) by a ground connection.

Comments are welcome, at