CW stations


Poul-Henning Kamp
phk@FreeBSD.org
Revision 2003-04-30

Continuous wave stations

There are a lot of other stations which carry usable frequency and time information, and provided they are inside my pass-band, it is possible to detect and even decode them.

There is a good and comprehensive list of timesignal stations at Klaus' Radio Page.

Here are som plots of amplitude (red, arbitrary units), and phase (green, -PI...+PI scale) of some of the more prominent signals. The X-scale is one second in all cases, and the start of a second is located at the same place in them all.

To create these plots I have just taken the raw A/D measurements, run them through a two-pole recursive peak-filter, mixed the signal down to baseband and averaged over a number of buckets, typically 1msec wide. The signals are not recorded with the same parameters, so don't compare them too closely.

Other more challenging signals like the 66.666...kHz RBU signal from Moscow takes much more heavy-handed filtering, but even that seems to yield to 351 sample FIR filters.

Here we have several seconds of the venerable and reliable 77.5 kHz DCF77 signal from PTB in Mainflingen.

The AM modulation with second markers of 100 or 200 msec are clearly visible in the amplitude, but more interesting is the 512 bits PRNG phase modulation which is barely visible.

PTB has sent me the specification of the PRNG signal, but it has yet to arrive. I have played with decoding it, and I think I have got it right but I will not publish the results until I have compared to the real specification.

Here we have several seconds of HBG signal at 75 kHz from Switzerland.

The AM modulation is almost, but not quite entirely identical to the DCF77 scheme, the second markers of 100 or 200 msec are visible. I have no indication that they employ the same phase modulation as DCF77.

The major problem with using a signal as weak as this, is that the number of "bit-errors" in the timecode makes it tricky. I have played with some code which tries to predict the timecode and does a correlation with it over the received signal. Despite the fact that this timecode was not built for correlation use, it actually has some pretty good correlation properties. More on this when I have time.

Here is the slightly quaint signal from MSF Rugby at 60kHz.

This signal is AM modulates with two bits per second. That means that each second can have the form -_---, -__--, -_-_- or -___-. In addition to this the top of the second has a 500msec pulse.

I often have a lot of interference on this signal, in particular I suspect that one of our domestic appliance monsters uses a frequency controled motor which operates on PWM in this range.

This is the French LW transmitter at 162kHz ("France Inter").

Since they are transmitting a voice and music signal in normal AM modulation, they have superimposed the time signal by phase encoding. Every second the phase will wander first a radian ahead, then one behind and then finally back to zero. The second zig-zag carries a single bit which encodes a DCF77 like timecode. I am currently not aware what the remaining bits encodes, it could be DGPS signals.

This is my local "LoudenBoomer", Kalundborg Langbølge sender at 243kHz. Here is an article about the history of the transmitter.

This is a pure voice/music transmitter, so the phase stays pretty steady.

Being located only about 40 km away, the signal is very strong. Their frequency is derived from a rubidium frequency standard, but their transmitter has a fancy signal-matching power supply which due to changing voltages and capacitances shift the signal about 100nsec forth and back. I hope to document it one day. They turn it off at night, which is nice because it makes it simple for me to see if their signal gets filtered out properly.

Ohh, and of course I can demodulate their signal and play it on my sound card. Been there, done that. Unfortunately they have the most boring programming in Denmark, fish-prices, weather reports, notices to ships etc.


Valid HTML 3.2!