From: MarkusVester@aol.com
Message-ID: <82.4ebf8ca.277f4842@aol.com>
Date: Sat, 30 Dec 2000 09:16:34 EST
Subject: LF: DCF spurious analysis
To: rsgb_lf_group@blacksheep.org
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Hi John, hi all,

G4CNN wrote
>  If you are sure that it is DCF49/42 please explain how you think the spur
>  frequencies might be derived. I would like to learn a little more.
>  Happy New Year, John, G4CNN

Yes John, I'm quite certain about the source of the problem. A while ago, I 
also  had a very interesting talk with Gamal Soegiono, who knows a lot more 
about these things than I do and lives much closer to Frankfurt. Yesterday I 
made some more detailed measurements of the relevant frequencies and repeated 
the attenuator tests to make sure the IM was not produced by the rx.

The two transmitters DCF42 and DCF49 are colocated at the same site 
(Mainflingen), using seperate antennas. The intermodulation is caused by the 
significant coupling between antennas and the nonlinearity in the tx output 
stage, in spite of the selectivity of the antenna matching circuits.

DCF 42 is used to transmit DGPS data and has a fairly complex spectrum, 
looking like a USB signal with a reduced carrier. This carrier (the "pilot" 
as it aids the synchronisation of the decoders) is at
 fp = 122.500 kHz, -15dBc.
The information is sent at a data rate of
 fd = 1187.5 Hz 
in a 2.3 kHz wide noise-like spectrum around
 fp+fd = 123.6875 kHz ("0 dBc")
in PSK (or is it MSK?). This contains most of the transmitted energy.

The signal is accompanied by a number of undesired sidebands. There is a 
significant line at 
 fp-fd = 121.3125 kHz, -48dBc,
caused either a second harmonic in the pilot generator or (more likely) by 
self-intermodulation of the data spectrum. This line again is accompanied by 
a set of satellites at
 (fp-fd) +- (fd/26) = (fp-fd) +- 45.67Hz,  -58 dBc
and smaller ones
 (fp-fd)+- n * (fd/104) = (fp-fd) +- n * 11.42 Hz,  up to -63 dBc.
These are apparently caused by 26- and 104-bit periodicities in the data. In 
the main spectrum you can actually find repeating frames of 88 ms duration.
The next harmonic to the pilot is also visible here at
 (fp-2*fd) = 120.125, -67dBc.

DCF49 is stronger, but simpler, and used for telecontrol of switched-tariff 
electricity meters and street lights. Most of the time, it transmits its idle 
frequency of
 f(DCF49) = 128.930 kHz,
accompanied only by the Luxembourg effect of several broadcasting stations 
(Deutschlandfunk on LF, Voice of Russia on MF). However, at least every 12 
seconds it is modulated by bursts of 200 bd, +340 Hz shift FSK. Like DBF39, 
the modulation is phase-continuous, so the the carrier does not return to its 
old phase after a data packet, which sets a 0.1Hz lower limit on the usable 
detection bandwidth for DX-experiments a la VK2ZTO.

The intermodulation spectrum is now easy to explain by the third-order scheme
 f(IMP) = 2* f(DCF49) - f(DCF42). 
All of these carry the DCF49 bursts at +680 Hz shift. The strongest 
components are the broadband data around 
 2*f(DCF49) - (fp+fd) = 134.1725 kHz, +41 dBuV/m = 38 W EMRP
and the pilot image
 2*f(DCF49) - fp = 135.360 kHz, +21 dBuV/m = 400 mW EMRP,
comparable to a good amateur station and thus perhaps a useful DX-indicator.

Some products in our band are 
 2*f(DCF49) - (fp-fd) = 136.5475 kHz, +1 dBuV/m = 4 mW ,
which was also noted by Alan 'NYK a while ago,
 2*f(DCF49) - (fp-fd) +- (fd/26) = 136.5018 kHz (!), -9 dBuV/m = 0.4 mW,
 2*f(DCF49) - (fp-fd) +- n(fd/104), eg. 136.4904 kHz, -15 dBuV/m = 0.1 mW.

The bad news of course is 136501.8 Hz, spot on VE1ZZ's frequency. There are 
only two things which could help us down here:
- Either one of the telecom technicians reads this and decides to readjust 
the traps at the tx location to get a few dB better decoupling,
- or Jack pulls his crystal a bit to qsy somewhere between .496 and .499 kHz.

Which of course brings up the matter of Loran lines. At my qth, I do see some 
in the upper part of the band, but below 137 kHz the noise appears to be 
higher, and they are quite faint if visible at all. Looking around in the 
neighbourhood of 100.0 kHz revealed three sets with periodicies of 5.5, 6.7 
and 7.5 Hz.

Ok, that was a lot of stuff, hopefully interesting to some.

73 es all the best for the New Year
Markus DF6NM