I did some measurements for a few days in november on ionospheric doppler using whatever "constant" signals have been available, like WWV, CHU, several broadcasters on 49 meters, including Radio Habana Cuba, whose transmitters are some 20 km away from my QTH and become a mix of NVIS and ground wave. I used Spectran to obtain the carriers spectrum, and I mean plural, because you get several "threads" dancing on the screen in frequency and amplitude. I shifted the carriers to 1 kHz, offseting the tuning by that amount.
The results have been interesting and a bit surprising. You may see several threads on the screen, spread around some 5 Hz. The frequency moves on the average at a rate of about 0.1 Hz per second (more or less, I was looking for absolute shift more than rate of shift, but both do matter). At times, some of the threads dissapear, and reappear shifted about 1 Hz per thread, sometimes jumping from one extreme of the carriers bunch to the other. The analysis bandwidth was between 20 and 50 Hz wide. I was also playing with Spectran and learning how to use it. That kind of behavior is a killer for many modulation schemes. As Patrick says, PSK10 suffers much more than higher speeds because the bit duration is an important fraction of the doppler shift rate, and you never get anything as stable phase. But also, when relative amplitude among the several threads varies, the carrier regenerator may jump from one to another. I did not look either for amplitude calibration during those first attempts. It may require periodic screenshots of the Spectran spectrum in the upper window. Reanalyzing some of those screenshots you may see amplitude differences as large as 20 dB at times and as small as 1 dB on some ocassions. But I was looking at the waterfall, and did not pay too much attention to amplitudes, and the screenshots timing was random between those patterns I found interesting in the waterfall. I don't know so far if Spectran might be able to produce a printed log of the amplitude vs. time distribution of the threads amplitude (possibly not...). Multicarrier modulations must also suffer, because if you use OFDM the jump from one of the "outer" threads to the other extreme, 5 Hz away, may be a cause for MT63 to refuse to decode at all sometimes. That is on the lower boundary of usable speeds. But you are bound also by ISI (inter symbol interference) at the higher usable speeds. One of the strategies of Pactor III is supressing some of the carriers at the lower data rates, and it sort of protects the signal by spacing the frequency bins a bit farther away, which may become more inmune of doppler causing the carriers to fall in the adjacent (wrong) bins. I did not measure delay spread, and would appreciate suggestions on doing so. And I only see the time standards (CHU, WWV) as suitable using the second ticks as time markers. I am open to suggestions. I cannot be conclusive about doppler spread now. It does not seem any worse on the higher frequencies, but then again, paths are different, and it does have an effect on what you get on the screen with Spectran. 73, Jose, CO2JA --- Patrick Lindecker wrote: > > Hello to all, > > There are also some basic axioms (I suppose so) as: > > 1) "Doppler iosnospheric modulation" acts on PSK modes (the phase is > "dancing" randomly, the Costas loop cannot stabilize and the PSK > decoding is quite impossible). The more the modulation speed is, the > less this Doppler acts on the phase difference (which determines the > bit). Hence, PSK10 is very sensitive to Doppler, PSK31 is less sensitive > and the effect is very weak on PSK125. > > 2) "Doppler ionospheric modulation" effect is linear with the HF > frequency (smaller in 3.5 MHz that in 7 MHz). It seems logical but I'm > not so sure of it... > > 73 > Patrick __________________________________________ V Conferencia Internacional de Energía Renovable, Ahorro de Energía y Educación Energética. 22 al 25 de mayo de 2007 Palacio de las Convenciones, Ciudad de la Habana, Cuba http://www.cujae.edu.cu/eventos/cier
