Below is an image transmitted from W0LMD, in Colorado, to W9NTP and KB4YZ, in Indiana, and to W8ZCF, in Ohio, during the first HF tests, on 29 Nov. 2000. At this time there was only one level of Reed-Solomon coding. Two methods were tested. One method used 4 subcarriers, with 7 different possible phase changes for each subcarrier. The other method used 5 subcarriers, with 11 different possible phase changes for each subcarrier.
Below is the same image, but with each pixel expanded by a factor of 4, both horizontally and vertically.
Below is a spectrogram of what was received as one of the test transmissions during the test, on 20 meters, of 29 Nov. 2000.
In this spectrogram, time increases from left to right, and frequency increases from bottom to top. The tic marks along the time axis are 1 second apart. The tic marks along the frequency axis are 200 Hz apart. The amplitude scale is the sequence of color bars at the left. Each color bar spans 6 db. Thus, the boundary between the yellow and green color bars is 6 db below the boundary between the red and yellow color bars. What is plotted is the magnitude, in db, of the energy of the received signal, as a function of time and frequency
This spectrogram is from a case with only 4 subcarriers. The energy distribution, due to the modulation of the 4 subcarriers, is seen as the 4 (about 150 Hz wide) bands running horizontally across the spectrogram.
Note the diagonal band running from near the beginning, in time, of the top subcarrier down through the next to top subcarrier. The brown portions of this spectrogram are 30 db down from the light green portions. Thus, during the first 5 seconds of this transmission, propagation effects were like a 30 db deep notch sweeping across the top two subcarriers.
Also note that during the last second of this transmission, there is curved band of attenuation affecting mostly the top subcarrier, and somewhat the next to top subcarrier.
In order to sample the frequency response of the channel, a chirp signal has been transmitted before and after each phase modulated transmission. The chirp signal consists of 3 audio subcarriers whose frequency is linearly increased with time. Each of the 3 chirp subcarriers takes about 0.31 seconds to sweep its respective portion of the audio passband.
Due to a mistake on my part, the data below 850 Hz in the chirp signal used in the tests of 29 Nov. 2000 are not valid.
Below is a plot showing the results from demodulating the top two chirp subcarriers from chirp signal leading the transmission whose spectrogram is shown just above.
Note the 30 db deep notch centered just above 1000 Hz.
Also, note the region just below 1400 Hz. This is where the end of the middle subcarrier and the beginning of the top subcarrier intentionally overlap in frequency. There is about a 0.3 second difference in time between when this overlapping range of frequencies is sampled at the beginning of the the chirp signal (by the top subcarrier) and when it is sampled at the end of the chirp signal (by the middle subcarrier). The plot below shows that the frequency response increased by about 6 db over this range of frequencies in about 0.3 seconds. This is a rate of change of 18 db per second.
Below is a plot showing the demodulated chirp signal leading an earlier transmission by W0LMD on 29 Nov., 2000. This shows that at this time, there was a notch about 9 db deep near 1000 Hz and another notch, about 3 db deep just below 1700 Hz.
Below is the demodulated result from the chirp signal trailing the same transmission from W0LMD as just above. Thus, this "trailing" chirp signal measured the frequency response about 10 seconds after the measurement made by the "leading" chirp signal.
Note the relative flatness of the frequency response, show by this "trailing" chirp signal, compared to the distinct notches shown by the "leading" chirp signal.
For comparison purposes, the image below is shrunken version of a spectrogram for the ideal case of a test signal used in the Australia to United States tests. This spectrogram is of a portion of the .wav file sent to the sound card supplying audio to the transmitter. Note that all of the tic marks did not survive the "shrinking" process.
The spectrogram is 986,141 bytes long, and has been removed, to save space.
For comparison purposes, the image below is shrunken version of a spectrogram for an actual received signal transmitted from Australia to the United States.
Compare this spectrogram to that of the ideal case above.
The full spectrogram is 1,012,358 bytes long, and has been removed, to save space.
Links to: