The 3 SSB modes seem very similar, the last one is slightly less good - which means that, to reduce receiving effects, "Probably Optimal" should indeed be optimal.
I've tried the 3 different SSB modes with strong multipath clipping.
Basically, I clip fast differences in the signal, which I
think is what causes multipath distortion.
To my surprise (although, not really, this actually matches a post I made a half year ago on some radio site), the first ('traditional') SSB mode seems to be WORSE then DSB, meaning that the multipath clipper has more effect! Here are all the results:
0. DSB mode.
This is what we will compare with.
1. SSB Traditional mode:
About 0.1 dB MORE highs loss. So this means that the waveform actually moves faster in this mode
2. SSB Probably Optimal mode:
About 0.1 dB LESS highs loss.
3. SSB Steep Optimal mode:
About 0.1 dB LESS highs loss.
2 vs 3: Nearly identical, but 3 is just slightly better (!). 0.03 dB or so.
My initial guess was that something happens around 12.667 Hz, because for a sound at this frequency, the LSB frequency is 38000-12667=25333 and the USB frequency is 38000+12667=50667, exactly twice as much. That's why I *think* that for frequencies above 12.667 kHz it's good to use SSB, for frequencies below it DSB is better, near 12.667 it doesn't matter match. That seems to correspond with the fact that method 3 seems to work best, this uses a filter that starts at 10 kHz and reaches SSB at 12.667 kHz.
It could however also be that 'tones' are better in DSB mode and hiss-like sounds in SSB mode. In that case, a separation at a slightly lower frequency (around 5000 Hz) would be better.
Real-world results would be very welcome to check if my ideas are correct!
Quote:
What my 'gut feeling' was telling me, and measurements that I did tonight now seem to confirm, is that SSB mode does not help against multipath distortion for lower frequency tones. A 1 kHz tone that's placed at 37 kHz, 6 dB louder, instead of at 37 and 39 kHz, doesn't help much, in fact overall it seems that it causes the FM modulation frequency to move faster than having both tones present at half-power.
What I think is that something special happens at 12666 Hz, since (38000-12666) * 2 = (38000+12666). So, my guess is that, to get less multipath issues, SSB would be useful for this and higher frequencies. On top of that, a clipper or limiter that limits the maximum modulation change in a given amount of time should help (although that also will hurt stereo separation and in case of very loud highs even mono audio quality).
I'll start working on the performance tomorrow, might be a bit problematic though.
Register
Login
FAQ
Search
