What I am about to write may not be very popular, and I am likely to be shot down in flames. Please believe me that I mean no offence, and would just like to share an opinion that has been developed over many years of HF SSB operation, both on the amateur bands and on military HF comms circuits.
While the new SDR world brings many huge advantages to the radio experimenter, we must be very careful not to throw the baby out with the bath water when we discard analogue techniques. IMO, the speech processing area represents the very crown jewels of an HF radio transmitter, as it is critical to the overall operation of the radio. I therefore believe that developers should be very careful when they tinker in this area, and such work should not be undertaken lightly. Before presenting new code for evaluation, the following bench testing should be done. (On the air testing is VERY time consuming, and requires contacts under the right propagation conditions which are not always available) Test 1. Adjust the OUTPUT level on the UUT (unit under test) to give the required peak output power while whistling into the mic Use an average reading power meter. Then say "aaaaaargh" into the mic, and adjust the INPUT level for half of this power. Eg, 100W indicated for a whistle, and 50W for "aaaaaargh". Then go back to the whistle if necessary, and re-adjust the OUTPUT for 100W. Iterate as necessary to achieve both these results. For an RF clipper, this represents approximately 15 dB of clipping. If this 2:1 power ratio cannot be achieved, then the UUT fails the first test. Note the input level for the whistle using a scope. Test 2. Using an audio signal generator at a frequency of 700 Hz, feed in the same level into the UUT as the mic was giving for a whistle. Do NOT adjust the input or output levels of the UUT. Check the output of the radio on a spectrum analyser, and ensure that the products at 1400 Hz and 2100 Hz offset are at least 30 dB down. This can also be done at some intermediate test point if available. If both the above tests pass, then it is worth while to try some on air testing. The other parameter that affects on air intelligibility is audio frequency response. The mic itself should peak at about 1 kHz (eg Shure 444), and there should be ADDITIONAL electronic pre-emphasis with a time constant of about 160 us. This gives a roll-off below 1 kHz of 6 dB per octave. Many operators (especially the "AM HiFi bunch") will hate what I have said, but under difficult weak signal conditions, a radio performing as above will give the most intelligible signal. From a performance point of view, it really doesn't matter whether this is achieved by analogue processing external to the radio, or internally in DSP. The result will be the same if the correct principles are followed. In my experience, it is very difficult to achieve these results with processing performed at baseband. Envelope tracking comes the closest. In principle they are easier to achieve with IF/RF processing. 73, Greg, ZL3IX Ian Scoble wrote: > > > > > Please can anyone else confirm whether they are having audio punch problems > on SSB, the problem seems to have occurred since the SVN upgrade on or > around 1840. I did put a bug up on the board but this seems to have been > deleted other people I know are having the same problem and I was wondering > how wide spread it is. > > > > Thanks Ian G0CTO > > > > _______________________________________________ > FlexRadio Systems Mailing List > FlexRadio@flex-radio.biz > http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz > Archives: http://www.mail-archive.com/flexradio%40flex-radio.biz/ > Knowledge Base: http://kb.flex-radio.com/ Homepage: > http://www.flex-radio.com/ > > > _______________________________________________ FlexRadio Systems Mailing List FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archives: http://www.mail-archive.com/flexradio%40flex-radio.biz/ Knowledge Base: http://kb.flex-radio.com/ Homepage: http://www.flex-radio.com/
