Hi Jim
> > This is > > important because often people are using compression and without > realizing > > that due to the SSB overshoot problem, the ALC now has to reduce the gain > > even more to avoid the splatter. In other words some of the dB you got by > > using compression, you give back because the ALC had to more aggressively > > reduce it due to higher envelope overshoots; but with CESSB you get to > keep > > all the compression dB advantages (plus the regular advantage of CESSB). > > Good points, Hans. But the important thing to realize, though, is that > audio is a very complex signal, whose spectral distribution, is > approximately logarithmic, as is the human ear/brain hearing system. And > as more than a century of studies of how we hear, engineers learned that > components of speech below about 500 Hz contribute almost nothing to > speech intelligibility, and while components above about 3 kHz do help > some, we can do quite well without them and save a lot of bandwidth (and > transmitter power). > > Bottom line is that steady state measurements of anything involving > audio are pretty meaningless. We can define system clip and assess the > noise floor, but that's about it. Two-tone tests for IMD tell us next to > nothing about what happens with program material. > Yes, all agreed. Only looking at steady state measurements is totally insufficient. Nevertheless the 2-tone IMD test is extremely useful! It's an easy measurement to make accurately on a spectrum analyzer, and even in real time so the effect of adjustments can be seen immediately. If you get good IMD results, then your TX has good linearity, and a lot of good performance flows from that automatically by itself. It's also very useful to be able to set up such tests which are effectively a sterile, lab test scenario which is repeatable, accurate and objective (no wishful thinking bias); and in the case of two-tone testing is quite a harsh test of a transmitter's capabilities. QMX can also measure its own transmitter IMD3, allowing it to self-calibrate to optimize IMD performance. But yes none of that has any relevance to CESSB nor Compression, equalization - that requires studies with real world speech samples. I had fun with a Winston Churchill speech https://www.youtube.com/watch?v=-vnJZ7-j6gc and the original QEX 2014 Hershberger CESSB article soundtracks (which are available for download from ARRL), even an AD-DC "Highway to Hell" track for ridiculousness (Video file: https://drive.google.com/file/d/1CWQJbARk5Rtt1Mz6utg0lcovryczoOXn/view?usp=sharing Audio only file: https://drive.google.com/file/d/1CWQJbARk5Rtt1Mz6utg0lcovryczoOXn/view?usp=sharing ) As well as many months of yelling into a microphone and recording the results. For most of the testing I used two transceivers with about 60-70dB of inline attenuation, using the transceiver's internal sound cards to provide sample audio tracks, and record the results for analysis. So I think you need both sterile artificial (but repeatable and objective) test cases - such as two-tone, as well as lots and lots of testing with realistic real world audio. CESSB, compression and equalization are all valuable for boosting the effectiveness of transceivers particularly at QRP levels. 73 Hans G0UPL http://qrp-labs.com https://www.buymeacoffee.com/g0upl ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[email protected] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to [email protected]
