When I operate SSB, I monitor the ALC meter and I modulate such that the meter frequently "bounces off" 0dB. This will create some very mild compression; however, it does NOT distort your signal in any material way. It does not broaden your signal. Also note that, unlike other radios, the ALC will NOT exceed 0dB. It is designed that way to avoid any possible "clipping" of the digital signal —- when you hit 0dB, there is no clipping but the signal is limited to that level.
To drive the audio chain in PowerSDR in the way the design intends, you should be frequently hitting 0dB on the ALC meter; frequently hitting 0dB on the Leveler if it is in use; and achieving 0dB or above on MIC if the Leveler is not in use. The speech processor we have in PowerSDR mRX PS is NOT the old compander that was used some time ago, it is a much different algorithm, as are the Leveler and ALC algorithms.
It's certainly possible that some other radios are peaking over 100W … some are known to have "ALC Overshoot." My bigger fear is that our operators are often backing off on gain, trying to stay away from the 0dB line, because of their prior experience with Flex or analog radios. Hitting 0dB does not cause flat-topping or any material distortion in the digital chain of our radios. That's different than the "old" PowerSDR.
As far as the ALC, it uses a short delay line. That enables it to look for a peak coming on the "horizon." When it sees such a peak that would exceed 0dB, it gradually decreases the gain such that when the peak arrives and its output is generated, it will peak at 0dB. After the peak passes, assuming another is not coming, the gain gradually returns to 1.0. Actually, the ALC just monitors the situation unless there's a peak that would exceed 0dB; then, it deals with the situation. There is no clipping anywhere in the audio pipeline.
(NR0V 29Oct14)
There is a very good video by G7CNF on audio setup using a Flex under PowerSDR. NR0V posted some OpenHPSDR-specific follow-up tips:
- It's certainly fine to have the MIC meter and EQ meter hitting in the same range, near 0dB. However, the only one that really matters in terms of what happens later is the EQ meter. Since there's no limiting of any kind until after the EQ, these are just two adjustable gain stages in series and the output level of the second is what's important.
- The leveler functions as a constant gain until there's a peak coming that would cause the output to exceed the maximum level (currently +0.4dB). In that case, it gradually adjusts the gain down, along an exponential curve, before the peak arrives. So, saying it another way, the leveler does nothing except provide a constant gain UNLESS you are peaking at +0.4dB, in which case, it performs some "leveling".
- The leveler also sets the level to be correct for the following compressor. So, if you're using the compressor, I recommend using the leveler.
- The ALC just monitors the signal until a peak is coming that would exceed 0dB —- in which case, it gradually reduces the gain, along an exponential curve, so that the peak hits exactly 0dB, no more, no less. Unlike the example in the video, adjusting the attack time has no effect on the level the peak eventually reaches. I recommend an attack time of 2ms. Adjusting the decay time could have some effect on subsequent peaks that are less than 0dB just because gain will be increasing during the decay period.
- Finally, having some peaks hitting 0.4dB on the leveler or 0dB on the ALC does introduce some mild compression; however, it does NOT introduce any "flat-topping" or any material distortion. There are reasons (some current and some related to future features) that I'd recommend operating with peaks at those levels or very close.
- In our software, there's no need to "worry" about exceeding 0dB ALC. It won't happen —- that part is taken care of automatically for you. Also, a bit of ALC compression is not harmful in this software. Said another way, our software could be viewed as being more automatic and user-friendly in dealing with the level / distortion issue. As was pointed out in the video, EXCEEDING 0dB in an SDR such as this can have very negative consequences; you don't have to worry about that with this software.
- Also note that a Compander is not used for compression in our software. Instead, we use the mathematical equivalent of an RF Speech Processor which is enabled by the "COMP" button.
- If you are hovering at 0dB, you will hear some compression. However, this compression does NOT create IMD (or worse). (The only exception is that since this increases the average signal output, it will increase any IMD generated by the analog amplifiers in proportion.) There will be no clipping. It's difficult to quantify in words at what point you begin to hear any compression as a result —- I'd suggest you try it. Just listen to your audio and see what you're comfortable with. The good news is that you don't have to worry about the very negative effects that result from clipping a sample stream or overdriving a DAC.
- Results and required settings vary quite a bit with voice and microphone. I normally use 5dB of leveler gain and 7dB of compression. People who know my voice say they don't hear much compression until I get to 10dB.
- In SSB, the block diagram is: MIC » 20dB HW Preamp » DSP MIC Gain » EQ » Leveler » SpeechProc/Compressor » Filter » ALC
- The amplitude of the signal from the MIC depends upon (1) the MIC, and (2) the MIC Gain. There is no inherent limiting of this signal unless you were to exceed the dynamic range of the MIC input hardware (very unlikely). The EQ does no limiting of amplitudes either. The Leveler limits its output amplitude to 1.0. So, if the Leveler is ON, you have a controlled input to the SpeechProc. If the Leveler is OFF, you do not. Therefore, with the Leveler OFF, how much compression you get will depend upon your MIC, MIC Gain, and EQ. An alternative is to turn the Leveler ON and set its gain to 0dB—in effect, it's controlling the output amplitude but not actually providing any 'Leveling' action.
- Mic Boost turns ON/OFF a 20dB analog preamp that is contained in the TLV320 CODEC chip. So, if you turn ON the Mic Boost, and have a high microphone level, you could overdrive the preamp and/or the ADC. MIC Gain is later applied digitally; so, there's no way you can undo that damage with a Mic Gain setting.
(NR0V 6Dec13, 11Aug14)
W1AEX's Quick Instructions
- with the TX multimeter set for "MIC", set the audio level so the peaks reach anywhere between -10 dB and 0 dB
- with the TX multimeter set for "EQ", set the EQ "Preamp" slider so peaks are regularly reaching close to 0 dB.
- this does not usually result in regular excursions to 0 dB on the ALC scale, so use the Leveler to bring the TX audio gain up to the point where you are pushing the ALC to 0 dB on most audio peaks.
comment: If you're using an outboard audio processor, you may want to turn off COMP and DSP Leveler.