Conclusions on the audio processing project

 I've spent some serious time with the audio processing monster I've created.... and described here and here.

Now, after working with it for a while, I've found a few surprises along the way... both good and bad.   Remember, that my object was to raise the average power output by several DB, and to ensure being heard more often. 

The result was somewhat less successful than I expected in some areas,  and wildly successful in others. Let's break it down.

As noted in the other articles, the average power in SSB signals in conversational transmissions, amounts to about 25% of rated transmitter power.  I calculated that the signal could be raised to around 80%  on average, assuming some serious manipulation of the audio stream prior to being input to the radio.

 Trouble was, as I noted in the other articles,audio overshoots, particularly on transient peaks.  Adding that 8db to my signal with the internal proc alone, added a mess of distortion, both on the audio and presumably on the RF signal as well.  since it-pounded the poop out of the ALC. Didn't do much for the intelligibility, either. The ALC and internal compressors were not up to handling the levels I was sending them.  That said, I found myself encouraged by the transmitter output,which seemed to have the headroom to accomplish most of the gains I was after. If I could control the peaks, I'd have it.  So, here we go.

I generally have a Heil proset-6 headset running to the mix console. Advantage: Close miking ensures a consistancy and purity of sound you simply can't get with a desk mic or a boom mounted mick, both of which I also have available. 

 Mix Console is a Gemini GEM05USB mixer to bring the mic up to lline level.

Understand that I'm bandpassing the audio at 100Hz to 2900Hz (-18db/Oct) at the Mix Console Outputs. No point in feeding the processors audio I'd not like to transmit anyway. And frankly, having the procs act on audio outside the radio's bandpass is counter-productive, and creates wholes in what Phil  Spector used to call "the wall of sound."

Within that bandpass,  I added an audio equalizer to the microphone amplifier which I have set for about a 5DB boost at 2100 cycles, and a similar sized cut to around 100Hz and below.The result of this setup is that the processors are working harder on the boosted highs than everything else... maintaining a consistently very bright and punchy sound.

From that EQ, the path runs to the 2600 pro, which is a noise gate/expander/ AGC/ limter/DeEsser

Understand, that the 2600 is designed as a Stereo unit, which can be configured as Dual, independent Mono. I'm using it in dual mono mode, with the output of the first channel driving directly into the second channel. The first channel is set for slow AGC and some fast De-esser action.  The second is pure peak limiting. 

 Both channels employ some expansion... the first slow, and the second somewhat faster. This is to counter-act action further along the chain. The function of the noise gate is obvious... it turns the gain down when I'm not saying anything which means I'm not picking up echoes in the room when I pause. Once the audio makes it over the gate, it's picked up by the AGC and amplified to a fairly tight dynamic... one the peak limiters and the clipper-follower can use.

The only fly in this ointment is the internal compressor in the 991 which I cannot defeat. I've been able, however to adjust the gates/expanders on the 2600 to counter the effect. To obtain full power, the internal compressor ends up being driven to around 15db of compression. A bit high for my tatse, particularly given what the rest of the audio chain is doing, . No getting around that. So I'm forced into dealing with that with a slight of hand in the 2600.... since that internal compressor in the 991 is rather slow-acting,  using the 2600's expanders to balance against that compression seems to work well.

Yes, the setup is complex, to be sure and literally a balancing act.  But, once you get it right, it doesn't need much daily tweaking, if any.

The results are encouraging: I'm finding far less trouble is cutting through pileups, even if my peak signal is smaller at the received end. The ALC meter is running, normally at around 40% of total swing, which is somewhat below the recommended 50%, while the wattage hovers around 80%-90%. The lack of overshoots has the opposite sideband leakage under control, and the reports I'm getting are "armchair copy", even when getting low S-meter readings.

Total investments:

The 2600 cost be $160.
The patch cables about another $76..(I had some of what I needed ion hand)
About a week of every day use to get things dialed in. Very well worth the results I'm getting.