The short audio path is the best audio path. Radio stations are generally not designed around this concept. Most of them to the extreme opposite.
Physically short? Electrically short? Geographically short? What means "short"?
With regard to the Compellor or any similar processor, your signal is going through many op amps and many coupling capacitors.
Ok, well I happend to have a 320A on the bench now with the manual open.
The opamp count: For a single audio channel, there's the input diff pair (1 or 2 depending on how you consider a differential receiver circuit), a differential VCA drive (again, 1 or 2), a VCA buffer, output gain amp (and inverter), and the output amps. So, that's 4 or 6 opamps for your "many". And one VCA.
Caps? There's one in each leg of the input for DC blocking, there's another in each leg of the output gain amp, one in each leg of the output amp, and...let's see...that's about all in the audio path. 6 total.
All of the op amps and VCAs are connected to power supply "rails" that are supposed to immediately provide a low impedance voltage that exactly matches the signal swing that the op amp needs at the output. The truth is that power supply rail impedance at audio frequencies in broadcast grade equipment is going to be much higher than we really need.
But is that really the "truth" or something somebody read or told you? Because, in the schematic there are local "audio decoupling" (what they're called in the schematic) caps that take that impedance pretty far down in the audio band.
Just put an oscilloscope probe on one of the op amp power supply pins (AC coupled) and watch all of the audio signal dance around riding on the DC power. It's the tail wagging the dog kind of thing.
Ok, well since I have it open and on the bench, let's just do that. Here's one of the opamps in the output circuit, arguable a higher current position. The scope is attached to the power supply pins on the IC, it's AC coupled, and input up so we can see the noise. Here a shot of one of negative supply rail, with no audio applied.
Here's the same neg rail with audio enough to deflect the output meter to full scale.
Here's the positive rail, no audio:
And the positive rail, same full output audio:
I don't see any wagging, sagging, lagging or dragging. All looks pretty much the same.
And I really need to clean off my scope screen.
That sloppy DC impedance and inadequate power supply decoupling lets op amps interact with each other at the power rail level when they aren't supposed to do that.
That might be true if there were:
1. Any such thing as DC impedance (which there isn't)
2. The power supply actually was not adequately decoupled (which it is)
3. Any condition actually existed that might cause opamps to interact with each other (which there isn't)
"But the distortion test looks good" is a bad excuse. Distortion is done with sine waves and nobody but your distortion analyzer listens to sine waves.
Well, I don't just test with sine waves, but they are the best test for linearity. In fact, there isn't a better one.
Music demands and instant steep power peak and the power supply is supposed to deliver that and still keep it's composure for the next few moments afterwards.
Um...well, no. Power is the result of voltage times current. In most of the opamp applications within a Compellor (or most other devices), there is voltage gain, but very little actual power distribution through audio in the circuit. So no, the actual current does not spike with music "demands". A power amp is different, and that's not what these things are. But also, there is no actual "instant steep power peak" anyway, it's all limited to the maximum frequency and amplitude of the audio waveform. There are lots of much faster signals, and they too can be handled.
Poor power delivery plus the fact that every single component that audio directly passes through adds distortion. That means capacitors, op amps and even resistors add distortion.
Yikes, what a horrible world we live in. But we don't actually always have poor power delivery, and not every component adds audibly to distortion. Capacitors, operated correctly, do not cause distortion. Only of the wrong types are chosen or applied wrongly in the circuit. Resistors distortion contribution is far, far below anything audible, or anything else in any circuit. Opamps distortion is well known, and also vanishingly small. Your distortion generators are all transducers, mics, speakers, headphones, phono carts, tape heads, and so on, and by several orders of magnitude. Discussing them as significant signal modifiers along with resistors is just overblowing the absolute into the absurd.
It has been measured and proven.
I don't think you really want to go down that rabbit hole.
Distortion and noise in an audio path is cumulative. The longer the audio path the more distortion and noise will add up.
I guess "longer" in this case means "more devices". Sure, in the absolute, but in the practical world how would one device with a noise floor at, say, -98dBU add to another noise floor at -98dBU? Assuming they are uncorrelated, but identical spectrum and amplitude, well that's easy, +3dB. But is that a problem? And does that occur? Because idencial noise levels is rare, as is an identical noise spectrum. 3dB add is worst case, are actually rare. Distortion doesn't simply add. To do so the specific nonlinearity that caused it would have to be identical.
If I can avoid using a Compellor I will do that.
Yeah, I think I got that part.
I did my critical listening with a Compellor set so that the signal path was through the VCAs and with input and output levels matched but without the dynamic side chain moving the levels. It was a fair in circuit versus bypass test and the difference to me was noticeable.
My next response, if I wanted to really persue this, would be to ask if the test was a true ABX/DBT. But lets not.
Look, you hate compellors and broadcast audio chains. Ok fine. Some have them in their lives, not by choice, but by necessity. Others can shorten the chain, and make everything a digital stream. But then, how does someone like you feel about 24/48? Resolution too low? Or realize that there actually is no 24 bit audio because there are not real 24 bit performance ADCs? Or is anything digital bad because of the mythical stairsteps (which don't exist) or the slicing and dicing (that doesn't happen)?
How can you tollerate this kind of audio business at all??
Amused, here.