Question about VCA sidechains and control voltages.

[Mr. Bill]

Hello,

I'm in the beginning phase of building a stereo VCA compressor just to see if I can pull it off and I have a couple questions. The signal path of most VCA compressors is pretty straight forward....in, op amp, VCA, op amp, (sometimes transformer), out. However, here's what I don't understand about most designs I have studied. The sidechain that gets fed to a second VCA in order to control ratio, attack, release, etc. by sending a control voltage back to the signal path is usually a cheaper or lower quality VCA. Does this affect the action of the signal path VCA in any way? Would the compressor track better if the higher quality VCA's were used in both the signal path and the sidechain? If the sidechain VCA is supposed to mimic the response of the signal path, then I don't understand why almost every commercial compressor uses a different type for the sidechain.

[Enzo]

Please post an example schematic.

The side chain is turned into a control voltage which then controls the VCA. The side chain doesn't have to react to every cycle of signal. It turns signal into an average.

[Mr. Bill]

Here is probably the most popular one that people build. It's a popular clone of the old SSL bus compressors. I guess I don't understand why the sidechain is taken from the audio path and fed to a separate vca summed to mono to gain the control voltage. I apologize if my ignorance is showing.

[Leo_Gnardo]

I don't understand why the sidechain is taken from the audio path and fed to a separate vca summed to mono to gain the control voltage.

If the control voltage wasn't "mono" to both vca's, you would have the situation where each channel's gain is independent and could easily result in dynamic unbalancing of the "stereo image" as if a balance control on a stereo hi fi amp was being randomly twisted left and right. Could make for an interesting effect though...

Those buss compressors were a hot underground item amongst the insiders in the mid 1990's. Chris Lord-Alge introduced me to his. Some technician at SSL apparently was ghost-building them and selling 'em out the back door. From what I hear he moved about 40 of them before he got pinched and then the party was over. These days I wouldn't be surprised if some outfit was making a PC board kit. There's lots of classic studio gear available this way.

[Mr. Bill]

On some material I could see the image shift being a problem. I record mostly rock music though, so the center elements are so strong that no noticable shift usually occurs. I prefer running most of my bus compression in a "dual mono" fashion as I like the way it sound better. Interestingly enough, I believe Chris prefers them the same way as he even un-links the focusrite red compressor he uses a lot. (no idea how he gets those knobs to match, must be a pita)

I do get how they usually will take a summed signal or at least a DC voltage representing the greater of 2 peaks like an absolute value for stereo linking. Summing before DC though seems like it could cause phasing interference in the detector. As in if you fed it 2 out of phase 1k sine waves panned hard L/R, there would be no compression no matter how much level was introduced.

[Enzo]

And the sine waves would trigger no compression anyway. It is changes in level that trigger compression, not the signal itself. In other words, a 1kHz sine wave would not be compressing and releasing 1000 times a second.

[Mr. Bill]

Pardon my crude drawing here, but for the sake of making sure I understand the basic function, here it is.


So in the scenario of a shared detector being fed from sum a, two 1k out of phase sine waves would show up as 0V at the input of the detector and trigger no compression, no matter how loud their amplitude. However, if each side had it's own detector that was then summed at sum b, then fed to each channel's vca, would cause the appropriate compression because the dc values would not cancel themselves. Is this not correct?

Note: I only drew it as one channel with it understood to be a stereo unit, hence the summed control voltage.

[Enzo]

Two out of phase 1kHz sines mixed together indeed comes out as zero. But then your pure stereo signal also cancels so if you had each coming out a speaker, you'd hear nothing in the center. The sides would cancel and you'd only hear a tone if you leaned towards one of the speakers.

So we are setting up a hypothetical that doesn't apply to a practical circuit. So if it has value in improving understanding so be it, but it won't be a problem in use.

[Leo_Gnardo]

So we are setting up a hypothetical that doesn't apply to a practical circuit. So if it has value in improving understanding so be it, but it won't be a problem in use.

True 'nuff. What the VCA responds to is the envelope of the signal the compressor/limiter is processing, not the signal itself.

Of course some other signal entirely can be applied to the return jack of the sidechain for special effects like ducking. An odd one I saw Carlos Alomar using was driving the sidechain return jack with the output of a drum machine, where he used a snare drum beat to gate his guitar sound. Bizarre, but that's what it takes when you're recording with the likes of David Bowie.

[Mr. Bill]

Okay, I think my original question was somewhat nonsensical because I wasn't quite grasping how these VCA circuits work. I apologize for that. The topic doesn't seem to have a lot of readily available literature, so I've tried to break it down section by section to grasp what's happening a little better. If you don't mind, allow me to check my current and evolved understanding of the circuit above.

1. Starting with the signal path, we have a differential amplifier set to output unity gain by making up the 6db difference in voltage drop from the divider with 6db of loop gain to drive the signal path VCA (with current?). The VCA modulates the signal based on it's control voltage input as a feed forward device and outputs this to a pair of op amps that re-balance the signal, again at unity gain.

2. The sidechain taken before the VCA is fed directly to another VCA with the same chip. This VCA is set up in a feedback configuration and receives it's control voltage from it's own loop. The output of this loop is fed to another pair of op amps which look like two full wave rectifiers, with one feeding the ratio circuit (no clue in the slightest how this part works at the moment) and the other feeding the time constants.

3. Again, I'm not grasping how the ratio part of the circuit works, but the attack seems to be a simple resistance to the rectified signal. I am honestly uncertain of how this resistance would delay the onset of compression, but I'm still learning. The release makes more sense to me. It appears that the time it takes the bypass capacitors to charge and release would determine that portion of the time constant.

4. This altered and rectified signal gets fed back to the control voltage of the detector circuit in an open loop and this loop is tapped and sent to the signal path VCA to modulate the actual audio signal based on this detection.

Am I close?

Just nix the original question, because I was pretty confused at that point. I felt all your concerned looks loud and clear though.