VCA compressor sidechain electronics in class a

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

My3gger

Well-known member
Joined
Nov 2, 2012
Messages
568
Location
EU
Hi,

i try to understand what is advantage of having VCA compressor in class a, especially sidechain electronics. It is about Focusrite Red 3, some of their statements:

The Focusrite Red 3 is a Class A VCA-based dual mono/stereo compressor/limiter.
The side-chain electronics are class A, ensuring superb transient response.
http://vintageking.com/focusrite-red-3-compressorlimiter-mh10505-used

As i understand it, signal after sidechain VCA is rectified into dc, timed and then returned to main VCA. Of course there is more to it, don't have scehmatics to see how it is done, so can't assume it is similar to SSL 4k or THAT papers.
How can class a electronics in sidechain control do any good if it is dc and not audio ac signal? I mean dc is dc in this case, can't see any advantage in using class a op-amps if this is truth in Red 3, only complications and possible problems with reliability in the long run. I won't start on other class a claims, it just makes no sense to me knowing newer Focusrite a bit.
 
  For me sounds as a marketing thing more than technical.

  That said, the rectified signal in the side-chain is hardly DC, it changes with the signal and depending on the timings it can be very fast. Then having a faster side-chain helps to follow that fast envelope better. Arguing class A is faster and you need one to do so, I don't know. Also, class A advantages are in the x-over region, or the lack of. Working with unipolar signals, are always on the same side of the zero, so it doesn't make much sense. Still, changes in current sign is what makes X-over in a class B, so still possible to have that in a rectified signal.

JS
 
My3gger said:
Hi,

i try to understand what is advantage of having VCA compressor in class a, especially sidechain electronics. It is about Focusrite Red 3, some of their statements:

The Focusrite Red 3 is a Class A VCA-based dual mono/stereo compressor/limiter.
class A is a specific topology of VCA, archaic now (IMO) with advent of reasonably priced and high performance modern generation IC VCAs from THAT.

Class A VCAs just like Class A signal handling circuitry are biased on with roughly 50% peak current at idle.  For a VCA this results in a much higher noise floor than class B VCAs where the operating current varies with signal. (FWIW in 1989 I got a patent on a variant class A VCA that didn't need trims, later after dbx/THAT started selling their VCAs to other manufacturers I abandoned my VCA that wasn't as quiet as theirs.)
The side-chain electronics are class A, ensuring superb transient response.
Cough...  try not to step in the cow patties. Speed in a side chain just causes more short term waveform distortion and is intentionally slowed by a selected or variable time constant (aka attack time). 
http://vintageking.com/focusrite-red-3-compressorlimiter-mh10505-used

As i understand it, signal after sidechain VCA is rectified into dc, timed and then returned to main VCA. Of course there is more to it, don't have scehmatics to see how it is done, so can't assume it is similar to SSL 4k or THAT papers.
How can class a electronics in sidechain control do any good if it is dc and not audio ac signal? I mean dc is dc in this case, can't see any advantage in using class a op-amps if this is truth in Red 3, only complications and possible problems with reliability in the long run. I won't start on other class a claims, it just makes no sense to me knowing newer Focusrite a bit.
The function of the side chain is to rectify AC (i.e. convert it to DC), then generate DC voltages proportionate to signal level, compare that DC to a DC threshold and use the result to manipulate gain. Class A in the side chain is meaningless (IMO).

JR
 
JohnRoberts said:
class A is a specific topology of VCA, archaic now (IMO) with advent of reasonably priced and high performance modern generation IC VCAs from THAT.

Class A VCAs just like Class A signal handling circuitry are biased on with roughly 50% peak current at idle.  For a VCA this results in a much higher noise floor than class B VCAs where the operating current varies with signal. (FWIW in 1989 I got a patent on a variant class A VCA that didn't need trims, later after dbx/THAT started selling their VCAs to other manufacturers I abandoned my VCA that wasn't as quiet as theirs.)

What exactly is a Class A vs. a Class B as it applies to a VCA? I'm thinking of an OTA with a LTP input such as the CA3080/CA3280 in which it appears everything operates in Class A. Would Class B be push-pull vacuum tubes, with the varying DC grid bias giving variable mu? It seems this would vary from Class A to Class AB or so. Is there a single-ended design (other than opto-LDR aka "vactrol,") I'm not thinking of?

As far as the use of the term "Class A," there's the electrical usage of an active device that conducts for 100 percent of the input waveform cycle, then there's the more general usage/implication of "best in class." I see way too much audio usage that confuses these meanings, whether intentional or not.
 
> What exactly is a Class A vs. a Class B as it applies to a VCA?

There was a many-issue squabble in RE&P some decades back which covered (did not really explain) how the (more complex than '3080) VCAs could run in A or (A)B, with different results.

IMHO, if you need more than '3080 performance, just get the THAT chips and let THAT worry about the bias.
 
benb said:
What exactly is a Class A vs. a Class B as it applies to a VCA? I'm thinking of an OTA with a LTP input such as the CA3080/CA3280 in which it appears everything operates in Class A.
An OTA is an operational transconductance amplifier, not a VCA (voltage controlled amplifier). As PRR noted back in the day there were VCA wars between competing approaches with several (most?) of the significant players long gone. DBX/THAT won the long war by surviving longer than their competitors, and their latest generation integrated circuit VCA delivers remarkable performance. Gary Hebert the lead VCA engineer for THAT published an AES paper on that new VCA, while it may not get into the nitty gritty of how VCAs work since that is relatively old technology today, while the incremental improvements were news.

If you understand how OTAs work, I will use them to describe a basic class A VCA (while there are probably multiple class A variants). Imagine two OTA amplifiers with their + and - input terminals connected in parallel. The first OTA , operating at a fixed nominal current is operated inside a feedback loop to linearize the input terminal voltage. The second OTA delivers a variable gain based on varying the operating current. There will be a first order cancellation of input stage distortion while the noise floor will be dominated by the first stage.  (My class A VCA patent was simpler using two LTPs cascaded  US04818951 1989).
Would Class B be push-pull vacuum tubes, with the varying DC grid bias giving variable mu? It seems this would vary from Class A to Class AB or so. Is there a single-ended design (other than opto-LDR aka "vactrol,") I'm not thinking of?
For a class B VCA visualize both npn and pnp LTP, biased on through resistors, to + and - rails. connected together. The collectors of the first top PNP and first bottom NPN connect together and return to the op amp - input. NF causes the VCA to swing and deliver a current equal and opposing the input resistor current. The second transistors from each LTP have their collectors connected together and that collector current output is the VCA output. This VCA output is fed to another inverting op amp with feedback resistor to convert back to a voltage.  As long as all the bases are at the same voltage, nominally 0V, the input current will match the output current. Introducing mV level offsets between the device bases will steer more or less current toward or away from first feedback path, effective boosting or cutting the output current. The base current control law is a very convenient  XmV/dB of gain change.

This is a simplified description and there can be class A operating currents inside class B VCAs that likewise affect noise floor and linearity. VCA technology is rather mature and there is little (no) reason to look beyond the latest generation THAT parts for transparent gain control IMO. 
As far as the use of the term "Class A," there's the electrical usage of an active device that conducts for 100 percent of the input waveform cycle, then there's the more general usage/implication of "best in class." I see way too much audio usage that confuses these meanings, whether intentional or not.
I have written about this before, here is a very old article I wrote while at Peavey.  https://peavey.com/support/technotes/poweramps/classact.cfm  judging from the amplifier models this was written sometime in the 1990s)

Class A generally describes a topology using only one polarity of active devices that either only source or only sink current. Most early active circuitry was class A by practical necessity.  Far from higher quality, cheap equipment used class A to reduce cost. The audiophools have spun class A as a premium approach because the power devices never shut completely off.  Circuit designers have managed to deal with that inconvenience delivering vanishingly low distortion in not class A circuits .

JR
 
This is slightly off topic, but I think that fascination with class A single ended circuitry has more to do with powering a circuit with only one power supply, and not what class A does to the amp's transfer function.

With a modern bipolar supply op amp circuit, a problem can result from the power supply currents that feed the op amp as it delivers current to the load. With a DC centered waveform, the transient power supply current sent to an op amp will be sourced by the op amp from its two power supplies as two 'rectified' components, each composed of the positive or negative going portions of the signal current sent by the op amp to the load. The problem is that these currents are split into two halves, and if power supply bypass caps couple these currents into signal ground, voltages could develop in the ground traces that can inject this 'rectified' signal into various portions of the circuit, injecting distortion into the circuit.

With a single supply amplifier, the current sent to the load is sourced entirely from one supply, so it is never broken into two 'rectified' pieces that might travel separate paths through the circuit's ground system before they combine with the return current from the load.

Essentially, it's easier to make a sloppy PCB or layout with a single supply amplifier, and have things work a lot better than if you do a sloppy layout with a bipolar supply circuit, where these rectified signal currents can flow through the wrong places and create voltage drops that induce high order distortion that is far larger than the basic linearity of a quality op amp.

A way to combat this problem is to try to get these two 'rectified' signal currents from one op amp to combine at a small bit of foil on the PCB. This can be as simple as laying out the ground pads for the supply bypass caps right next to each other on the PCB, and having nothing important reference ground near that part of the ground foil. This can also be done by using local regulators and only small HF bypasses to keep the amplifiers stable, so that the supply currents largely sum into the regulators, away from the possibly sensitive ground nodes of the amplifier.

Back to VCAs and compressors, these control signals and voltages do need to get isolated from the actual signals, since they are by definition highly nonlinear. However, "class A" does not seem to be necessary, nor is it sufficient. Isolated ground schemes, and careful attention to where these nasty currents are sourced and where they end up is what needs to be addressed, and "class A" is just some cutsheet drivel.
 
Monte McGuire said:
This is slightly off topic, but I think that fascination with class A single ended circuitry has more to do with powering a circuit with only one power supply, and not what class A does to the amp's transfer function.
While you are entitled to your own opinion, Class A power amps can be made with single or split supplies. If the output is DC coupled the class A amp will need + and - voltage PS rails. I have also seen modest sized class AB made with single rail (but not many).

Active devices can not turn on or off arbitrarily fast so a significant benefit to class A the active power device never turns off. In fact it only turns completely off at saturation or clipping. The device current is strong (1/2 full scale) and device response time fast, at the sensitive region where audio swing is smallest.  A great deal of design effort goes into minimizing this turn-on/off time issue in class B or AB amplifiers that must pass current delivery back and forth between devices at this most sensitive small signal region.
With a modern bipolar supply op amp circuit, a problem can result from the power supply currents that feed the op amp as it delivers current to the load. With a DC centered waveform, the transient power supply current sent to an op amp will be sourced by the op amp from its two power supplies as two 'rectified' components, each composed of the positive or negative going portions of the signal current sent by the op amp to the load. The problem is that these currents are split into two halves, and if power supply bypass caps couple these currents into signal ground, voltages could develop in the ground traces that can inject this 'rectified' signal into various portions of the circuit, injecting distortion into the circuit.
Of course PS ground and signal low or 0V are two different things. Besides corrupting grounds Cherry wrote an AES paper, actually wrote more than one, but this one was  about the effect of significant rail currents in power amplifier layouts. IIRC this paper was probably published back in the 80s.
With a single supply amplifier, the current sent to the load is sourced entirely from one supply, so it is never broken into two 'rectified' pieces that might travel separate paths through the circuit's ground system before they combine with the return current from the load.
The output current received by the load is very similar for different classes of amplification. In higher power amplifier design managing all currents are important to deliver good distortion specifications.
Essentially, it's easier to make a sloppy PCB or layout with a single supply amplifier, and have things work a lot better than if you do a sloppy layout with a bipolar supply circuit, where these rectified signal currents can flow through the wrong places and create voltage drops that induce high order distortion that is far larger than the basic linearity of a quality op amp.

A way to combat this problem is to try to get these two 'rectified' signal currents from one op amp to combine at a small bit of foil on the PCB. This can be as simple as laying out the ground pads for the supply bypass caps right next to each other on the PCB, and having nothing important reference ground near that part of the ground foil. This can also be done by using local regulators and only small HF bypasses to keep the amplifiers stable, so that the supply currents largely sum into the regulators, away from the possibly sensitive ground nodes of the amplifier.
Signal low or 0V is a true input just like signal hot and will affect the output.
Back to VCAs and compressors, these control signals and voltages do need to get isolated from the actual signals, since they are by definition highly nonlinear.
The control signals directly modulate the primary audio path gain, so artifacts in the control chain will appear superimposed on the audio.
However, "class A" does not seem to be necessary, nor is it sufficient. Isolated ground schemes, and careful attention to where these nasty currents are sourced and where they end up is what needs to be addressed, and "class A" is just some cutsheet drivel.
We agree touting class A for side chains are a nothing burger.

JR
 
I suspect that the Red3 has a class A differential amplifier where the direct signal enters the inverting input, and the non-inverting input receives the VCA output. That would make sense to me, and is a very clean way to archieve gain reduction.
As Joaquins said, a sidechain signal has a CV only, and rarely crosses zero.
To archieve good transient response, the envelope detector diodes should be very precise and fast, but sounds like marketing speech to me.
 
Back
Top