## A Discussion About True Power Summing for Stereo Compressors

### Re: A Discussion About True Power Summing for Stereo Compressors

Hey, another application might be for a feedback compressor with variable compression ratio; the stand-alone timing constant needs to be multipied by the compression ratio for a consistent rms timing value at all ratios, yes? That is if you could scale it correctly after ganging it to the ratio pot...

### Re: A Discussion About True Power Summing for Stereo Compressors

Huh?

Visit https://circularscience.com to hear what properly "cleared" drums sound like.

### Re: A Discussion About True Power Summing for Stereo Compressors

Hmm, I guess I'm not understanding the rms thing yet LOL - seems to me that in feedback mode, the timing cap value would be dependent on the compression ratio to keep the same timing value (?)...I need to learn me some math...JR. wrote:Huh?

### Re: A Discussion About True Power Summing for Stereo Compressors

Let me think about that... AFAIK those dots are not normally connected but who knows.?

JR

=======

OK, giving this a little time to ferment... I don't see any strong association between compression ratio and attack/release rates.

Attack and release speed affects the amplitude of the AC component of control voltage. Faster side chain means higher amplitude CV modulation. Compression ratio influences how much of this side chain modulation makes it's way into the the VCA control voltage. Indeed the higher compression ratio causes a higher percentage of this side chain modulation. Does this mean we want to slow it down for high compression ratios?

That is counter intuitive. It seems slower response for higher compression ratios would lead to larger or more time in error. i.e. more overshoot from transient steps.

I don't have a definitive proof, but my gut feeling is that attack/release time is far more dependent on the program material than the compression ratio. Any adjustment to normalize the modulation would be less than 2:1 and in the wrong direction (IMO).

This is again something folks can experiment with and empirically determine if there is any first order speed tweak desirable wrt compression ratios. My gut feeling is this is not a rich vein to mine, but don't let me stop anybody.. just do it and let us know what you think.

JR

JR

=======

OK, giving this a little time to ferment... I don't see any strong association between compression ratio and attack/release rates.

Attack and release speed affects the amplitude of the AC component of control voltage. Faster side chain means higher amplitude CV modulation. Compression ratio influences how much of this side chain modulation makes it's way into the the VCA control voltage. Indeed the higher compression ratio causes a higher percentage of this side chain modulation. Does this mean we want to slow it down for high compression ratios?

That is counter intuitive. It seems slower response for higher compression ratios would lead to larger or more time in error. i.e. more overshoot from transient steps.

I don't have a definitive proof, but my gut feeling is that attack/release time is far more dependent on the program material than the compression ratio. Any adjustment to normalize the modulation would be less than 2:1 and in the wrong direction (IMO).

This is again something folks can experiment with and empirically determine if there is any first order speed tweak desirable wrt compression ratios. My gut feeling is this is not a rich vein to mine, but don't let me stop anybody.. just do it and let us know what you think.

JR

Visit https://circularscience.com to hear what properly "cleared" drums sound like.

### Re: A Discussion About True Power Summing for Stereo Compressors

A lot of gates / expanders allow for linear or log (or something in between) AR times. Anyone try that with a comp?

Just a thought.......

Bruno2000

Just a thought.......

Bruno2000

The large print giveth, and the small print taketh away - Tom Waits

### Re: A Discussion About True Power Summing for Stereo Compressors

With the DBX detector IC the integration is performed in the log domain at the x^2 stage of the conversion. IIRC the cap is driven by a pure current so attacks with the signal squared, and decay's linearly, but since the control voltage is in the log domain the linear decay appears exponential to the outside world.

Note: per the THAT corp data sheet attack is (1-e^-t), i.e. faster than linear which is >> faster than simple exponential when translated back to linear world. http://www.thatcorp.com/datashts/2252data.pdf page 4-5

Regarding this comp I'd be inclined to use existing CV approach but with perhaps a general speed switch/pot (or not).

For dynamics in general, there are many many variants that can be and have been explored. Different flavors for different stews.

JR

Note: per the THAT corp data sheet attack is (1-e^-t), i.e. faster than linear which is >> faster than simple exponential when translated back to linear world. http://www.thatcorp.com/datashts/2252data.pdf page 4-5

Regarding this comp I'd be inclined to use existing CV approach but with perhaps a general speed switch/pot (or not).

For dynamics in general, there are many many variants that can be and have been explored. Different flavors for different stews.

JR

Visit https://circularscience.com to hear what properly "cleared" drums sound like.

### Re: A Discussion About True Power Summing for Stereo Compressors

O.K. Sensei, one last ponderable and I'll stop bugging you about it (probably... LOL):

That all makes sense to me when it's feedforward, but... when the compressor topology is feedback, does it really "see" that timing cap as being the same value as when the detector is feedforward?

That all makes sense to me when it's feedforward, but... when the compressor topology is feedback, does it really "see" that timing cap as being the same value as when the detector is feedforward?

### Re: A Discussion About True Power Summing for Stereo Compressors

Off the top of my head I suspect its pretty similar.

Without digging too deeply, for a similar compression ratio and result, and ASSuming a similar VCA topology, by extension the VCA control voltage will be similar. What is different is the side chain input. In feed forward it is the raw uncompressed input, in feedback it gets the compressed output. So the hotter input gets a smaller multiply factor than the cooler input . I suspect its a wash. In price sensitive low end designs the feedback topology wins because the side chain only has to operate on a reduced dynamic range.

There is also a variation between connecting the VCA in the input leg or feedback leg of an active gain stage. There are subtle differences here too where one is increasing conductance to lower gain, vs dropping conductance. Similar to the dynamic range considerations of the side chain the VCA must handle a wider dynamic range in the input leg vs. feedback leg.

Inexpensive companders for cheap tape NR always put both the gain element and side chain in the feedback side to take advantage of reduced dynaic range.

JR

BTW I'm speaking in general terms and broad strokes since I Haven't looked at the PICO schematic recently. (Maybe I should shut up at some point).

Without digging too deeply, for a similar compression ratio and result, and ASSuming a similar VCA topology, by extension the VCA control voltage will be similar. What is different is the side chain input. In feed forward it is the raw uncompressed input, in feedback it gets the compressed output. So the hotter input gets a smaller multiply factor than the cooler input . I suspect its a wash. In price sensitive low end designs the feedback topology wins because the side chain only has to operate on a reduced dynamic range.

There is also a variation between connecting the VCA in the input leg or feedback leg of an active gain stage. There are subtle differences here too where one is increasing conductance to lower gain, vs dropping conductance. Similar to the dynamic range considerations of the side chain the VCA must handle a wider dynamic range in the input leg vs. feedback leg.

Inexpensive companders for cheap tape NR always put both the gain element and side chain in the feedback side to take advantage of reduced dynaic range.

JR

BTW I'm speaking in general terms and broad strokes since I Haven't looked at the PICO schematic recently. (Maybe I should shut up at some point).

Last edited by JR. on Fri Mar 27, 2009 5:27 pm, edited 1 time in total.

Visit https://circularscience.com to hear what properly "cleared" drums sound like.

### Re: A Discussion About True Power Summing for Stereo Compressors

Ah, so I was comparing the JR var-cap to the non-linear cap circuit and stumbled upon where I must have gotten that timing cap/ratio idea:

http://www.thatcorp.com/datashts/600034-1.pdf

http://www.thatcorp.com/datashts/600034-1.pdf

*The above calculations assume a stand-alone RMS*

level-detector. When the detector is placed in a feedback

compressor topology, the effective time constant

that results is calculated by taking the level detector’s

stand-alone time constant and dividing it by the compression

ratio. Therefore, if we plan to operate with

a compression ratio of, say, 20:1, we will need to increase

the timing capacitor by a factor of 20. So, for

our design the timing capacitor, C2, becomes 220 uF,

the nearest standard value.level-detector. When the detector is placed in a feedback

compressor topology, the effective time constant

that results is calculated by taking the level detector’s

stand-alone time constant and dividing it by the compression

ratio. Therefore, if we plan to operate with

a compression ratio of, say, 20:1, we will need to increase

the timing capacitor by a factor of 20. So, for

our design the timing capacitor, C2, becomes 220 uF,

the nearest standard value.

- mediatechnology
**Posts:**3849**Joined:**Sat Aug 11, 2007 2:34 pm**Location:**Oak Cliff, Texas-
**Contact:**

### Re: A Discussion About True Power Summing for Stereo Compressors

Here's an application note "The Mathematics of Log-Based Dynamics Processors" that may be useful to our readers.

It only touches upon the time constant differences between feed-back and feed-forward and deals primarily with steady-state threshold and ratio.

http://www.thatcorp.com/datashts/an101a.pdf

It only touches upon the time constant differences between feed-back and feed-forward and deals primarily with steady-state threshold and ratio.

http://www.thatcorp.com/datashts/an101a.pdf

**https://ka-electronics.com**

**Meme World**https://www.memeworld.com/

**Project Veritas: Google "Machine Learning Fairness" Whistleblower Goes Public, says: "burden lifted off of my soul."**https://www.youtube.com/watch?v=g1VeElB ... e=youtu.be

**If banned by YouTube:**https://www.bitchute.com/video/g1VeElBAeas/