Compressor Attack Release Time Measurment

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

bluebird

Well-known member
Joined
Jun 11, 2004
Messages
1,070
Location
Los Angeles
Does anyone have a test already made up for an Audio Precision One that can give me an accurate reading for attack and release for a compressor? Guessing it would be a burst and an amplitude/time measurement. I could wrestle around with it and figure something out eventually but if anyone has gone down this road already, I'd appreciate some advice.

Thanks!
 
After decades of designing sundry dynamics processors I am not aware of a single common standard for specifying Attack/Release times. I always kind of winged it.  Especially when sidechain had multiple moving parts involved.

Risetime of an RC is often considered the time it takes to get from 10% to 90% of final voltage, so for a side chain using simple RC that might work.

[edit] another significant ratio is 1xRC that for simple exponential decay falls some 69.3% or roughly -10dB in one time constant.  While looking at it from the other direction 69% is roughly - 3dB from endpoint (100%). [/edit]

JR

 
Thanks John,

Its a Fairchild style side chain with a transformer and diode bridge. I guess I would have to factor in the transformer too.  I can use a DAW (pro tools) to make bursts and record the resultant wave form as well.

edit: I'm sure I could actually spend some time searching all the fairchild posts and find the exact formula to use....
 
I'm curious too what people use for the actual definition for attack and release.

If you have the capability I would use a pulse for an input with a fast rising and fast falling edge.  Then look on the scope to see the reaction time. 
 
I've used cubase for checking out the envelope from my builds of a few of the classics ...

Basic stuff, just pre- preparing of a few tracks of sine waves of various lengths and ampliudes (and freqs) .. then playing back thru the unit, with periods of silence interspersed ..

I kept manual notes (on the computer) with screenshots for a few different scenarios  etc.

...

Its certainly interesting and useful to do so ...  I am still to figure out how close to 'real' my own fc660 type build gets.

What I see from the cubase files is that it seems to massively overshoot the program material, then release slightly then massively overshoot the next threshold transient etc.

It's kind of maddening really ..  at least for me. I'm certainly no expert on measurement of dynamic limiter characteristics.

I do have both 'ac threshold' and 'dc threshold' controls on my interpretation, and it is possible I simply haven't got it set up right.

It's definately a journey as to using and understanding ,even crude stylings of the classic.

 
There is simply no common-ground to how manufacturers specify attack/release timing. I was once told that our national radio DR (who are responsible for the rather high quality of Traditional Danish Audio Stuff) once made a whitepaper addressing this problem, but I never managed to dig it up.

If you decide to try going the measurement way, please keep in mind that timing is co-affected by the ratio setting in feedback architecture compression - which makes the number you come up with even less informative ;D

On the other hand, if it's just for relative work, comparison and/or go/nogo in calibration, then I'd go with the above-mentioned 10%>90% transition timing. In a fairchild-style compressor, you'll still have to keep ratio the same for comparable results..

Jakob E.
 
bluebird said:
Does anyone have a test already made up for an Audio Precision One that can give me an accurate reading for attack and release for a compressor? Guessing it would be a burst and an amplitude/time measurement. I could wrestle around with it and figure something out eventually but if anyone has gone down this road already, I'd appreciate some advice.

Thanks!
PM me and I can walk you through the process.
Duke
 
Over the decades I killed many brain cells designing sundry side chains for dynamics processing. Probably a little more work on companding noise reductions than single ended dynamics but it is all pursuing the same end point... smooth gain changes in response to changing levels, without audible artifacts.

In many cases I incorporated adaptive level dependant, nonlinear attack and release mechanisms. Human hearing perceives large and small level changes differently, not to mention so time interval phenomenon, that can be incorporated into side chains to make them more transparent to human listeners.

The bad news is there is no way a single number could characterize attack or release for all, let alone most, such cases using nonlinear sidechain tricks.  I put numbers on some compressor controls because the market expected numbers, but when I could I just labelled controls with fast and slow ends.

JR

PS: There is merit is quantifying some side chains for comparison between builds, but be very certain to use identical stimulus signals for such measurements.  I built a custom tone burst box with zero crossing synchronization and multiple adjustments to properly exercise side chains when designing circuits using them.
 
Here's the standard electrical RC-time-constant thing:

https://en.wikipedia.org/wiki/RC_time_constant

And as said, compressors (or "good sounding" compressors) don't follow the ideal exponential decay of a capacitor discharging through a resistor, but the time for a 63 percent change can still be a decent measure that you put on the front panel for different attack/release settings.

It's common to cascade several compressors, often with different attack timings as well as different compression ratios, and this could conceivably be done in the sidechain of a single compressor. Optical ("Vactrol"-based) compressors are inherently nonlinear in several ways, and thus (even) harder to measure  - several short repeated attacks and releases have different time constants than the first signal after a period of silence.

But you can measure the time constant and end up with a "number" (with a larger number meaning a longer attack or release) even though this number may be only grossly comparable between different compressors.
 

Latest posts

Back
Top