Feedforward / look-ahead compressor

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
hazmo
Posts: 19
Joined: Wed Dec 23, 2009 7:31 am
Location: Germany

Feedforward / look-ahead compressor

Post by hazmo »

Hi gentlemen,

could somebody point out some example circuits using a delay in the signal path? It's only of academic interest to me, but I'd love to see what different approaches were taken. Allpass filter, BBD (?), ...


Volker
hazmo
Posts: 19
Joined: Wed Dec 23, 2009 7:31 am
Location: Germany

Re: Feedforward / look-ahead compressor

Post by hazmo »

Found this thread where several pieces of equipment are mentioned. No gory details though. I'll start by looking for the Neumann, might be the easiest to come by for me.
hazmo
Posts: 19
Joined: Wed Dec 23, 2009 7:31 am
Location: Germany

Re: Feedforward / look-ahead compressor

Post by hazmo »

Thanks, yes, that should be easy enough. But as I said, it's not the actual sound processing I'm after.
I would be mainly interested in seeing how the delay is achieved in analog hardware.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Feedforward / look-ahead compressor

Post by JR. »

I don't have any first hand experience or schematics, but I am pretty sure you won't find many (any?) examples of BBD delays used in front of a compressor, since the dynamic range of BBD paths is poor to start with, and compressing that would make it even less. In fact it is routine to compand (compress before and expand after) BBD paths. I do have experience with BBDs

All pass filters could deliver reasonable dynamic range, while the amount of delay is modest. Perhaps as a very short look ahead for a broadcast limiter. Typical all pass delay was used by Rane in early loudspeaker crossovers to provide driver alignment delay, while the task was made easier by the limited bandpasses involved. Wide band all pass delay typically involves combining several filters.

This technology, while such an obvious application for delay has not been widely considered until we reached a critical mass of signal path already in the digital domain. I recall an early stand alone digital compressor, with delay as a feature but this was pretty much dismissed by sound quality conscious consumers since compression just reduces the already marginal dynamic range (early days).

Rane has some commercial success with a noise gate that has a digital pre-delay built in. A couple mSec give or take path delay is not a huge issue for live sound as long as signals are kept segregated and not mixed with non-delayed versions of themselves.

JR
Cancel the "cancel culture", do not support mob hatred.
hazmo
Posts: 19
Joined: Wed Dec 23, 2009 7:31 am
Location: Germany

Re: Feedforward / look-ahead compressor

Post by hazmo »

Thanks for the input!

I got hold of the service manual of the EMT 266 Transient Limiter. Here are the block diagram and part of the schematic. Two blocks of four 2-pole allpass filters are utilized for an overall delay of 300µs. Also note the fairly elaborate lowpass in front, probably in order to get a passband where the delay will be homogeneous.


Volker
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Feedforward / look-ahead compressor

Post by JR. »

Yes, all pass filters "was" the only way to get reasonable fidelity short delays, but there is no free lunch so there are phase response considerations. Curiously all pass filters are still used in modern SOTA digital crossovers but not for delay, which is trivial in the digital domain, but to tweak phase response of individual drivers so they combine better with the other drivers outputs.

In the context of limiters, it becomes a tradeoff between the amplitude artifacts of very fast attack times, vs phase errors associated with all pass filters. My original suggestion that this may be used in broadcast because they have littler tolerance of overload, and perhaps a relaxed appreciation for fidelity.

Note: I am not an all pass expert so it may be possible with very complex alignments to approach pure delay over the full audio passband, but I expect that will require several very precise networks summed together, ironically something that analog circuitry also lags behind digital in (precision).

Depending on the application it seems operating in the digital domain where it should be relatively trivial to give the side chain unlimited lead time without any fidelity compromise seems one obvious solution.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Feedforward / look-ahead compressor

Post by JR. »

raf wrote:Except that DSP/plug in compressors don't sound as good as analog hardware...
Since I don't see a smiley face, I'll bite. Why, can't be done, or just hasn't been done yet?
=====

I have been tempted to make one, mainly for the added flexibility and control of the side chain. In my "shiny hammer- universal comp" thread I have shared my opinions at length.

I have had good luck with using digital pots for a nominally analog path with digital control, but I'd be tempted to go all digital for finer gain control resolution.

The primary limitation from where I sit is the control interface. If I want to cover every compressor ever made, as the starting point to improve upon, I need a sh__ load of knobs and switches, that can only be done practically with a soft computer interface.

But I don't think the world needs yet another comp, even if it promised to be their last one ever.... 8-)

JR

PS: I almost suggested rolling a simple delta mod delay line for the OP.. delta mod is almost analog and could be good fidelity for short delay using high clock rate. Don't know if we can even buy 1 bit wide digital memory chips these days.
Cancel the "cancel culture", do not support mob hatred.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Feedforward / look-ahead compressor

Post by JR. »

Are you suggesting that digital conversion technology plateaued at some inferior level short of analog performance?
=====
I am not a student of the current dynamics market. I haven't designed a compressor myself since the '80s, and one done by an engineer working for me after that was probably lifted from THAT app notes verbatim, I didn't even review it closely.
======
While i have not done a commercial design for decades (and I am not enthusiastic about doing one now), I am of the opinion that a properly executed digital path (or plug-in) could mimic any analog compressor technology. I'm not sure we want to copy the flaws of old VCAs, while the flaws of some classic legacy gain cells are surely part of their signature sonic attraction.

I wish you continued success with you compressor, but i attribute that success to the specific combination of components and design decisions you made. Not some superiority of analog over digital technology. In fact I am of the opinion that digital "could" deliver some remarkable behind the scenes gain manipulations and smart non-linear situational decision making, only limited by the programmers willingness to experiment and invent.

I have one foot in both domains and digital is so much more flexible than analog it is no comparison (IMO). Now digital keeps getting cheaper and cheaper. At some point i expect the market for analog VCAs to decline, but they will linger on for some time. I am working on a design right now where I am using DPOTs in place of VCA for simple level/gain manipulation, but this is not a compressor (while I could add one into it).

JR
Cancel the "cancel culture", do not support mob hatred.
juniorhifikit
Posts: 139
Joined: Sun Oct 04, 2009 8:37 am
Location: Oakland/Paris

Re: Feedforward / look-ahead compressor

Post by juniorhifikit »

JR. wrote:While i have not done a commercial design for decades (and I am not enthusiastic about doing one now), I am of the opinion that a properly executed digital path (or plug-in) could mimic any analog compressor technology. I'm not sure we want to copy the flaws of old VCAs, while the flaws of some classic legacy gain cells are surely part of their signature sonic attraction.
I can't add much to this discussion, except that I think digital can, and someday will mimic any analog compressor. The designers just need to figure out what to model, and then have enough processors cycles to do it. Several plugin designers are personal friends of mine, and we often discuss what's good and bad about their designs. Often they only model "part" of a device - the part that they feel is most important. Some of them are not life-long mixers, and so perhaps don't fully understand what it is that mixers love about each prized piece of analog gear. For example, one company who modeled the SSL stereo bus compressor decided to only model the "knee" and gain reduction characteristics, and not bother with the gain staging and input/output amps. In polite use, the plugin mimics it's analog counterpart very well. But we're making art, and polite is not often appropriate. It's the creative abuse of analog gear and the way they fold that makes them prized -

It's their flaws that we love.

So I agree with both of you :)
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: Feedforward / look-ahead compressor

Post by JR. »

I don't want to hijack this thread (oops too late).

As I have written elsewhere one conflict or internal argument I see about dynamics design is arbitrarily making decisions for the end user, or giving them full control. One can imagine tens of parameters that can affect the "sound" of gain manipulations. Consumers will never tolerate that many decisions, and I don't blame them. The approach I wanted to pursue, back before I knew how to do it, was to call up presets that approximated popular comps as a starting point and then let the operator tweak from that presumably good starting point. I was going to save the semi-custom program data to the recording tape leader (yes I was thinking about this that long ago.).

This is still probably TMI for most users. A plug-in based on my theme would require an interface with tens of twiddles.... I can image a few extreme users who might embrace this... If your plug-in buddy is interested, either start a new thread or go to my shiny hammer thread...

To bring this almost back on topic the plug-in could use look-ahead... unfortunately not analog delay...

JR

PS: I also had a none-of-the-above compression approach that I actually built into a commercial unit back in the early '80s (LOFT) but the company went belly up before then comp got a fair test from the market.
Cancel the "cancel culture", do not support mob hatred.
Post Reply