Page 1 of 6

new shiny hammer- or Universal comp

Posted: Mon Jun 30, 2008 2:34 pm
by JR.
I realize I'm probably starting to sound like the boy with a new hammer who sees himself surrounded by nails.

I came to a realization and defined up a product back in the early '80s that I am only now able to execute. I wanted to make a soft or programmable compressor with literally everything adjustable or selectable. While this much flexibility is barely handled well by comp designers, the customer would surely be overwhelmed. So the product would come loaded with sundry presets that mimic popular designs. The user could call up a favorite industry standard then go in and tweak that favorite to be "just like the XYZ but with rms instead of peak, or faster/slower whatever". The icing on my decades old cake, was to be a FSK audio save of the comp settings that could be printed to a recording tape leader like a slate tone, so in subsequent mixdown sessions you could recall the exact compression variant you used.

Now with the benefit of decades of hindsight, and the evolution of recording, makes the FSK save seem a bit archaic, but I think there is still merit in making a soft/flexible comp platform that can be as easy to adjust or hard as the user wants .

While I see Cool Audio is selling a (cyrstal?) knock off 24B CODEC for $.90 in high quantity, and DSP can be had for a few bux, I'm inclined to use a real analog VCA but do the gain control crunching inside a cheap digital micro. These cheap micros have a bunch of A/D inputs so there's no problem accommodating feed back, feed forward, and even side chain inputs. Peak, RMS, whatever lin/log attack, release, hold, ripple filtering, soft knee, whatever... If you can imagine it, you can probably do it in the code.

A single dual VCA, and handful of opamps to provide bal in/out and buffer the insert would be modest in cost. A single <$2 micro could handle the crunching, we are back where chassis, PCB, buttons, leds, et al would be the expensive part. This really screams for an inexpensive USB port for a computer interface to accommodate managing all the variables for tweaking.

As I read sundry posts about different comp topologies or rectifier idiosyncrasies I can't help thinking that would be pretty easy in software. (of course easier to talk about than do).

Ironically this probably isn't viable as a commercial product since one could do almost the same thing with plug ins, if not now, probably some time soon. My time might be better spent learning how to write a plug in or a plug in generator. (different kind of hammer).

JR

Re: new shiny hammer- or Universal comp

Posted: Mon Jun 30, 2008 8:01 pm
by mediatechnology
Go for it!

Do you think you would need to be able to, in DSP, model the non-linearities of the various models even with them operating in bypass? What I'm trying to say is that the transformers, electronics etc. might be part of, or at least thought to be part of, the sonic signature.
this probably isn't viable as a commercial product since one could do almost the same thing with plug ins
What about latency in plug-ins?

Re: new shiny hammer- or Universal comp

Posted: Tue Jul 01, 2008 11:04 am
by JR.
I would not attempt to mimmic transformer or VCA distortion. In a compressor you have more than enough distortion from fast release times and unfiltered gain control modulations to listen to.

The significant sonic signatures of a compressor that I'm interested in, are fairly obvious. I would aim for a neutral analog path, and clean SOTA VCA.

I was thinking about this some last night after beer o'clock and this screams for a multi line LCD display to handle all the attributes.

I guess to make this friendly it needs a fair complement of controls, while I can imagine a lot of set it and forget it attributes.

Ideally operation should be as simple as, call up "easy over"... and either just go with the standard preset, or dig into the minutae and see if you prefer how it sounds with sundry tweaks. I suspect closer inspection might reveal flaws in some of the classics that can actually be improved upon, and alternately rough edges that are musically useful.

Having multiple layers of control voltage crunching allows simultaneous downward expansion noise gating, peak limiting, compression, even de-essing. Could even use peak detection for limiter, while ave/rms for comp. I designed a compressor back in the '80s (LOFT) that I really though was a better way to do it, but it never got a good shot in the marketplace as the company was fading at the time. My comp was basically full range with no threshold. Compression ratio compressed the whole shebang. This eliminated the need to deal with threshold adjustments, but resulted in an elevated noise floor. I built in a smart downward expander that did have a threshold and slope, and in this case, the threshold actually defined where it passed through unity gain again as the expander unwound the gain added by the compressor above threshold and reduced gain and noise below threshold at a rate defined by the slope. A peak limiter was also overlaid on top of the whole thing. IMO pretty logical.. set comp ratio to get the sound right, set the noise floor where you want and you're done. This was unconventional enough that it wasn't immediately understood or embraced.

I would also use my peak/VU meter to monitor output of compression. You should be able to literally see the result of reducing the crest factor in real time.

There are a few things that true digital domain processing could do that digital controlled analog can't. Dynamic EQ, or processing within specific frequency bands is one. Another is use of delay look ahead, RANE makes a very well respected noise gate using this capability. I think Valley audio did a digital domain comp years ago, but they were too far ahead of the market (perhaps still today) and never gained traction.

Latency exists in real life... the speed of sound in air is rather slow compared to electricity so small time delays are always with us. Of course it is an issue to be managed.

JR

Re: new shiny hammer- or Universal comp

Posted: Tue Jul 01, 2008 11:29 am
by JR.
raf wrote:John

My take is that seperates sound better than plug ins when it comes to comps and EQ, even if the code is similar. Seperate conversion?

Have you ever messed with a TC Electronic Finalizer? It is a dsp based multi band / Eq / normalizer etc. Good idea but I am sure it could be improved soundwise. TC is deathly afraid of distotion figures > .0000000000000001%. Ok, I was kidding but I have a feeling that is why I don't like to use the compressors, too sterile sounding. I still use it as a CR appliance, DA for Cds and ultra pure 24 bit test tones, phase meter, PPM meter, fader. Pretty handy. I'm glad I didn't dump it.

Now if someone could make a really good non sterile sounding dsp super comp, I would be very interested. Wouldn't be able to add much to the design discussion since this type of dsp is not my forte, but interesting nonetheless.

I'd say go for it!
rf
I never listened to the old Valley unit but knew the guy who designed or worked on it. He was a bass player (Jason).
-----
I am pretty ignorant about plug ins but the trend I see in big dog live SR, is using the same big digital consoles as in studios with the same plug ins and sundry tricks available. If the plug ins are somehow flawed maybe thats an area for research and opportunity.

I am just dipping my toe into DSP waters, my first effort will be pretty crude FFT analysis of drum resonances for tuning purposes.

I did spend some time bench testing and working on a friend's DSP platform. While the unit used a decent CODEC and DSP, I was never satisfied with the raw audio quality. Nothing obvious like frequency response or distortion, but something just wasn't right. Of course his product was doing some rather unconventional processing, that was likely the cause.

Making a simple digital compressor doesn't even need a DSP, it's just some number crunching and a multiply. Of course if you get fancy with filters you'll need the DSP trix.

Since cool audio is selling a 24B codec for $.90 it might be cheaper to make a digital comp... but I would hate to think about what kind of sound quality $.90 buys you.

JR

Re: new shiny hammer- or Universal comp

Posted: Wed Jul 02, 2008 12:36 pm
by JR.
After another session of soaking this idea in beer to see what floats up to surface, I'm coming to several realizations.

I can do far more tricks in the software than I can easily provide a user interface to in the hardware.

Several options are fermenting...
1) a master with tons of controls (LCD screen, sundry rotary encoders and tact switches) , and reduced control slaves that can be programmed by master. Slaves would be free standing and functional just not as flexible, without the master. Master and slave would talk over proprietary serial com using simple phono cable.

2) A reduced control slave-like only module but with a USB or appropriate computer interface, so PC or MAC can be used to vary the tens of parameters. I'm reluctant to use the simpler rs-232 since it's getting rare on modern computers

3) do the whole thing as a plug in...or plug in maker

Looks like I'm not finished learning how to do silly digital tricks since there's sundry pieces in there I haven't mastered yet.

Oh well, time to back burner until I pick up some more hammers.

JR

Re: new shiny hammer- or Universal comp

Posted: Wed Jul 02, 2008 1:38 pm
by mediatechnology
I like the "glass cockpit" approach with USB to the actual compressor engine. Those knobs switches etc. will be the bulk of the cost.

I'm always amused when I see the tear down cost breakdown of flat panel TVs. Usually the second highest cost, after display is plastic.

Re: new shiny hammer- or Universal comp

Posted: Wed Jul 02, 2008 2:33 pm
by JR.
Yes, an interesting dynamic is cost of stuff that does something, vs the cost of stuff to control and house that stuff that does things.

I was working at peavey when we first got involved with Media Matrix The big computer based audio control systems for big instals.

The established old school architects and consultants didn't want to deal with Peavey, but the cost savings from eliminating racks full of signal processing, and miles of wire to patch it all up, and man weeks of labor to wire it all up, with human errors. Cost savings on that scale are huge. and they had to swallow their pride and spec the Peavey, or lose the job to somebody way under their bid.

In recording and live PA consoles , outboard processing will find it's way inside the box on pure economics, and consistency between recording and live environments.

Regarding cost of electronics vs box and PS...like I mentioned.. 24bit codec<$1, dsp $2-3 crazy...

I need to start programming plug-ins but figure out some way to avoid getting ripped off...

JR

JR

Re: new shiny hammer- or Universal comp

Posted: Thu Jul 03, 2008 3:43 am
by XAUDIA
raf wrote:I have a Nord Modular synth that uses MIDI to connect the core to PC for editing and it is quite the cool synth. Better to avoid legacy serial like RS232 and MIDI.
(Slightly veering off topic - I have one of these too! They're great little synths and the software is cool. I had to buy a USB-midi adapter for it. I think the newer models are now USB.)

Re: new shiny hammer- or Universal comp

Posted: Thu Jul 03, 2008 8:35 pm
by JR.
as long as we're veering off topic, a friend of mine in Oz designed a guitar synth controller and he uses the nord for output.

JR

Image

name of product is Chicipik, but I don't think he's actively marketing it anymore.

Re: new shiny hammer- or Universal comp

Posted: Sat Jul 05, 2008 10:57 am
by JR.
OK I cranked out a rough faceplate... about a million LEDS but thats not a huge deal with micro.. All digital rotary encoders.

Display and two encoders on left set and indicate thresholds for noise gate (left) and limiter (right). Top legend is for limiter, bottom legend for gate.

Next switch selects between L and R, in master/slave mode this might step through multiple slaves.

Next switch selects status of group of 4 rotary encoders. These controls can adjust these 4 parameters independently for the noise gate, limiter, and a full range comp that operates between gate and limit thresholds.

Control 1 is slope or ratio, 2 is attack, 3 is hold, 4 is release.

L & R output meters always active.. Uses my simultaneous peak/VU display (dot/bar) So you can see crest factor directly.

Last rotary is level or make up gain. But this type comp doesn't typically need make up gain.

------
There are a few things odd about this approach. The full range comp, is softer than soft knee,, I should call it "no knee", It just compresses full range.

Limiter is conventional above threshold limiter

Gate-Downward expander is again unconventional. Threshold actually sets where it passes through unity gain, if the full range comp has added boost, this unwinds boost at a rate based on slope to intercept unity gain again at gate threshold. Below that threshold it reduces gain for noise floor reduction.

Note: for old fashioned fuddy duddies, they can use this like a conventional above threshold comp by ignoring the full range comp and adjusting the limiter for use as an above threshold comp.

I've actually done something very close to this in analog but I didn't have full parameter control over different dynamics sections. Limiter was set fast attack high ratio, no hold, Still was a full house of knobs and switches.

I'm biased but IMO this approach is logical to use and sections don't interact.

First set limiter for desired peak output, set noise gate for desired noise floor, and crank in full range comp to get sound as squashed or not as desired...

VCA status (gain) display indicates both boost and cut. The full range comp has an implicit threshold or 0dB gain inflection point. Below 0dB comp is adding gain to raise quiet parts, above zero it reduces gain. This is same thing as goes on in typical comp but it's concealed in limiting gain reduction and later make up gain... In my old analog version I did the gain display with a meter that started at bottom and ramped up for boost, started at top and ramped down for attenuation. The VCA gain display is by the thresholds so you can see it respond as you tweak thresholds.

This is going to be a $5 comp with $20 worth of knobs and display...

I would like to add more options, another switch in the control section could select between peak, ave. or RMS response for each function. Another switch in the same area could select between input or sidechain for each function, so you could EQ just the limiter or gate for improved selectivity.

Anything missing?

JR