dual / multi band compressor control options?

[samhay]

My never-quite-ending dual-band compressor design has floated to the top of the pile again, but I am a little stuck on how much control is too much for a bass / guitar compressor. I figure you good folks might be able to offer some suggestions / experience.

I like the idea of having an adjustable crossover frequency, so at a minimum, I guess it needs 5 controls: Crossover frequency; mix (between each channel); comp (high-pass); comp (low-pass), and level.

Where I'm stuck -
I am not sure whether having a separate envelope detector for each channel is necessary (I guess I need more breadboard time too). I can see it being nice sometimes, so could add this as a switchable option, but then we have more control.

I like attack and release control - if I have separate envelope detection, do I then have separate attack and release control? If so, we are up to 9 knobs and maybe a switch.

Hmm. Thougths most appreciated.

[Mark Hammer]

We're most familiar with multi-band compression in the studio context.  But keep in mind those units are designed in anticipation of grappling with much wider bandwidth than what our measly instruments provide.

That is not so suggest that multi-band compression is pointless.  Rather, I think once we get to two bands, we've pretty much covered the bases for bass or guitar or voice.  Drums or synth may be a different matter, but stringed instruments and vocal cords would, I suspect, be amply handled by two bands.

But which two?

Ideally, the different bands use different time constants.  One would likely need a full array of continuously variable attack and release controls if you have no idea what the specific basspands are going to be.  But if you have defined bands, then I'd imagine you could probably make do with fast/slow switches for each band, since those speeds would be tailored to the characteristics of that band.

[samhay]

Dual band compressors are fairly popular amongst bass guitarists. Not planning on going to 3-band at this stage (more of a thought experiment).

I am looking to use a 2-pole Sallen-Key subtractive crossover with a crossover frequency that is variable from about 100Hz to 1kHz. I guess a 3-way switchable attack and release could work, but this is probably a better solution for a fixed crossover frequency.

[tubegeek]

I know that the Rane DC-24 is a now-discontinued, still-popular two-band compressor with enormous flexibility, bass guitarists love the thing. The schematic and manual are up on Rane's website. Maybe they might inspire you?

Schematic:
http://www.rane.com/pdf/old/dc24sch.pdf

Manual:
http://www.rane.com/pdf/old/dc24man.pdf

[Johan]

You could probably fake a multi Band compressor by placing a single band parametric eq before a regular compressor. ... Just a thought. .

[PRR]

> not sure whether having a separate envelope detector for each channel is necessary

IMHO, separate level detectors is the WHOLE point of 2-band.

> we are up to 9 knobs and maybe a switch.

There does come a point that a box has so many knobs you can never find the sound you want.

> I guess I need more breadboard time too

Very wise.

And don't discount simple plagiarism. You are NOT the first down this path. Start with the trail blazed by those who went before you.


> the Rane DC-24

Looks to me like a stereo single-band compressor.

True multi-band processors split each channel into bands, process each band, then sum. In AM radio this specifically addresses the problem of the vocal dipping-out on every thud of the bass.... they are in separate bands, the voice-band processor does not hear the bass thuds.

You can't fake this with EQ in front of a single-band.

Multi-band processers were VERY hot-stuff in AM radio some decades back. They still are, but DSP technology allowed 13-bands, nonlinear action, and peak-skewing to MAXimize AM loudness (at the cost of listenability?).

I have a strong feeling that we don't want 2-band guitar processing for the same reason that guitar speakers are almost universally one-way (not woofer/tweeter). The guitar is a very holistic instrument. Breaking it up sounds wrong.

An additional point: unlike the AM broadcaster with the hot 45 in hand, _you_ have separate control over each note. (I am well aware that this is a Skill, and one I lack.) If bass thuds are overwhelming your soprano notes, don't pluck the bass so hard.

[StephenGiles]

An additional point: unlike the AM broadcaster with the hot 45 in hand, _you_ have separate control over each note. (I am well aware that this is a Skill, and one I lack.) If bass thuds are overwhelming your soprano notes, don't pluck the bass so hard.

If bass players stuck to bass and didn't try to play lead guitar on it (John Entwistle & al!!), problem wouldn't arise - ducks and runs for cover!!!!!

[electrip]

If bass players stuck to bass and didn't try to play lead guitar on it (John Entwistle & al!!), problem wouldn't arise - ducks and runs for cover!!!!!

Thou shalt not take the name of the master bass player for any joke. 

The world would be better if those doom guitar players wouldn't tune down to C#
and use the worst sounding fuzzes with up to 1mF input capacitator so any bass player is swamped out into oblivion.

electrip

[midwayfair]

This is what I'd do:

-Two envelope detector circuits.
-Attack control in the treble band with a fixed short decay of about 50-100mS. Use a FET for the variable resistance element -- it's faster than a photocell. You want the attack control on the treble band because a faster attack swallows more treble -- this will double as a type of tone control.
-Decay control in the bass band, using a photocell -- this ensures a smooth decay even at lower settings. This is where most of the "sustain" is heard anyway.
-Two threshold controls.
-Master volume.
-Input gain trimmer if you want to handle higher output pickups.
-Fixed "ratios" for each band; the treble band should probably be no higher than 4:1 (and personally I'd prefer something closer to 2:1), and the bass band should probably go a little higher, probably at least 8:1, but something more like 12:1 would be awesome. I'm just talking about the total gain reduction here, so you can probably handle this with a couple resistors in series or parallel with the FET or photocell.

There you go -- five knobs and a trimmer. The controls are all where they do the most good.

If you really, really want to take things a step further, don't focus on what controls you've got in the envelope detector but how you handle the signal. In a dual band compressor, the most important thing is going to be the separation of the bands. So use multiband filtering at the envelope inputs AND when you split the dry signal. I would say that for a two pole filter, ~98-147Hz is probably good for a bass band (depending on whether you want it at the G string on a bass or the D string on a guitar), which is about where I tend to set the low band on the multicomp on recordings.

[samhay]

Thanks for the comments.

I have a fairly good idea how to build this thing and am leaning towards a dual OTA rather than FET or LED/LDR.

>  I would say that for a two pole filter, ~98-147Hz is probably good for a bass band
Thanks Jon. I was hoping someone with some experience would wade in, and you seem to have more than most when it comes to comps.

> There does come a point that a box has so many knobs you can never find the sound you want.
That's the crux of my dilemma at the moment. I think I will put it back on the breadboard with as much control as possible (will post a schem once it is known to work as intended) and try and use it for a while to see what can be paired away.

Existing designs haven't been much helps, as these both seem popular:

[tubegeek]

> the Rane DC-24
Looks to me like a stereo single-band compressor.

No, there's more to it than that, but it has a kind of an odd implementation. It has a crossover, as well as these features/controls:

"To use the DC 24 's crossover bandsplit mode, be sure the rear panel CROSSOVER ENGAGE switch is in. Connect the input to CH 1/CROSSOVER IN. With the rear panel OUTPUTS switch set to SEPARATE, split outputs are available at the CH 1/LOW OUT and CH 2/HIGH/COMBINE OUT jacks. With the OUTPUTS switch at COMBINE, use only the CH 2/HIGH/COMBINE OUT jack for a mono sum of high and low channels."

"SEPARATE / COMBINE OUTPUT switch: In the COMBINE mode, the outputs of channel 1 and channel 2 are added
together and delivered to the CH 2/HIGH/COMBINE OUT connector. This feature is supplied to allow an input to be split
by the crossover, high pass and low pass delivered to separate processor channels, and recombined at the output of the unit. The
COMBINE mode has no effect on the output of channel 1."

"CROSSOVER FREQUENCY control: This rotary control selects the crossover frequency. What did you think?"

[chicago_mike]

HA! SSM2018 is the same chip inside the Hartke amps!