Companding in a distortion pedal?

[Connoisseur of Distortion]

What could it do? I'm thinking that you could kick back the noise level significantly, and/or use it to compress the signal for a smoother attack and more sustain. besides, futurlec is selling those SA571 units for just about free, so it wouldn't be too bad to try...

this is an idea for my homebrew distortion pedal (the one with 5 stages of a 4049). i think i could use it to get a lower noise floor, and maybe use it to change the gain and have two separate channels.

opinions? if i recall, some of you guys know these things inside and out...

[Ben N]

Isn't there an EA fuzz article around with just such a circuit?

Ben

[Connoisseur of Distortion]

i see mention of this thing, but i have never seen any information about it. does anyone have the schematic around? or a soundclip? or something?

[Mark Hammer]

It was an ETI project, and also a PE project.  You can find it here: http://elektra.mine.nu/anders/pdf/susfuzz.pdf

First, let's distinguish between a fuzz with companding, and a fuzz that uses a "compander chip".  "Companding" is a process that constrains and restores dynamics in a complementary way such that the final output has dynamic properties virtually identical to the input.  Chips like the NE/SA570/571/572 are *usable* for companding, but they are open-ended enough that they can be used for other things too.  So, sticking a 570 in a circuit does not mean that companding is used in the circuit.

There was a circuit in ETI waaaaaayyyyyyy back that was described as a "clever fuzz box".  It used a 570, but did so in order to have a fuzzed output with the same dynamics as the input....the legendary "fuzz with dynamics".  The circuit I've linked you too does not do that.  Rather it uses one of the 570 section to assist in generating distortion, and the other to impose constant signal level (sustain).  Clearly, even though a "compander chip" is used, there is no companding going on. That doesn't mean it is necessarily a bad circuit/project.  But it does mean that this is a different beast than, say, the E-H Graphic Fuzz that  restores dynamics, post-clipping.

[Ben N]

Well, I've never built one and don't know anyone who has, but according to the text of the article, the control input of the "follower" section is switchable between the compressed signal, for "fuzz with sustain" and the buffered input signal, for "fuzz withoiut sustain--essentially restoring the opriginal signal dynamics to the squared-off signal--if I am reading this correctly.  Anyway, it does seem to be somewhat along the lines of COD's inquiry...

[Mark Hammer]

Not exactly, if *I've* understood it right.  There is the rather consistent signal level the distortion section spits out.  This is no different than the "naural compression" just about any clipper produces, whether it is deliberately aiming for compression or not.  On top of this, though, is the degree of consistency in the signal you feed the distortion circuit.  Most folks here know that if you plunk a compressor in front of a distortion box, you can get the same qualitative distortion over more of the lifespan of each note plucked (i.e., things fuzz the same way for a longer duration), but you likely won't get the particular attack tone typical of plugging directly into the distortion.  I believe that is what this unit does.  It's a bit like cramming a compressor and distortion pedal into one board.  The old Big Muff Pi Deluxe had a compressor (Soul Preacher) on board, but from what I gather ( http://ampage.org/hammer/files/Device1-7.PDF ), these were wired up in parallel, kind of like the way a clipper and clean path coexist in a Voodoo Labs Sparkle Drive.  In the case of the 570-based "susfuzz" project, it would appear they are wired in series.

[Brian Marshall]

isnt the 571 a dual compander....

i think it could be pretty sweet... have a toggle switch for a noise gaite/noise reducer, and compression.

[Mark Hammer]

Well it IS a dual compander, but here the rectifier circuit in the second section is being used as the fuzz source.

[stm]

I'd rather say it is a single Compander (not dual):

It has two sections that when one is configured as a Compressor and the other as an Expander can form a Compander (COMpressor-exPANDER).

You can have a dual Compressor or dual Expander if both sections are configured for the same task.

[Mark Hammer]

Absolutely correct.  I must have built my first compander around 1978 or so, for tape purposes (as poor man's dbx), using a 570.  Because it is so easy to reconfigure each section for the one function vs the other with a mundane pushbutton or toggle switch, circuits of the time described the chip as a dual compander, because it was assumed the two sections would be used for the same purpose at the same time: compress in stereo to tape, and expand in stereo when playing back tape.

But the less familiar one is with the chip, I guess the more important it is to stress that, yes, each section can only do one thing at a time - either compress OR expand dynamics.

[Joe Kramer]

What could it do? I'm thinking that you could kick back the noise level significantly, and/or use it to compress the signal for a smoother attack and more sustain. besides, futurlec is selling those SA571 units for just about free, so it wouldn't be too bad to try...

this is an idea for my homebrew distortion pedal (the one with 5 stages of a 4049). i think i could use it to get a lower noise floor, and maybe use it to change the gain and have two separate channels.

opinions? if i recall, some of you guys know these things inside and out...

Hi!

If you use the 570 just as a compressor, you might be stuck with a fixed 2:1 compression ratio, unless you know of a way to vary the ratio on that chip.  (I don't know how, but someone here probably does.)  If you use the chip as a compander around a fuzz circuit, you may encounter two problems.  One, you'll have to pump up the distortion capacity of the fuzz, because the compressor will knock down the peaks that the fuzz normally works on.  Two, the expander will cut the sustain back to the original input signal, so you'll lose the noise, but also any sustain added by the fuzz.  FWIW.  Anyway, give it a try--it might sound great!

Joe
 

[Mark Hammer]

If you use the 570 just as a compressor, you might be stuck with a fixed 2:1 compression ratio, unless you know of a way to vary the ratio on that chip.  (I don't know how, but someone here probably does.)

I don't think the compression ratio can be altered, but where that ratio is applied can be.  So, applying a 2:1 compression ratio to just the peaks of transients will sound different than applying the same ratio to anything above a whisper.

If you use the chip as a compander around a fuzz circuit, you may encounter two problems.  One, you'll have to pump up the distortion capacity of the fuzz, because the compressor will knock down the peaks that the fuzz normally works on.  Two, the expander will cut the sustain back to the original input signal, so you'll lose the noise, but also any sustain added by the fuzz.  FWIW. 

Again, yes and no, depending on how it is set up.  Keep in mind the chip has two independent rectifier/followers.  If the rectifier in the expandor half was working from the same signal as the one in the compressor, you would be correct in assuming that original dynamics would be restored/replicated and the sustain added by clippiing would not be that apparent. However, the rectifier in the expandor half can also take its feed from whatever you feed it.  And if the dynamics of what has already been compressed is further squished by the clipping process, then the dynamics will more closely approximate what the fuzz did, when it gets expanded by 1:2.  It the fuzz essentially eliminates any dynamic variation as you hold a note, then the expandor will have no dynamic differences to accentuate.

I guess think of it this way...A guy walks into a store with a wad of one dollar bills and they make him change all his money from dollars into $5 bills.  He walks around a bit and as he leaves, they change all his $5 bills back to $1, such that the pile of money was smaller for a little while and was restored.  Another guy walks in to the same place, gets his ones changed to fives.  But then he goes to a sales clerk that changes everything into $50 bills.  As he leaves the store, somebody changes his much smaller pile of $50 bills into a slightly taller pile of $10 bills.  In the grand scheme of things, the pile going in and the pile coming out have been compressed/converted and expanded by the same ratio.  But the difference in absolute height between piles, prior to and after leaving the store will be different, because of what has gone on inside the store.

Make sense?

[Joe Kramer]

However, the rectifier in the expandor half can also take its feed from whatever you feed it. 

Hey Mark,

Right, a point I forgot.      So, off the top of your head, from where in this hypothetical fuzz circuit would you take a feed for the expander rectifier, so that the noise would be quashed but the sustain of the compressed-fuzzed signal would still be apparent?

Joe

[Zero the hero]

Maybe a bit off-topic, but this circuit uses a compander in the fuzz section. I haven't built it yet...

http://img.photobucket.com/albums/v368/zero_thehero/schematics/GSP05.jpg
http://img.photobucket.com/albums/v368/zero_thehero/schematics/GSP06.jpg

[Mark Hammer]

As shown in the susfuzz schematic, the switches formed by IC3a and IC3d are complementary - when one is open the other is closed.  This does a few things.  When switch IC3a is closed and 3d open, the rectifier is fed with the input (uncompressed) signal.  When 3a is open and 3d closed, the second rectifier is fed with the audio output of the compressor.

Now, the second 570 stage is configured as expandor.  When it is fed the output of IC1, it functions as a true single-ended expandor, accentuating dynamics.  But the first stage, configured as a compressor, has its rectifier fed with the same signal, which ought to, in principle restore dynamics, minus whatever it is that little single transistor "fuzz" section takes away.

The rather interesting thing, though, is what is happening on pin 16 vs pin 1 of the 570.  The 2u2 cap on each of those pins sets the attack/decay parameters (which as the little text box in the article notes, is set internally at 5 times the decay time, compared to attack time).  BUT, the 1M resistor in parallel (R12) means that the cap (C16) discharges just a bit faster on expansion than on compression.  This provides a wee bit of what essentially amounts to downwardexpansion thatgets the expandor to shut down a little earlier than anticipated, so as to reduce gain during the very quietest parts, thereby eliminating  residual hiss (or trying to).

Okay, taking everything into account, it seems to me that if you wanted the MOST squashed signal dynamics you could get out of this circuit, you could do one of three things.

• take C14 and feed it directly to pin 9 of IC3 (the switch section labelled IC3c) such that switching between the outputs of IC3C and IC3b gives you completely clean (IC3b closed, 3C open) or a compressed, fuzzed (but not expanded) signal (3b open, 3c closed)
• reconfiguring the expander half to take its rectifier input from the wiper of RV1, such that whatever clamp on dynamics this "fuzz" imposes will further impact on the rectifier to produce *some* limited expansion but not enough to restore original dynamics; of course what is interesting here is that the signal feeding the rectifier would vary with RV1 setting, wouldn't it?  could be interesting.
• alternatively, reconfigure the expandor half to be a second compressor by tying the + sides of C14 and C15 to pin 10, and feeding RV1's output directly to pin 11 (cutting the connection between pin 11 and 10)

[A.S.P.]

I don't think the compression ratio can be altered, but where that ratio is applied can be.

page 10-7, fig. 18




[edit]: Compander Cookbook

[Mark Hammer]

Yup.  saw that after I posted. Although, apparently it can get pretty involved.

[Joe Kramer]

Compander Cookbook

Thanks for the link!  IIUC, the ratio can be changed to less than 2:1 but not more.   

Quoth Hammer:

# reconfiguring the expander half to take its rectifier input from the wiper of RV1, such that whatever clamp on dynamics this "fuzz" imposes will further impact on the rectifier to produce *some* limited expansion but not enough to restore original dynamics; of course what is interesting here is that the signal feeding the rectifier would vary with RV1 setting, wouldn't it?  could be interesting.

Yes, driving the rectifier of the expander with the fuzz's output seems the intuitive thing.  By the description in the article about the "long delayed attack" when the unit is set for fuzz but not "sustain," this suggests a nice "swell fuzz" effect, which someone was seeking in another recent post.  Could be interesting indeed. . . .

Joe

[Ben N]

Not exactly, if *I've* understood it right.  There is the rather consistent signal level the distortion section spits out.  This is no different than the "naural compression" just about any clipper produces, whether it is deliberately aiming for compression or not.  On top of this, though, is the degree of consistency in the signal you feed the distortion circuit.  Most folks here know that if you plunk a compressor in front of a distortion box, you can get the same qualitative distortion over more of the lifespan of each note plucked (i.e., things fuzz the same way for a longer duration), but you likely won't get the particular attack tone typical of plugging directly into the distortion.  I believe that is what this unit does. 

Right, BUT if you have an expander/envelope follower after that compressor/clipper chain, and it's control input is getting the full dynamic range of the original uncompresssed, unclipped signal, then that will (in theory, at least) restore some, although not all, of the dynamics to the distortion product.  I am under the impression that that is what this circuit aspires to do, when switched to the non-sustaining fuzz setting.

BTW, if I ever get around to breadboarding this, I'd be curious to see how the concept works with something other than just a heavily overdriven BJT in the loop--like, say, a BMP clipping stage, a differential distortion or an octaver.  In general, though, I prefer no compression ahead of my distortions because I do like the way the distortion blooms with the envelope of the note.

Ben