Very simple compressor. Works well IMO

[brett]

Hi.
I find most compressors either quite poor (orange squeezer) or quite complex (Dyna).
Now the Dyna is even harder to make coz the CA3080 seems to be out of production.

This simple circuit works quite well...
(Apologies if this or something very similar has been posted before).
http://members.optusnet.com.au/~jethro.dog/mypic168.jpg
By replacing the 1k resistor between pins 1 and 8 with a 100k pot, you could make gain (and therefore compression) variable.
Also, you might consider replacing the 1N4148 with a Ge or Schottky diode.  However, I've tried this and didn't like the extra compression.  I suspect that the MOSFET spent too much time "on" (or too "on"), and the input impedance was reduced too much for my pickups.

cheers

[MartyMart]

That looks sweet and .... simple ! 
Wouldn't the 1k work as well with say a 4k7 pot ? 100k pot seems
quite large !

Thanks for sharing

Marty.

[bancika]

Any clips? 

[MartyMart]

Hi Brett,
Just breadboarded this and have a couple of questions :
First, it's working quite well, though HB pups make it break up slightly
For short "picking" style its fine but on slow chord strums, the "bloom"
of the compression is about 1.5 seconds.
Would reducing the 22uf cap to say 10uf or less help, or would it cause too
much ripple in the release stage ??

Thanks,
Marty.

[Arn C.]

Anything new on this Brett or Marty?   Looks real interesting.  I will also build this up and try out a few things when I get the chance.

Thanks for the new design!
Peace!
Arn C.

[Johan]

..shouldnt it be a resistor between inputcap and FET for it to make it work like a pot?...now it just loads the guitar down..might still work, but will behave very different depending on whats feeding it...

johan

[vanhansen]

Hey, I'm likin' the looks of this one.  My kind of circuit.  Low parts count to keep it simple and yet still keep me busy..lol.  Moving at my own pace to higher and higher part count circuits.

I'm interested to know what else you've found out about it, Marty, and what Brett has to say about your observations.

[Arn C.]

I think that I would try a 5k pot first, instead of the 100k pot and also try values like 2.2uf up to 100uf instead of the 22uf.    Just my two cents!  I will try this once I get it built...

Peace!
Arn C.

[MartyMart]

Well here's how i improved it with my HB pup GTR and rig :
Changed the 22uf for a 10uf, this does speed up the "bloom" of the compressor
Changed the original 1k5 ( all I had ) for an 820 Ohm in series with the "new" 10uf between
pins 1 & 8
Stuck a 100k resistor between the 1M to ground and input cap ( no more break up )
I did try a cou[ple of pot values ( 25k and 100k ) as suggested, but with "so-so" results
making it not worth using up an extra pot.
I'm quite happy with it now, it's more of a "use it gently" compressor, if you crank it too
much, it just gets a bit "ugly"  !!
Dialed into where you just feel it happening is sweet

MM.

BTW: It has a LOT of boost available, which is a nice additional feature !

[David]

Brett:

Good on you, mate!  Nice work.

Well, we've got the Ross Comp and we've got the Dyna Comp.  I think we should call this the "Aussie Comp"!

[Steben]

Luv it, mate.
Simple notes and hints:
-One could put a pot between the BS170 and ground, acting as a compression line control (to set max gain loss).
-Nice move using a overdrive-tone-if-blown 386 for this.
-Definitely use a 5k pot on the 386 as a sensitivity control.
-The 22u cap can be left, but accompanied by several others with a switch for a "release" control. What exactly is that 10k pot for?

[Arn C.]

-One could put a pot between the BS170 and ground, acting as a compression line control (to set max gain loss).

I will assume between source and ground.  Is this correct?

Thanks!
Arn C.

[MartyMart]

What exactly is that 10k pot for?

That's controling the "amount" of compression for me !!
Sends signal through the diode to the BS170 ....

MM

[brett]

Hi.
Thanks for all of the positive comments!!  This amazes me...posting a schematic and immediately it's built and being modified and improved.

RE:

For short "picking" style its fine but on slow chord strums, the "bloom"
of the compression is about 1.5 seconds.  Would reducing the 22uf cap to say 10uf or less help, or would it cause too
much ripple in the release stage ??

As the later posts noted, a 10uF cap reduces the release time from about 2 seconds to 1 second.  I get good sustain with my guitar, so the bloom didn't need to "kick in" so soon.  (I use heavy strings)  Replacing the 1M "release" resistor with a 1M pot would give adjustable release time, but my concept was for a really simple compressor.

The idea of replacing the 1k feedback resistor between pins 1 and 8 is a good one.  The 1 resistor sets the gain of the 386 to about 40.  A 10k pot would make it adjustable from about 25 (10k) to 200 (0k).  As someone noted, the 386 does break up a bit with the 1k resistor, and a better fixed values might be 2.4k to 4.7k, to reduce the gain a little to suit gutars with humbuckers, playing through boosters, etc.  You'll also notice that turning the "compression" pot to maximum tends to limit distortion, because there's faster, more pronounced compression of the signal.

Thanks for the interest.

PS  I like the name "Aussie compressor".

[StickMan]

I'm way too new at this to have any kind of really informed opinion.  I'd appreciate any comments to my suggestions/ideas listed below from people who have more experience than myself:

First off, this is very cool.  I've been doing some reading, and pondering why I've never seen a compressor using a mosfet as a variable resistor anywhere.  This looks like proof that it can work, and work well.

Just off the top of my head, it looks like the circuit might benefit by having an input buffer to isolate the guitar from the varying input impedance of the circuit as the mosfet gate voltage changes.

Essentially, the mosfet is going to behave very close to the led/ldr setup for most optical compressors, in that the mosfet can only source current, and without biasing the gate signal you're only going to be compressing the +ve side of each wave.  The diode plus the 4.7uF cap is probably smoothing out the -ve side of the waves, and holding some +ve gate voltage as the waves drop -ve.  At the expense of a little more complexity, it seems to me that if you put a full wave rectifier on the gate circuit, and a slightly more sophisticated peak hold scheme in there, the performance of the envelope follower part of the compressor could be improved dramatically.

I'm also confused about the use of the 386 here.  It's simple, but it probably sucks way to much juice out of the battery, adds a lot of noise and might have a crappy slew rate for this kind of application.  Especially with the 1-8 jumper in place, there's a lot of gain coming out of there that needs to be shed before you go into another pedal or an amp.  It almost looks more like a compressing amplifier than an effect pedal the way it is.  I'm wondering if the circuit was changed to use a TL072, split the input immediately into two parts, run one through 1/2 the op amp + rectifier stuff to supply the gate voltage and then run the other part into the other half, with the variable resistance of the mosfet affecting a feedback circuit on that 1/2 the op amp would give better performance.

Finally, I'm with Brett 100% on how cool it is that a circuit can be invented, and posted and within hours people have actually constructed it, tested it, and are suggesting improvements.

dave.

[Steben]

-One could put a pot between the BS170 and ground, acting as a compression line control (to set max gain loss).

I will assume between source and ground.  Is this correct?

Thanks!
Arn C.

Yes, you could call this one "depth" too. Since the MOSfet is a variable resistor, with a theoretical minimum Z of zero, placing a pot in series raises the minimum Z (reducing the max compression). But don't use small pots (1k-10k), they won't stand up to the impedance, try at least 100k.
I guess the 10k pot there acts more as "treshold" then or something likewise?

Just off the top of my head, it looks like the circuit might benefit by having an input buffer to isolate the guitar from the varying input impedance of the circuit as the mosfet gate voltage changes.

Very true. Saving space on the board, it might be one buffering half of an opamp, the other half could substitute for the 386!

Essentially, the mosfet is going to behave very close to the led/ldr setup for most optical compressors, in that the mosfet can only source current, and without biasing the gate signal you're only going to be compressing the +ve side of each wave.  The diode plus the 4.7uF cap is probably smoothing out the -ve side of the waves, and holding some +ve gate voltage as the waves drop -ve.  At the expense of a little more complexity, it seems to me that if you put a full wave rectifier on the gate circuit, and a slightly more sophisticated peak hold scheme in there, the performance of the envelope follower part of the compressor could be improved dramatically.

Yes, the MOSfet gets its guiding curent from the positive rectified signal, but since this determines a neutral "Z", the MOSfet does load the complete signal. The positive half "represents" the total amplitude, but doesn't tell the MOSfet what to load, just how much.

I'm also confused about the use of the 386 here.  It's simple, but it probably sucks way to much juice out of the battery, adds a lot of noise and might have a crappy slew rate for this kind of application.  Especially with the 1-8 jumper in place, there's a lot of gain coming out of there that needs to be shed before you go into another pedal or an amp.  It almost looks more like a compressing amplifier than an effect pedal the way it is.  I'm wondering if the circuit was changed to use a TL072, split the input immediately into two parts, run one through 1/2 the op amp + rectifier stuff to supply the gate voltage and then run the other part into the other half, with the variable resistance of the mosfet affecting a feedback circuit on that 1/2 the op amp would give better performance.

The amount of current drawn from the 386's supply is more determined by the primary use in power amps (driving low impedance of the load eg. speakers). I don't know what the "idle" current draw is of a 386 with high impedance output loads.
Your opamp suggestion might rip the simplicity in pieces. I would go for the buffer + gain opamp idea (which I came up with after your buffer suggestions anyway  )

[brett]

Hi.
As with all things simple, there's some complex thinking in the background...

RE: Why a 386, not a TL072?
Because the TL072 input is at Vbias (usually 4.5 volts), while the 386 input (and output) is at ground, so there's no need to isolate the IC.  See the note below about impedance for another advantage of the 386.

RE: Does it compress only half the signal?
The 22uF capacitor and 1M "bleed" resistor hold the MOSFET "on" for a long while (a few seconds ?), so there's no rapid changes in the release of compression.  In fact, it probably holds on a bit too long.  Maybe the 4.7uF cap should be lower (1uF), or even switchable, depending on whether you are playing hard and fast (4.7uF is then ok), or playing a soulful solo and want "bloom" without waiting too long (probably want 1uF then, or reduce the 1M to 330k or 220k).

RE: a buffer
This can be tested, but I think it will work much better *without* a buffer.  One of the reasons is that the 386 has only 50k ohms of input resistance, and when compression kicks in, this is reduced further.  This is a benefit, not a problem.  The fuzzface has an input impedance of approx 20k, and is probably a reason why it "blooms" so well.  I expect that when you're plugged into the Aussie Compressor, and hitting your strings hard, the input impedance is quite low, and a large proportion of the signal is lost in the internal resistance of the pickups.  Ease off and the signal starts to flow through again.

cheers

[MartyMart]

I attached a 1M pot this morning, in exchange for the 1M resistor from the mosfet, this allows
much greater control
I also upped the pin 1-8 resistor to 4k7 as suggested by Brett.
Perhaps ( and I know its getting more complex ) a switch to allow a 1uf/2.2uf/4.7uf cap here
would also provide some nice "smooth ---- choppy" settings
Somehow, I'm still getting a small bit of "breakup" though .... ?
If I could get it to be 100% clean, it woud be fantastic

MM.

[MartyMart]

OK, solved it .... so to "sum up" my changes :

100k input resistor before first 0.1uf cap
8k2 and 10uf between pins 1 and 8
10k pot and 100k vol pot "as is"
1M log pot replaces the mosfet 1M to ground, wired as variable resistor ( lug 3 -- lugs 1&2 joined  with lug 1 to ground )
4.7uf cap changed to 2.2uf

I now have a "nice" response, no breakup unless comp pot is "maxed out" and I have a "threshold" pot, which is what
I'll call the new 1M pot from the Mosfet's gate to ground.
Very cool, there's a sweet spot with comp at around 2/3'rds and threshold at 3/4's which just cuts back when you
hit hard 

MM.

[brett]

Hi.
You're a champion MartyMart.  Just one more question: is that 100k in series with the input, or from the signal path to ground?  In series?

If it's in series, it'll lessen the load on the pickups.  This might help prevent any treble loss.  But a series resistor will substantially reduce the signal going into 386.  That's probably why the distortion has disappeared. 

What I'll try, or someone alse could try, would be to have no 100k input resistor, and also no resistor and capacitor linking pins 1 and 8.  That should reduce the gain by about as much as the 100k input resistor would do (I think ??). 

There might be some trade-offs between having a constant 100k load on the pickup to maintain tone, and not adding the resistor and getting some "bloom" as the MOSFET turns off and the pickup gets "unloaded".  Some weekend testing might be in order.

thanks again for the interest
cheers
PS.  I've just realised that this concept would work around a single JFET or BJT amplifier, too.  A JFET should work really well, as the gate is usually at ground, and any distortion sounds very mild ...I'll get back to you tomorrow (it's 11pm here).