Orange Squeezer Questions

[ptrx]

to get more compression from my orange squeezer i would:

a) use germ diodes instead of silicon ones.
b) use more then one diode in parallel (reverse / non-reverse)
c) use more then one diode in series
d) this has nothing to do with the diodes

any thoughts on these reflections? :?:

[cd]

B.  Put them in parallel, cathode to cathode, anode to anode.

[Processaurus]

D) isn't the trim pot for controlling how much compression there is in the OS?

[ptrx]

indeed, but it doesn't do a right lot on mine. Connections seem to be ok, pot is working. I heard from other people that there is supposed to be a "sound to no sound" region where you should leave the trimmer, but none of that occurs. Anyone else had this problem? I know a lot of you people built this unit.
(mind: there is def compression going on, it's just not very heavy and at a fixed ratio).

pat

[R.G.]

The orange squeezer is one of those devices that takes advantages of the quirks of a few components to do the effect. As such, it is limited in how and how much effect it can do. If you want lots of control of the characteristics of the compression, you probably want a different compressor. If you want some compression in an easy build, the OS is what you want.

The correct answer is closest to (d) - this has nothing to do with the diodes. (a) is quasi-correct; germanium has a lower forward conduction threshold than silicon, and the original pedal used a germanium diode. Silicon makes the compression cut in at a higher signal level. (b) will remove any effect, since what the diode does is to rectify the signal out of the opamp. Using two diodes in anti-parallel lets it rectify one half-cycle, then the other polarity signal sucks out of the rectifier what the first put in. (c) has even more of the problem that was noted for (a). Higher diode threshold raises the cut-in threshold for the compression even more.

Where the compression actually occurs is at the junction of the 82K resistor and the first JFET. The JFET can change its drain-to-source resistance as a result of a changing gate-to-source voltage.

For no signal, the source of the first JFET is held up at some ill-defined positive voltage determined by the current through the second JFET (the one with its drain connected to +9) and the trimpot. The gate of the first JFET is held down at ground by the 100K resistor in the rectifier section, so there is a negative voltage on the gate-source of the first JFET. This turns the first JFET off, and so any tiny signals too small to push the rectifier into conduction go through the 82K, the 0.047uF cap, into the + input of the opamp, and are amplified by the opamp's gain of 22 and appear at the output pot un-compressed.

The threshold of compression occurs when one of those signals is sufficient to forward bias the diode into conduction and start filling the 4.7uF signal rectifier cap. When that happens, the voltage on the cap rises, and that raises the voltage on the first JFET's gate towards the source; that is, in the direction of making it conduct more.

The JFET's channel resistance then acts like the lower leg of a voltage divider with the 82K and cuts the signal back some. Because there is a gain block (the opamp) and a feedback loop (the rectifier/filter/JFET), the thing tends to cut the signal back to a level close to the one that just cut in the rectifiers. The forward gain is low (22) so the feedback can only reduce the amount the signal is cut, not completely flatten it.

The gain of that opamp block is the closest thing to a compression control that the OS has. If you could run that gain up to several hundred or a few thousand, then the signal would get very, very compressed. Unfortunately, this is going to be very difficult to do in the real world because you'll run into oscillation problems fairly quickly, I think.

There's another "gain" here - the amount of drain-source resistance change per volt of change on the gate-source. JFETs with high "gain" (that is, go from on to off with only small Vgs changes) will increase the overall gain of the compression path and give more compression. However, they will distort the signal more because the ripple on the simple rectifier will also change the signal gain, and this will be heard as distortion. The JFET itself has a distortion mechanism because the gate voltage is really to the entire channel, not to just the source, and there is a V-drain-to-gate effect that causes distortion in variable resistor uses like this one.

The OS actually *decreases* the gain of the JFET with those two 470K resistor attached to the gate of the JFET. This reduces gain, but linearizes the JFET response.

And that's the whole secret of the orange squeezer - the more signal that you put in, the more it feeds voltage into the rectifier and the more it turns the first JFET on to cut the amount of signal getting into the opamp.

Notice I haven't said anything about that trimmer. The trimmer is there purely to adjust for the wide variation in Vgs sensitivity in same-part-number JFETs. The signal-cutting JFET needs to be biased just... barely... off to work well. Production JFETs have spreads of Vgsoff of 3:1 up to 5:1 these days. So the 9V connected JFET is hooked up like a constant curretnt source feeding that trimmer pot. The pot sets the bias voltage on the signal JFET to a point where it can cut in on small signals.

The trimmer is just to compensate for the variation in JFETs. It's NOT a compression control, an attack control, or any other control having to do with the compression other than just setting it to the right place to work.

The OS is a simple compressor. Its effect is all bound up in the gain of the opamp, the rectifier, and the JFET. If you just must play with it, the 1.5K in series with the diode is as close as it has to an attack control, the 100K and 4.7uF cap are as close as it has to a decay control. The gain of the opamp partially controls compression ratio, but has some definite limits on how high it can be raised. The voltage-to-resistance "gain" of the JFET partially controls the compression, but is nonlinear, so it's better to use some of that "gain" to linearize itself.

Does that help?

... I guess I just wrote the Technology of the Orange Squeezer...

[ptrx]

whoa
we have a winner  
thanks RG for your extensive reply, i hope it's of much help to many following OS builders.

ps: I actually like the compression a lot, it's very musical. There is some clipping, but i'll just experiment with different diodes to deal with that (got some germs coming up), what keeps me slightly worried though is that my trim doesn't do a thing for no apparent reason.... i guess i'll just blankly stare at it for a while to find the problem  :shock:  :idea:  :wink:

[R.G.]

what keeps me slightly worried though is that my trim doesn't do a thing for no apparent reason.... i guess i'll just blankly stare at it for a while to find the problem

Try using your voltmeter on the top of the trimmer pot while you rotate it. Does the voltage change with trimmer setting? It's possible that there's a circuit flaw that keeps the trimmer setting from making any changes in voltage.

Also, it's possible that a combination of the second JFET and the trimmer are not functional.

Since the only function of that trimmer is to get the thing to do compression, to move the first JFET into a range where it makes musically pleasing distortion, it may be that you're one of the lucky people where  every setting of the JFET is good. In that case, you may not *want* to know why the trimmer doesn't change things.

[sean k]

 So after reading your post Mr Keen I came up with this way to change the R and C in the attack and release sections you outlined.Think it'll work?
The switches are actually the centre off types.

[jc]

Great post on the OS RG

[rubberlips]

RG,

Any thoughts on what voltage we should be getting to setup the FET for compression with the trimmer? I know it depends on what type of FET you're using, but if you were looking at the data sheet what would you be looking for??

Cheers

Pete

[Mark Hammer]

Nice when someone who knows a lot has time on their hands to disseminate it.  Nice one, RG.

There is, in a sense, another type of "compression amount" control, which you alluded to indirectly.  If the 82k resistor and JFET form two legs of a virtual pot, then reductions in JFET resistance will have an impact on the signal level fed to the op-amp depending on the value of the resistor which is currently 82k.  If that resistor is increased to, say 100k, the JFET will have more impact on signal level.  Whether the response is a musically satisfying one, and whether the increased series resistance going into the op-amp takes away more than it gives, is something that I can't speak to, but in principle it should yield a more noticeable effect.

The bigger question for me is whether the OS should be tweaked and retweaked to have more compression.  It just seems to me to be analogous to taking a flathead screwdriver and grinding/machining it to be a Philips head unit.  Why go through all that work?  Why not just make a Dynacomp?

That being said, I stand by my earlier suggestions, distributed elsewhere here, to consider:

a) Tacking on a better input stage ahead of the 82k/FET "divider".

b) Not packing ALL the gain needed to drive the rectifier into one op-amp stage, but perhaps turning the one we see into a x4 stage (instead of x23) that feeds an audio output (though see below) PLUS another modest gain stage that feeds the rectifier directly, and whose gain can be adjusted to yield different compression curves without an impact on overall level.  At present, reducing sensitivity via reducing gain in the op-amp has an impact on output level.

c) How the OS might be used as a clean booster, by tacking on an additional gain stage AFTER the one we normally see.  The advantage of this is that tonal changes can be introduced between the two audio gain stages without having to sacrifice output level.  A common byproduct of compression is often a perceived loss of treble and bite.  The OS seems to be a bit less susceptible to this, but even so, it happens.  Following the existing gain stage with another permits the placement of a treble compensation network between the stages.

Overall, this suggests a revised OS that employs a quad op-amp: one stage for input buffering, one for post-FET gain, one for follower-drive, and one for audio-output/tone-shaping.

Now here's a question I'd like answered:  Why is the input biasing network on the stock schematic an "asymmetric" one involving a 390k and 470k pair?  Is this intended to offset what the diode subtracts?

[freeride]

I too have had my issues with wanting the orange squeezer to compress more to give some more spank to the tone, a la country lead.  The catch 22 here is that I use single coil pickups, so the input signal is lower and can't push the pedal as much.  I'm thinking about trying a booster in front, perhaps just a simple op amp.  Does anybody have any suggestions here?  As far as the biasing for the JFETs, don't forget that capacitors block low freqs from passing (ever heard of coupling?), so those spot are going to be DC-zero biased.

Great explanation on how the pedal works up there in the thread^^^.   

-zach

[matt239]

I don't really understand how the second jFet functions.
What is it doing & how does it do it?
Can anybody enlighten me?
Thanks.

[.Mike]

I don't really understand how the second jFet functions.
What is it doing & how does it do it?
Can anybody enlighten me?
Thanks.

Have a look here: http://www.diystompboxes.com/smfforum/index.php?topic=54308.msg414842#msg414842

[matt239]

Excellent! Thanks!
I could've done a search..
I thought i had, but I guess not really..

What would we do w/o R.G.?