orange squeezer question

[guitarkill]

I am having a problem with distortion, I can't get rid of it. Built it using the GGG PCB layout, and have double checked everything partswise. I'm using a 1N34A diode until I can find suitable replacement for the unobtainable 1N100, and using 2N5457 FETs. I found where it is mentioned that a 1N270 should be an OK substitute for the diode, but I am wondering: has anyone has found it necessary to change the 10K trimpot to something bigger to stop the distortion? Just thinking ahead a little bit in case the 1N270 doesn't fix the problem....

[theehman]

Send me your address via email and I'll send you a 1N100.  I'd be interested to see if the diode makes a difference.  FWIW, I've built several Orange Squeezer clones using 1N100 and have never had a problem with distortion.

[guitarkill]

Send me your address via email and I'll send you a 1N100.  I'd be interested to see if the diode makes a difference.  FWIW, I've built several Orange Squeezer clones using 1N100 and have never had a problem with distortion.

Cool, that's nice of you. I'll send you a PM.

Yeah apparently it has to do with the forward voltage of the diode. My 1N34As all measure 400mV or higher, which is like twice what the 1N100 is. So I surmise that is because the diode is not really whacking the positive peaks like it's supposed to be doing per "the technology of the orange squeezer" here:

http://www.diystompboxes.com/smfforum/index.php?topic=54308.0 (see post #6)

[GGBB]

The distortion problem is not likely due to the diode, and it's not that uncommon.  I've read more than a few build reports that complain about distortion, and also experienced the problem first hand.  It is normally resolved by raising the 10K R10 to 12K or 15K.

As for the diode, a genuine 1N34A should be fine, but these have been known to be counterfeited.  I had very good results with a BAT48 Schottky.  1N5817 will also work well.

[guitarkill]

The distortion problem is not likely due to the diode, and it's not that uncommon.  I've read more than a few build reports that complain about distortion, and also experienced the problem first hand.  It is normally resolved by raising the 10K R10 to 12K or 15K.

I've tried different opamps, I get distortion with all of them. That is wierd, why would it work fine for some people and not for others?

[GGBB]

I've tried different opamps, I get distortion with all of them. That is wierd, why would it work fine for some people and not for others?

Try increasing R10, not changing the op-amp.  The gain of the op-amp is set by R9 and R10 - which with stock values makes the gain 23x no matter what op-amp you use (1 + R9/R10).  Pretty much any op-amp can handle that without problems.  I'm no expert here, but I think the problem is that if your input signal is large enough (i.e. your pickups are hot enough), that can be enough gain to cause clipping, and remember that the original OS design was likely for low-medium output pickups of that era.  So by increasing R10 you lower the gain and avoid the clipping.  You can also make the op-amp gain adjustable with an external pot or internal trimmer - replace R10 with a 4.7K-6.8K resistor in series with a 10K-25K pot wired as a variable resistor.  This works like a threshold control for the compression and is quite useful as a way to control the amount of compression effect as well as to avoid clipping.  Have a look at Mark Hammer's take on the OS, and other refinements of his ideas:

http://www.generalguitargadgets.com/diagrams/orangesqueezer_marks_variable.gif
http://hammer.ampage.org/files/Tangerine_Peeler.gif
http://s1190.photobucket.com/user/GGBB1/media/Pedals/Compressor/OrangeSmoothie_Rev3A_Schematic_zps97630085.png.html?sort=6&o=5

[midwayfair]

There's no reason to fret about the diode or chase an unobtainable one. It does have an effect on the threshold, so shooting for a low Fv is a good idea, but there's nothing magical about the 1N100 and I don't even think it's the best choice. There are plenty of Scottky diodes that will have the right Fv.

I have a bunch of 1N100s, but I still put a 1N60P in there on the best OSs I've made. It's a 4c diode that works even better -- and I don't have to worry about temperature drift making the threshold change with the temperature.

Here's a list of subs in the right Fv range that will work better than 1n34a and are easier to find than 1N100:
Smallbear's "generic" germanium diodes
1N60
1N60P (Schottky workalike)
1N192
1N270 (top of the "good" range)
Almost all Russian detector diodes -- e.g. D9E, D9B, D9V, etc.
some BAT41 (I've gotten these with Fv .35 and .25 averages ...)
BAT43
BAT85
1N5817

Pretty much all of these diodes can be gotten any day of the week, and most of which can be gotten cheaply.

It's possible that a lower Fv diode WILL affect the distortion by triggering the compression at a lower threshold, but it's not going to affect transients before the compression turns on. But you shouldn't be able to hear them anyway. You will likely hear a bit of distortion regardless, though, as it's a known 'feature' of the OS ... personally I feel that if you want a completely clean compressor, this isn't the one to build and you might just get frustrated trying to push it beyond its design limitations.

[guitarkill]

It's possible that a lower Fv diode WILL affect the distortion by triggering the compression at a lower threshold

That is what I was thinking initially, but it sounds like the opamp has too much gain for whatever reason. Which leads me to believe that maybe my FETs are not operating in the range in which they should be, which is why I thought maybe I need a bigger trimpot. Still not completely convinced that I don't, but will try these other suggestions first.

You will likely hear a bit of distortion regardless, though, as it's a known 'feature' of the OS ... personally I feel that if you want a completely clean compressor, this isn't the one to build and you might just get frustrated trying to push it beyond its design limitations.

I'm OK with a little distortion, but mine has too much and it's just not very musical sounding. I guess I'll try messing with the value of R10 and some different diodes if that still doesn't get me where I want to be. I know that the opamp has a fixed gain of 23 and it's supposed to provide the extra signal level to bring up small signals to a constant level, which is why I'm leaning towards it being the diode, FETs or some combination of the two. I don't hear much compression either so something is not right.

[guitarkill]

I noticed that there is no compensation cap on the opamp on the original design. Can that also contribute to having more distortion?

[Mark Hammer]

Not always, but sometimes, what some people describe as "distortion" in effects that use a sidechain, such as compressors or autowahs or gates, is basically envelope ripple.  As the note decays, there are random deviations from average in the amplitude, that occur at a high enough frequency that they move into the audio range.

One way to determine if this is what is happening in your case is to temporarily tack on another electrolytic cap in parallel with the 4u7 that's already there, like 4u7 or 10u.  That will change the feel of the compression, but if it eliminates what you are describing as "distortion" then we've narrowed down what it is.

[GGBB]

Here's a list of subs in the right Fv range that will work better than 1n34a and are easier to find than 1N100:
Smallbear's "generic" germanium diodes
1N60
1N60P (Schottky workalike)
1N192
1N270 (top of the "good" range)
Almost all Russian detector diodes -- e.g. D9E, D9B, D9V, etc.
some BAT41 (I've gotten these with Fv .35 and .25 averages ...)
BAT43
BAT85
1N5817

Of the BAT41, BAT42/43, BAT46, and BAT48 that I tried - my BAT41s had the highest Fv (~380mV) and they didn't quite get the job done for me.  All the others were in the low 300s or below and the BAT48s were the lowest.

[induction]

I don't hear much compression either so something is not right.

It sounds like maybe your bias trim is set incorrectly. On one extreme it should be a cleanish boost and on the other it will kill the signal completely. There's a very narrow band between the two extremes where the volume changes drastically with small tweaks of the trim. Turn it to the clean boost side, then adjust it until the volume just starts to drop and leave it there. That's the sweet spot where it will leave the smaller signals alone, and compress the louder ones. The compression should be pretty noticeable, but not extreme like a maxed dynacomp. The stock circuit is not super fast, so the attack of each note is retained it will sound kind of clicky, which makes clean notes pop nicely and works very well for chicken picking. If you play with more dirt, the clickiness will be reduced by compression from the dirt.

Setting the bias correctly may also help get rid of the distortion. Mine had some nasty clipping with loud signals that went away when I set the bias correctly.

[guitarkill]

Not always, but sometimes, what some people describe as "distortion" in effects that use a sidechain, such as compressors or autowahs or gates, is basically envelope ripple.  As the note decays, there are random deviations from average in the amplitude, that occur at a high enough frequency that they move into the audio range.

One way to determine if this is what is happening in your case is to temporarily tack on another electrolytic cap in parallel with the 4u7 that's already there, like 4u7 or 10u.  That will change the feel of the compression, but if it eliminates what you are describing as "distortion" then we've narrowed down what it is.

Aha! That is good to know. I wonder if variability in ESR of various electrolytic caps comes into play here?

It sounds like maybe your bias trim is set incorrectly. On one extreme it should be a cleanish boost and on the other it will kill the signal completely. There's a very narrow band between the two extremes where the volume changes drastically with small tweaks of the trim.

All I get is the same amount of distortion and no change in the sound. I am beginning to wonder if it's because of my 4.7uF caps - it just so happens that these caps are in both aforementioned areas of the circuit. They're some boxy odd things that were given to me a long time ago. I think I'll try changing all those caps tonight and see what happens.

[midwayfair]

I wonder if variability in ESR of various electrolytic caps comes into play here?

No.

The only electrolytic that plays into this scenario, even if this was a concern, is the hold cap, which already has a resistor in parallel.

If your trim pot is not resulting in near silence at one extreme, then something is wrong with your FETs. I think it might be time to post voltages.

The total range of working compression is perhaps 700 Ohms of your 10K trim pot.

[jsleep]

In my experience with the Squeezer, I would first suspect a problem with the FETs. 

JD

[PRR]

The "bias" is CRITICAL.

Much more so than the diode.

One general plan is to insert a *steady* tone at a low level and trim bias for a slight reduction in output. But you may not have a steady tone or be sure how much is "slight reduction".

Here's a very different way(*) to set bias:



Jumper the C1 R2 node to the 9V supply.

Volt-meter the other end of R2.

Trimming the bias pot "should" change this voltage from nearly 9V to very low.

Set it for 8V. This is equivalent to 1dB initial gain drop-- it is "on the edge".

Remove jumper and meter. Try it.

That should work. With extended use you may find that you need a little one way or the other. But it will always be near the point that this node falls-off.

(*) The idea is that the FET does not know the difference audio or DC. (The bias trim has a small difference for DC, true.) We have a steady DC. We have a volt meter which can read "slight reduction" (about 10%). And it happens that a single jumper can insert DC without other problems. So setting it on DC is simple and reasonably accurate.

Also a lack of response to bias trim shows that you have something ELSE wrong. Wildly mis-matched FETs, wrong pin-outs, and the ever-popular wiring errors and solder blobs.

[guitarkill]

Thanks for all the help guys. After spending like 90 minutes on it and being stumped when no kind of FETs worked and I had 0 volts on the gates of the FETs, it turns out that I found a track on the PCB that connects the gate of Q1 to the source of Q2 was missing when I made the PCB. So I made that connection and now it works like it should. 

I also added a pot for an attack control and one for a release control. So far I like it, which is saying something because in general I hate compressors.

[Kipper4]

Well done Guitar kill.
Thanks for the tip to PPR .
Excellent I love it when a pedal comes together.

[guitarkill]

Just thought I'd give a little report about the 1N100 diode I received from a forum member. It actually measured a little higher than the 1N270s or some the 1N34As I have lol.

I read somewhere that the 1N100 diode is supposed to have a forward voltage of 200mV, but that must be on the very low end and not the norm...

[duck_arse]

@ guitar, and OT, what's that bird?