Orange Squeezer - compression ratio sucks?

[blackieNYC]

I built an orange squeezer from the madbean cupcake schematic.  I'm getting a maximum compression ratio of 2:1, and that ain't right.  And I'm tweaking the bias away from distortion such that I have to have the output volume up pretty high.  Still distorts on chords.
But please check my testing for me.  It doesn't sound like it is compressing, and a scope and oscillator reveal the same.
With a 100mv increase in input, I get a 60mv increase in output. So it is compressing, but barely.  I'm feeding 200mv in, then adjusting the trimmer so the tips of the waveform don't clip. Measure the output, then I back down to 100mv and take another output reading.
I've tried a variety of MPF102s and 5457s in Q1 and Q2.


I'm citing RMS AC voltages, not p-p.  200mv RMS is about 500mVp-p, a little hotter than average gtr. I don't think that's affecting things.
voltages on Q1   D=1.3v   S=1.3v   G=~0v
Q2   D=9.6v   S=2.1v   G=1.3v

I have 5.2v on the op amp pins 1,2,3, ground on 4, +9.6 on 8, 5 is ground, 6&7 are shorted.

[GGBB]

Without compression occurring, there is enough gain in the circuit to clip 500mVp-p. If you are adjusting the trimmer to the point where things are just out of distortion, that may not be enough. I believe going farther in the same direction with the trimmer adjustment should give you more compression.

[PRR]

The output stage has gain of 23. If you put 0.2V in, it "should" make 12V out. It doesn't. So it IS compressing. Some output numbers would help. But assuming you get 0.2V out, it is already cutting-back 60:1. Which is a lot.

I would also ask what the voltage is on the 4.7u cap at the diode. This point should rise and fall with input level.

[blackieNYC]

GGBB - thank you.  I have turned that trim down until I have to crank up the output volume a lot.  At that point, the compression is just absent.  It seems that the compression happens in the distorted (trimmer)range, and the clean range has no compression.  I hoped to find the trimmer's sweet spot (that you may have mentioned in a thread) but once I get rid of the farty distortion the compression has long since been tweaked away. 
IIRC - you have actually given up on the OS altogether, or am I thinking of some one else?  You came with one, no?  I will try it - but for this purpose I'm looking just to stick a set-n-forget compressor at the front end of another circuit.

Paul - one FET pair gave me this typical outcome: 200mv in gives me 145mv out, and 100mv in gives me 70mv out.  input change of 100, output change of 70, worse than 2:1.  Am I doing that right?
I'll get that diode voltage for you tomorrow. Thanks much.

[DavidRavenMoon]

What if you pad the input and then crank the trimmer up? Modern pickups are a lot louder than when that circuit was designed. 


Sent from my iPhone using Tapatalk

[blackieNYC]

Found a solder glob!
These side-chain builds can really throw me.  If you have something wrong in a distortion you'll probably know it, but with a mistake in a side chain circuit you can be misled by slightly odd behavior.
The squeezer sounds good when tweaked, but it is awfully touchy To adjust the trimpot. I thought the 500k pot to ground on the input was helpful, but really it turns out to work just like the trimmer. They seem redundant - Turn it up and it gets loud and distorted, turn it down and it gets quiet and less co pressed.  With a variety of guitars, it seems like one of these pots will have to be external. This build is just to add sustain on the front end of a gator or slow gear. I don't really want to be playing with input trims and output trims to find a sweet spot.  Hard to believe the OS was originally a no-knob effect. I'll play with it and see if I like it, it if I need a dynacomp.

[blackieNYC]

Here's a question I've thought about with a couple compressor designs.  I do get a bit of pumping when ringing out a chord. It's the subharmonic beating frequencies - the speed of the change in compression changes with every chord. Apparently the beat frequencies have enough amplitude to dominate the compression level.

These are low frequencies - how can I filter them out of the side chain, so they are ignored, but leave the audio path unfiltered?  I suppose it would be right after the split point.

Mark Hammer has brought this up - there are frequencies that we don't want affecting the threshold, but we do want them to pass thru the audio path. Like a gate perhaps, where you could try getting the side chain to only attenuate hissy highs and/or buzzy lows.

[mgsrake]

Dear, black
i've built on of them and sounds pretty.
I had to do a trick on the Comp pot because it's started sounds only at half of the scheme value.
It was discussed on guitarfx forum by user mirosol

http://tagboardeffects.blogspot.it/2012/02/orange-squeezer.html?showComment=1339528517621#c3998240838034615644

[Mark Hammer]

Just for the hell of it, try splitting the op-amp output so that the audio output goes through the 4u7 and 10k volume pot, but the sidechain leaves pin 1 via its own cap.  You could select the cap value to roll off low end and keep the nasty sub-harmonic wiggles to a blessed minimum.

I'm accustomed to looking at other drawings of the circuit.  So when I look at the one you posted,  My first impression is that the audio output has a sort of clipping circuit hanging off it, in the form of the rectifier.  I'm naively guessing that isolating the rectifier from the audio output by giving it a separate DC-blocked output from the op-amp may clear some things up.

Finally, keep in mind that the FET and 82k resistor form a virtual voltage divider (attenuator).  Making the 82k resistor a bit larger (e.g., 100k) will increase the attenuation, and conceivably the audible compression.

[GGBB]

Finally, keep in mind that the FET and 82k resistor form a virtual voltage divider (attenuator).  Making the 82k resistor a bit larger (e.g., 100k) will increase the attenuation, and conceivably the audible compression.

My experience with the Orange Smoothie suggest that making the input resistor larger will decrease compression for two reasons: the FET is affecting a smaller portion of the signal; and the input signal hitting the opamp is lower thereby triggering less compression by the FET. And I'm not sure whether the former is true, since all that really matters is the effect the FET has on the signal hitting the opamp. In the smoothie variant, the input resistor is not varied, but instead the FET connection is moved along the resistor. In this case, moving the FET toward the opamp end of the input resistor lowers compression. But I could be wrong about this - I haven't made any measurements or used a scope to analyze the signal - I am just going by how the compression knob works in the smoothie, and it is somewhat on the subtle side.

The smoothie also has the bandpass feature which does have a notable effect - I always use it. I've also found that lower sustain helps to lessen the pumping effect.

[blackieNYC]

All done. Maybe I can save somebody a little trouble: as driven by the rectified audio signal,when the voltage on the diode jumps up to about 120mV DC, (no more, no less) I'm getting the max compression. It is quite a sweet spot - the trimmer measures 3.6 k in circuit. And having found this sweet spot, there doesn't seem to be much of a pumping problem. Maybe that's my imagination - while I didn't get much benefit from splitting the output as Mark suggest above, I think that's still worth trying. In hindsight (that unique perspective one gets when their head is in their hind) I'll use this for the specific purpose I have in mind, but I think other compressor designs might be a little more efficient and less demanding. I could probably replace the 10k pot with a 5k resistor and 5k pot, but you just don't know that going in. A multitude pot might be helpful, but they have their own frustrations, too - the lack of visual indication can be a little challenging.
Odd - I don't think I need to bother with an input pot. Once I found a setting I liked, I tried my high output Seymour Duncan HBs, some 1966 vintage single coils, and a 10 dB booster. The distortion is really a product of the trimmer setting much more than of the input signal. Which is great. All stock values, and 5457s.