compressor expandor mod from mark hammer

[Ansil]

http://www.geocities.com/austenfantanio//compressorexpandor.htm

i redrew the flatline compressor and added the mod switch from mark to make it an expander also.

[Mark Hammer]

Thanks, guy.

[Ansil]

no problem man..  i love your work and i thought hey why not since it is next on my build list.  

[petemoore]

Does the 10k actually connect to the 220k and the LDR, if so [with the switch the way it is] the both sides of the LDR would be connected ?
 Ok I see IC1a's output drives the LDR when the switch is in the other posiition...the LDR's not being used with the switch in position shown.
 Is that 100k between VR and the 10k a compression level?
 Are there soundclips of this ckt?

[Ansil]

hmm i would like to hear a soundclip. i just read the mod page on marks site. and redrew it so i could inorportae it in my next design

[Mark Hammer]

Actually, there IS an error in the drawing.  The 220k resistor needs to be connected to pin 2 at all times, and the LDR is either placed in parallel with the 220k resistor for compression function, or in series with the 10k resistor for expansion.

Bear in mind that the LDR resistance goes low when the input signal goes high.  When the LDR is placed in parallel with the feedback resistor, this resistance drop reduces the gain in response to signal peaks.  When the LDR is in the path of the resistance to ground from pin 2, the resistance drop in response to peaks *increases* gain.  

Although the use of the LDR for compression is pretty straightforward, the use of the LDR for expansion is a little trickier.  First, although you can allow the feedback (pin 6 to pin 2) resistance to drop to zero, you *can't* allow that to happen for the pin-2-to-ground resistance or else you move into the equivalent of open-loop gain and oscillation.  That means that there absolutely *has* to be some fixed resistance in that path to set the maximum gain.  In this instance, with a feedback resistance of 220k and a resistor to ground of 10k, that means that maximum gain is set to 23, regardless of where the pot is set, or the present status of the LDR.  If you want the maximum gain higher, you can tinker with the values of either the 220k or 10k resistor.  It might even be smart to stick in a 4k7 fixed resistor and 25k trimpot instead of the 10k resistor.  Instead of going directly to pin 2, the 220k resistor goes to one end of the trimpot, the 4k7 goes to the other outside lug, and the wiper of the trimpot is tied to pin 2.  That lets you tweak the relative values of the two gain-setting resistances for maximum gain of 53 ( [220k+25k+4k7] / 4k7 ) or a minimum gain of 8.4 ( [220k+25k+4k7] / 29k7 ), depending on how much of the trimpot resistance is in the feedback path or the path to ground.

Of course how MUCH increase in gain in response to peaks makes for a palatable expandor is another thing.  My gut sense is that people will tolerate much more compression than they will tolerate expansion, so the effect of the LDR on gain will need to be played with.  It may well be that the LDR needs to go in parallel with the pot so that the pot adjusts the impact of the LDR on resistance to ground.  Assume we go with the existing component values and the LDR and pot are paralleled.  If the pot is set to half resistance (50k) and LDR+pot resistance is, say, 45k (assuming the LDR has a resting value of 500k).  Then with no signal input the gain is set to [220k+10k+45k]/[45k+10k] =  5.  If an input peak drives the LDR down to 1k resistance, then the LDR+pot resistance is now equal to 980R, which sets gain at [220k+10k+980R] / [10k+980R] = 21.

That's a pretty big leap in gain, if you ask me.  Probably not that problematic if it is in a *downward* direction, but perhaps irritating if in an upward one.  My gut tells me that you wouldn't really want the gain to change much more than by a factor of 2 or 3.  So, if it is a gain of 10 at rest, then it probably shouldn't go much higher than a gain of 20 or 25 on peaks.  If 5 at rest, then don't aim for much higher than 10-15.

This is why I say that adapting it to be an expandor is a little trickier, because you need to keep the properties of the LDR and all those other fixed and variable resistances in mind.  It's *feasible*, to be sure, or else I wouldn't have suggested it, but it is far from a no-brainer.

Which brings us to the topic of the LDR itself.  In the case of compression, sluggish LDRs are quite acceptable, and perhaps even preferable, since they reduce envelope ripple over the sustain portion of the note.  In the case of expansion, you probably want a faster LDR that will react to peaks and then get out of the way quickly.  I gather that can be offset to some degree by the value of the cap in the diode bridge (currently 100uf), but you still need some help from the LDR properties themselves.  I know some of the Vactrols specialize in fast response times, so these might be productive paths to pursue.

Boy, I sure made it sound easier when I first suggested it, eh?  

[petemoore]

Not too worried about expansion...
 So the 220k and LDR [parallelled] go between pin 2 and IC1A output...what does the 10k/100kpot from ground do then...
 It must be early or something I'm still a little confused...

[Mark Hammer]

The resistance to ground (10k + 100k pot) does what it has always done, namely adjust gain.  In the case of the compressor arrangement, though, it plays the role of adjusting how much gain reduction there is.  If the LDR is at max resistance and the pot is set for 0 ohms, the gain is pretty much at 23.  Once the LDR swings low, that gain drops dramatically, producing a compression effect.  If the pot is set for maximum resistance, the gain is [220k+110k]/110k = 3, which means that changes to the LDR resistance will not drop gain all that much, yielding little audible compression.

Note that this is different from the Orange Squeezer scenario.  Although, by coincidence, it too has a 220k feedback resistor and 10k resistor to ground in the non-inverting op-amp stage, the OS uses input-signal attenuation via a FET, rather than gain reduction  The amount of gain is constant, but the signal that gain is applied to is reduced.  Same net outcome in the final analysis.

[petemoore]

basically the 10k connects to only the left pole of the spdt, on the schematic?

[Mark Hammer]

Ideally, the schematic should be like this:

a) 220k between pin 2 and 6....always

b) one end of 100k pot goes to pin 2, other end to 10k resistor

c) 10k resistor goes to ground...always

d) one end of LDR tied to pin 2....always

e) SPDT switch flips uncommitted end of LDR to either pin 6 (for compression) or to 100k pot/10k resistor junction (for expansion)

If you want to get fancy, use a 3-position on-off-on switch to route the LDR.  That lets the unit function as either a straight preamp (LDR out of circuit), a compressor, or expandor.

Make sense?

BTW, you don't have to depend entirely on the IC1a resistance values to set the amount of compression or expansion.  It takes two to tango here, and the relative gain of IC1b, as determined by the 10k and 47k resistors, is also part of the equation.

[petemoore]

On todays flatline build ... the LED dont light....the legs are tied together dohh, woops better double check which side of the Ge's that light goes on the schematic...might as well draw you a new I think it's ready now...just orient and connect to where the 100uf does.

[petemoore]

Real nice slice Post FF compression. Really diggin' in, sharpened the mud right up.
 Gotta try a diff OA, by itself it oscillates...could be lack of box etc.
 The 100k controls the frequency of the osckllation...it's not terribly obtuse, and seems to disappear completely when driven by another ckt.
 Who was it recently said something like: I leave ckts that work but don't work perfect alone if I like them...[er smg to that efkt]...anyway probably will go with FF for that super slice lead searing into harmonic feedback boost I like very good...
  {haven't TB'd it, so don't know,  but seems like ample output].

[Ansil]

i am working on a pushpull circuit that lets you have three knobs .. one for overall volume.. one for compression and one for distotiron.. so when yo upull that knob up you get a nice booster that is playing sensitive.. but i had to add about four parts. to  get it to do what i wanted it to do.

[Ansil]

finally foudn the shcematic i was working on.. here ya go..

http://www.geocities.com/austenfantanio//compstortion.htm

[petemoore]

You'll still have to move that LED to where the cap connects, it's pointing the right way but as shown...it's legs are connected.

[Ansil]

why. i don't think i am completely awake here..  i am wanting it to go between in the opamp feedback loop and in parallel with the 10k resistor.

the 10k sets the gain. untill the ldr drops below 10k.. am i missing something. here

[Ansil]

i am getting about there can't be a zero gain. that is why i don't ahve the pot turned up maximum ever..  as long as i keep resistance to ground in mind according to what i got otu of marks post i am ok.. i will check and see.