compressor works orange squeezer

[nisios]

Can anyone explain  me how a simple circuit like orange squeezer works? or any other....although i like fet compressors and this one seems really simple.
I got lost arround what seems a current source on the sidechain part. the entiry sidechain is a bit blurry now.
my goal is to make a simple compressor with attack and decay pots......
heres the schematic from www.tonepad.com

[Mark Hammer]

I just finished doing that.  See "compressors" thread.

[nisios]

i saw your post after making this thread.
but......if i understood you right the sustain part of the sidechain is on the 100k 4.7uF network.
but where is the attack part on this one? that what i dont get.
and that current source with the other 4.7uf is still unclear to me.

[Mark Hammer]

In many other sidechain-operated circuits, like the Dr. Q, you would normally see a small-value resistor between the diode and the cap to ground.  The product of the resistor and cap would set the charge-up or "attack" time;  higher resistor value, or higher cap value, or both would mean longer charge-up time.  Longer charge-up time means that the pick attack would be going for a little while before the audio signal level was reduced.  With compressors, the slower the attack time, the more you actually get to hear the pick attack. 
The OS has no series resistor so the cap charges up as fast as it possibly can (what you would get with a zero-ohm resistor).  That gives it an extremely fast attack.  Because it has a fairly quick decay too, that means its principal action is on the initial peaks of notes, making its behaviour a little closer to that of a limiter, rather than a compressor.

[nisios]

thanks very much.
Ill experiment a bit then to see if i can make it variable attack and release.
maybe a fullwave rectifyer too.
il post a schematic then

[nisios]

I have done some reading and came to the conclusion that the attack time gets affected by the release resistor....in other words: as i lower the discharging resistor and as i have to consider the capacitor as a load on a complex RC circuit, i have to calculate the charging equivalent resitance by paralleling the charging effective resistance with the dircharging one.

Is there any way to overcome this problem?

[Mark Hammer]

The 100k resistor only indirectly affects how the 4u7 cap charges up, the same way a leak in a rowboat affects how quickly the rowboat fills up with rain.  The leak, however, does not determine how much water falls from the sky into the rowboat, only the extent to which that rate of rainfall can approach the limit of the boat.

In the case of the OS, there is zero resistance between the current source at the output of the diode, and the input point (positive terminal) of the cap.  So, what you can vary with the parallel resistance (100k in the stock design) is simply how quickly the cap can be drained.  If the resistance is higher then the cap holds stored charge longer.

In theory, you could place a variable resistance between the diode and the cap, in order to achieve variable attack time.  The problem is that this places a limit on the amount of current there is to be stored, and you'll need more "drive" to overcome that.  Although it  is in the context of a filter rather than a compressor, anyone who has attempted to install variable attack time on the DrQ/Quack or Nurse Quacky (which use essentially an identical envelope follower/rectifier configuration) will know that as you increase that series resistance to slow down the charge-up time of the averaging cap, you also have to increase the gain of the stage preceding it or the sensitivity control to provide more signal to that gain stage.  In the absence of that level compensation, what you'll hear in the auto-wah situation is not only slower sweep but less sweep overall.  In the case of a compressor, you'll hear this as less compression, and subtler compression.

You can do the same thing, to some extent, with the OS, by increasing the gain of the op-amp stage.  The default gain is x23, but you can vary that by either reducing the 10k resistance on the ground leg or increasing the 220k feedback resistance, or both.  Do keep in mind, though, that since any gain added to the stage will affect both the envelope signal being rectified AND the audio output, you'll need to turn the volume down.

[nisios]

is there any way of achieving a true independent attack and release even though with a more complex circuit?
becuase i was analyzing a different compressor wich can have variable attack without voltage loss on longer atacks throu the use of a very low impedance source but in this case the release is  dependent of the attack although variable.
the circuit is this one:
http://dt.prohosting.com/hacks/joecheep_schem.gif
this is a very interesting circuit by the way..........the signal path is passive!!!!