Diode Based Compressor(Passive to Active)

Advertise with us
Shop deals on ebay
shop deals on amazon
GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Where you put clip goes the attack, where you put attack changes attack and release, where you put the input, shouldn't be nothing, input should be somehow before DOA1, maybe an H-PAD is the way to go to keep it simple. R17/18 Vs R9 controls the ratio, you could play in order to get different compression ratios at both sides. Threshold could be changed changing the type of diodes or adding diodes in series.

Any of the controls could be controlled independently for the positive or the negative side of the waveform, match everything to get less distortion. Don't expect clean sound from it.

JS

 
joaquins said:
Where you put clip goes the attack, where you put attack changes attack and release, where you put the input, shouldn't be nothing, input should be somehow before DOA1, maybe an H-PAD is the way to go to keep it simple. R17/18 Vs R9 controls the ratio, you could play in order to get different compression ratios at both sides. Threshold could be changed changing the type of diodes or adding diodes in series.

Any of the controls could be controlled independently for the positive or the negative side of the waveform, match everything to get less distortion. Don't expect clean sound from it.

JS
Click to expand...

I changed circuit and IN is better now.Removed clip because timing is not stable when use it.

For now it's better than old version.

Actually yes,it's really dirty compressor but it's good for me.I want something like "bus/mix warmer".About ratio i want only compressor and limiter options but i can't calculate.Is there anyway to see compression knees with ltspice?Or what should i do? I think smaller values are compression and bigger limiter...Right?

 
R6 is attack for both sides, Rel is ok there, the caps would change both times.

when R17=R18 ratio is equal for both, for now on just R17. R17/(R9+R17)=Ratio (in times, V/V, not dB) 20Log of that will give ratio as you expect. R17 should include dynamic resistance of the diode, let's neglect it. Then, R9 much bigger than R17 is limiting, similar values compressing.

I still don't like the level there, still affecting the impedance seen by the device and changing times, for fast attacks you need really low impedance, Split R1 and R2 by half using two of half the value and between them use the level control there. You need good driving capability from DOA1 to get fast attack without bad distortion (current clipping there) so look for it. A lot to play with this simple topology, breadboard it and start playing around!

JS
 
It's almost ducking, I don't remember if I get this far with mine, but as it's feedforward is something that can happen. I told you is a nice one to play with, sounds really nice for some applications, I wouldn't do a mastering on it, but certainly over compressing drums for a parallel processing.

A nice thing to do, adding another opamp, IC for this probably, maybe a TL741, 07X has to low output capability, and from there drive R6. So there you'll have a sidechain insert point, to add a filter or trig from a different signal, though I don't know if it's useful for this comp (the key signal, not the filter) and also add some gain so you'll have an independent control over threshold which may be a good thing for the S/N ratio, but I don't know how much you can take from it since you need to have low signal on the other side where the diodes are compressing to be working as dynamic resistance in a specific operation point rather than traveling around in some non linear fashion. Using a couple of diodes in series will get the threshold up and really help in S/N reduction, by using 2 diodes in series you have twice the level, so keeping all other stuff the same you have 6dB less noise, for 6dB less you need 4 diodes, and probably won't get much better than that before run out of headroom in your feeding amp. Maybe with Ge diodes you get a softer knees, with LEDs maybe harder knee, something to experiment, also would change the threshold as expected.

JS
 
I don't understand why did you do that, driving the output transformer from a TL072 can't be better than a decent DOA, poor driving capability, if you get long cable or low impedance low you'll probably have a problem. The opamp I purpose to add was in the place of R6, as buffer, so you can leave your level pot as it is and play with it's gain to control the threshold.
You still didn't change the attack time to R6 instead of caps and still have the input level pot before the diodes. Try measure your time constants: attack when changing the input level (max and -6dB will show max change) and release when you switch the caps for your attack switch. I think both will change, when they shouldn't.

JS
 
New Samples

Fixed reduction with 1kHz Sine Test Tone to -6 dB.

Fixed Attack

Unprocessed
https://dl.dropboxusercontent.com/u/91809016/Diode%20Comp%20Sample/NEW.wav

Fast Recovery
https://dl.dropboxusercontent.com/u/91809016/Diode%20Comp%20Sample/Fast%20Recovery.wav

Medium Recovery
https://dl.dropboxusercontent.com/u/91809016/Diode%20Comp%20Sample/Medium%20Recovery.wav

Slow Recovery
https://dl.dropboxusercontent.com/u/91809016/Diode%20Comp%20Sample/Slow%20Recovery.wav

My design is done! It's working perfectly! But...

I need idea about GR meter.I have no meter now and i can't try.Where should i add meter to diode compressors?

dynamic-compressor-schematic.gif
 
I don't see the schematics...

There is no simple way to do that, and this is a simple compressor which shouldn't be looking for that. Between P1 and R4 you will have some voltage changing in some function of compression, but I wouldn't use that more than for knowing it's compressing or not, maybe one led called 'activity' or something like that. Some math needed but using those voltages (I'd buffer them in any case to not affect the compression functionality) and two more diodes and resistors, as R5, R6, D3 and D4 and R3 to form a divider from a reference voltage will act as the compressor divider which will make a decent DC for GR meter, but 3 opamps needed for that, change the ref to set zero and gain of the buffer to track. One to buffer the voltage, one to invert it and another to drive the vumeter which would be needed in any case and already there for the signal vumeter. R3, R5 and R6 should be equal than the ones in the vumeter section, which will make a PIA to work with if changing those for ratio.
  Another way is to compare input and output level, but not as directly as it appears to be I guess. I'd do rectifying for both in a place input and output gain don't change the relation between them, would be after input control and before makeup gain for each one, once rectified when no compression should be the same voltage in both, except for the attenuation which should be small if loading the circuit with high impedance, then the difference will show the compression in a linear scale, not really useful since XdB at some level would differ XdB at some other level and in other configuration. To make it work you need a divider, which isn't easy or two log converters and then the subtraction, it isn't funny either.

So, if I would go for a decent meter I would use the other set of diodes to make the other divider for DC. Maybe could use the opamps inside the diode arrange and use them to buffer the CV inside, then the vumeter only needs the divider. In any case you have 3 opamps used for this.

JS
 
I added circuit.And i know it's not "normal" compressor but i liked it!

You can see circuit and tell me something about GR meter,i guess.
 
Well, now you don't have input control, so your threshold is fixed and depending on the level at the input. I insist for last time in using R6 as attack, since changing caps would also change release time. What you added is just a input/output level meter, not GR. Take the voltage across release pot+R10 with a differential opamp, then invert it, attach two diodes between them with two resistors in series, as R18, D3, D4 and R8 (I'd use R9 to change ratio instead of R18 and R8 so you always keep tracking in the GR and easier for everybody) Then use a dual pot as R9 in order to do the GR meter track properly when ratio changes, replicate the circuit, D3, D4, R8, R9 and R18 and put this new R9 (let's call it R9', and all of them) so now you have a voltage to send to the meter, it will go to the in/out meter as you used. change the reference voltage applied at R9 to set the zero on the meter, change the gain that's sensing the voltage across R10+release pot to get proper tracking. This amp is differential but no need to be balanced, just make sure that the input impedance is high enough, so no affecting the release time or nothing else. I'd go for maybe 1MΩ input resistors and then play with the feedback to get the proper gain. Other way would be to use an input stage (the first two opamps) of something like an instrumentation amplifier, to get really high input impedance. which may be a good option, although, if you need to get lower gain, for proper tracking, at this stage you won't be able in a simple way, you may need to tweak R8', R18' and a new resistor across them, to make an equivalent value for the diodes to be seen but attenuated. (higher R8' and R18' with a resistor across them, in parallel with the diodes) to get this proper tracking.

JS
 
Thanks for explanation!

Yes,using R9 to reduction is better idea.I'll change like you say.

About GR meter...I think i understand you.I'll replicate compression circuit to feed meter.But what about attack and release?Should i use dual pot and 4pole switch to control attack and release of meter?

EDIT

Actually no...If i use R9,everything changes.Using other two resistors is better.R9 works like "blend".
 
R9 shouldn't work like blend, maybe you are off in the value range for it. Take the voltage between D1/C8 node and D2/C4 node. I thought those two opamps at the end were for I/O meter, no need to do the buffer, just add impedance balanced output and go for the next, use those opamps for other task, like the meter! Also keeping your path as simple as possible is good for you hearing experience.  :eek:

So, no problem with attack and release because you are taking the reference after they are acting, so no problem with that, the only problem still is the ratio, you should try R9 again and see why it acts as blending and not ratio (sometimes is similar) You should keep values high enough so dynamic impedance of the diode is used in a pretty small region. Let's say 1M pot for R9 with maybe 100K series for 1 to 10 ratio variation, R8 and R18 probably about 10K. It was acting as limiter in the original circuit, so maybe more than 10K can go there, they that out, Maybe 22k as the original is fine, maybe 47k is also fine and your max ratio will be the original. You need some input level control, add an H pad instead of R1/R2. if your level is too high diodes will not work dynamically but will start to work in an un-linear way, no bueno.

JS
 
Polarity of C4 and C8 in the schematic, in the first post, look to be backwards. I would reverse them.

The AA119 diodes used in the original schematic are germanium type, which will have a turn-on voltage that is about half that of the 1N4148 diodes that you are using. They also have a longer knee range. These two specs will significantly change how the compressor reacts.

Best regards, Jack


 
he AA119 diodes used in the original schematic are germanium type, which will have a turn-on voltage that is about half that of the 1N4148 diodes that you are using. They also have a longer knee range. These two specs will significantly change how the compressor reacts.
Click to expand...

What about 1N34A? When i use 1N34A,it changes attack & release.Faster responses...It's easy to find germanium diode for me.

Now i don't have AA119 but it's easy to find,too.I can't find spice model for AA119.I want to try with spice.
 

Similar threads

jBam
Diode resistance question...
Replies
6
Views
625
jBam
jBam
K
Compressor issues
Replies
3
Views
530
MagnetoSound
MagnetoSound
Z
Tube base extender
Replies
19
Views
1K
Whoops
Whoops
OneRoomStudio
One-Tube Opto Comp
2 3
Replies
44
Views
4K
OneRoomStudio
OneRoomStudio
E
power supply questions for Gefell MV202
Replies
0
Views
355
erikx
E

Latest posts

Back
Top