Featherpillow's Simple VCA Compressor

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

featherpillow

Well-known member
Joined
May 14, 2005
Messages
214
Location
USA
Okay, so over the weekend, I had the chance to do some prototyping of a simple VCA compressor. I wanted to use parts I had laying around, and keep the cost low. I'm certainly not an expert in circuit design, but I do know a bit about VCA compression, so I thought I'd give it a shot. Plus, THAT has very good datasheets.

It's nothing fancy, it's a feedback design (I like the way these sound in comparison to feedforward), and it uses a 22M resistor tied between pins 1 and 7 of the 2252 to compensate for errors. In terms of control, it only has Ratio (1:1-1:infinity) and Makeup Gain (-20dB to 20dB), although it's easily expanded. I used an OP275 in the CV signal chain, since they're low noise/low distortion, although one could easily sub a 5532 or TL072.

It uses a 2N3819 for an input buffer, and, instead of using the traditional "opamp as a trans-impedance stage" for the VCA, it uses a FET/PNP Sziklai follower. It's designed with assuming an Idss of roughly 5ma, since I have tons of 3819's with roughly that Idss.

Overall, I think it's a pretty nice and simple comp, it's a soft-knee design (see D2), but it's easily changed to something faster if needed. It doesn't distort or pump at all, due to the values used for timing and bias on the 2252. It probably won't out-do an SSL clone.

I was able to build it on a proto-board. It's designed for use with a +/-15vdc supply because I have plenty of those, although it's easily adapted for use with +/-24vdc supplies.

Anyway, if anyone else is interested, the schematic is here
 
Is it ~easy to make attack and release adjustable?
It's not terribly difficult. If I've got some spare time this week, I'll post a revised schem with attack and release controls.

How does it sound?

It's nice--I like it better than the traditional opamp/VCA/opamp topology.
Of the different compression methods out there, VCA isn't my overall favorite, but I find it useful. It sounds better than any of the DBX stuff I've used and thus far I like it more than the RNC. The Sziklai follower is something I enjoy toying with because it's simple and fairly new to me.

The level detect obviously shapes the overall sound alot. A poor choice of component values in the LD circuit can cause distortion in the CV signal, which then gets multiplied in the VCA's internal gain cell, and shows up as noise or high-order distortion in the audio signal.

Also, I've never been a fan of compressor pumping (I know some people like it for effect), so this one won't do it. The LD is biassed and timed to have a release rate of roughly 120dB/second, because that introduces the least amount of error into the CV signal. Also, that 22M feedback resistor helps correct any additional errors, but it does it at the expense of low-level accuracy.

It seems to handle low frequencies pretty well, which I would consider to be at least one benchmark of compressor performance.

I haven't tried it on drums, or across a buss yet.
 
I think I have a few of those chips around... i might be able to build one and run it up on the Audio Precision... The FET path intrigues me... that's not something that you usually see, rather a couple of hastily tossed-in op-amps instead.

Keith
 
Thanks!

It's really simple, nothing fancy, no tricky absolute value circuit or esoteric VCA. I kind of feel like a beginner guitar player showing off his ability to crudely play "Smoke On The Water," to a group of professional players.

I am somewhat interested in this FET/Opamp VCA design I stumbled across on the Electronic Design site. It, too is nothing new or fancy. It just intrigues me.

Figure_01.gif
 
I really like the feedback approach. I've modded some of my compressors for feedback instead of feed forward. Sounds much more musical to me, especially on Guitar and Bass.
 
Looks very interesting!

Thanks Jakob!

As promised, here's a revised version that includes threshold, attack, and release controls. It's untested as of yet.

*edit* schematic has been edited to include blocking diode after U4A, and omits the 1M input resistor to U4B. The added controls are still untested.
 
A couple of things:

1. You will need a blocking diode on the output of U3a, otherwise, the release C will discharge through the input of the opamp, not throught the resistor string including the release control.

2. R34 is marked as 1M, and this is too high for the input to U3B. 1K would be better, though none is really needed.

3. Now that you've added timing circuitry after the "makeup" gain pot, you could potentially have a static voltage on the timing caps. I'm not sure what the result would be; you might want to breadboard and check it. You could add additional circuitry after U3B and connect R20 to it, but this may add unwanted complexity. However, it will enable you to meter the gain reduction on the output of U3B, and will keep 0 voltage on the timing caps during times of no compression

4. FYI, the overal circuit now closely resembles the basis for a multiband compressor I designed about 5 years ago, and have sold to several radio stations. It works very well.

Nice job!
 
You will need a blocking diode on the output of U3a
Thanks for the suggestions--this hasn't been tested yet with the added controls. Also, I notice now, looking at the schem, that I've got two "U3" opamps--that second one, with the release and attack controls--should be U4.

I'll prototype it with your suggestions in mind, and leave out R34. You may be right about the standing voltage, and I may have to move the makeup gain pot elsewhere in the circuit.
 
Hi JDiamantis,


Maybe a little off topic, but still...


I'm looking at multiband comp designs for a while, I want to design one to use in an in-ear finalizer (see http://www.groupdiy.com/index.php?topic=2892&highlight=finalizer )


Would you be willing to share some info about your approach in multiband comp design?

Following questions come to mind:

-Would you use different attack, release and ratio settings for each band of a three-band design? I guess att and rel could be slower for low band and faster for high band compared to mid band.
Any suggestions regarding time constants?

-Would you use one set of attack, release and ratio controls (three-deck potmeters) offsetting this parameters for all bands , or rather an attack, release and ratio control per band?

-What cross-over frequencies would you use? Would you make them user adjustable?

Note that this questions are relating to me using the multi-band compressor on full mixes (mastering), not on separate channels or subgroups.


Thanks in advance!


When other people are interesting in this matter, maybe we can start a new thread regarding multi-band comps.


Best regards,


Rogy
 
Below a copy of my reaction on your explanation in the "Featherpillow simple VCA compressor" thread:


I'm looking at multiband comp designs for a while, I want to design one to use in an in-ear finalizer (see http://www.groupdiy.com/index.php?topic=2892&highlight=finalizer )


Would you be willing to share some info about your approach in multiband comp design?

Following questions come to mind:

-Would you use different attack, release and ratio settings for each band of a three-band design? I guess att and rel could be slower for low band and faster for high band compared to mid band.
Any suggestions regarding time constants?

In the particular design discussed in the thread, I used the dbx 4301 chip that contains a VCA, 3 opamps, and an RMS detector. I set up the structure as a "feedforward design". I picked different time constants for the RMS detectors, each of which was followed by an averaging section, whose time constants were identical for each band. I did not offer an attack control, as I decided to use the RMS detector as the controlling element, and it's tough to use this as designed and provide an attack control.

-Would you use one set of attack, release and ratio controls (three-deck potmeters) offsetting this parameters for all bands , or rather an attack, release and ratio control per band?

I used one set of release times for all three bands, adjustable via a pot on the front panel. I've experiment with different release times for each band, but usually this was for use in a multiband LIMITER. For agc use, identical release for all bands sounds best to me. Ratio was changed simultaneously for all bands, via a rotary switch: 2:1, 4:1, 6:1, 10:1. There were adjustable output controls for the low and the high bands: ~+/- 10dB.

-What cross-over frequencies would you use? Would you make them user adjustable?

I used 200Hz for the low band, and 2.5kHz for the high band. For use in mixes and mastering, you might choose other freq's or allow for user adjustment. AS this unit is used in a broadcast environment, the fewer user adjustable controls, the better. I used a "derived" or differential crossover, where the mid-band is derived from a summed input of the wideband signal, positive polarity, plus the low and high band signals in opposite polarity. The result is a crossover, when resummed, that is phase coherent, and flat in frequency response. Similar to the one used in the dbx 3bx, Aphex Dominator, and some older National Semiconductor application notes. Response slope is 6dB/octave.

-Note that this questions are relating to me using the multi-band compressor on full mixes (mastering), not on separate channels or subgroups.


-Thanks in advance!

I'm currently prototyping another design, using feedback detection, and an Analog Devices VCA. I should know soon how this unit fares.

John Diamantis


-When other people are interesting in this matter, maybe we can start a new thread regarding multi-band comps.


Best regards,


Rogy[/quote]
 
I'm really curious if any of you still have the schematic because it seems like a very interesting project!

Thanks and regards,

Tom
 

Latest posts

Back
Top