Suggest me a compressor project - a bit special perhaps

[hhelland]

Hi all,

I've been wanting to build a compressor. Not one of those "really nice" "can't even hear it's compressed!" designs, but rather one that I can use as an effect. I am a bass player, and I like how Tony Levin uses compression as an effect. In fact, I like everything TL does.  Anyway, one thing he does is compress the signal so it gives a nice click at the attack of the note, then has a fat round sustain. I've been able to get quite close to that sound experimenting with the Carl Martin Compressor/Limiter pedal, but I find it hard to understand it - I much prefer the standard controls of Attack, Decay, Ratio, Treshold, Gain..... the Carl Martin has something else (I don't even remember what it's called). Anyway, do you understand what I'm after? Something that can radically rearrange the dynamic envelope of my signal.

So, it'd be nice if someone has built something like this before.... I'm not much of an engineer, alas.... I imagine it'd need to be a feed-forward design of some sort. It'd have to work well with the bass guitar of course.

Any suggestions?

I remember finding such a project once, but I didn't bookmark it and don't know where it was.....

Thanks in advance.

[MartyMart]

http://dt.prohosting.com/hacks/what.html

Try this .....
There's also a "lite" version around here somewhere !

MM.

[petemoore]

   Ross Compressor.
 Can be set to sound 'clunkier' [where the comp amount can be heard 'moving around' ], and slow enough to let a bit of sharper attacks slip through making it a dynamic sounder. There are many mods [dealing with attack/release and voicing, clean blend also] discussed in recent thread.
 Perhaps someone with more 'senior' understanding will make suggestions.

[Mark Hammer]

A good place to start would be o check out some of the on-line articles on compression and limiting.  I'm not suggesting this to imply that you need to go back to basics.  Rather, I've found that often such articles will say that such-and-such a compressor is great for drums, or bass, or voice, or guitar, etc., for such and such a reason.  Producers seem to have their preferences for source/compressor pairings, and their preferences might suggest some paths to pursue.  Not so much for brands and such, but for processes to keep an eye out for.  I belileve there is at least one excellent article like this in the archives kept by Electronic Musician magazine (from early 2001 if memory serves).  I imagine many other recording/guitar/bass oriented mags have similar articles.

My own gut impression is that you likely want something with an LDR as part of the formula.  For several reasons.  One is certainly that LDRs tend to average out envelope ripple in the sidechain over and above whatever the rectifier circuit does, and such ripple can be a real annoying issue with bass compression.  Another reason is that the "click-plus-ultra-squish" you seem to allude to is aided by having a brief but noticeable attack time, followed by a long decay/recovery time; like a turtle that jerks its head back into its shell and seems to take ages for the head to emerge again.  LDRs frequently do not respond as quickly as FETs or OTAs, so the sluggishness that I suspect is critical to your target tone is kind of built in.  Finally, LDRs have essentially infinite clean headroom, which can be essential to processing bass.

Much like the way distortions pedals are designed to provide a certain degree of clipping based on anticipated input level, compressors are designed to provide X amount of gain/level reduction for Y amount of input signal.  The ridiculously obvious compression that some folks like may not be achievable from stock pedals unless there is either a very hot signal fed in or the gain of the envelope follower in the sidechain is cranked so that much more drastic degrees of gain reduction are attainable.  That's easily done, though.  My point is simply to remind you that what you want may well be available in a number of stock designs, with a minor change here or there.

(Post-edit)

It may also be that the tone you want comes from a dual or multi-band compressor, with light compression for the highs and radical volume-suck for the bottom.  That, too, can be achieved with changes to some standard circuits, but obviously those changes will result in a more complex circuit that ain't gonna fit in a 1590B!

[hhelland]

Thanks a lot so far: the What compressor looks promising. I'll probably breadboard one. One question about the What though: is there any particular reason why I shouldn't replace the MC3317B dual op-amp in the input section with for example an NE5532? (I've never used the MC3317 and don't have any lying around.... isn't it designed for single supply?) And, Is there some magic smoke in the VTL5C9 LDR/LED combo, or can I expect decent results from a DIY LDR/LED doohickey?

Thanks again.

[Processaurus]

http://www.elixant.com/~stompbox/smfforum/index.php?topic=40292.0

This is a ross compressor blend control I did for a friends X-mas present.  It works, I don't mind saying so, really well, in terms of getting different attack sounds from the ross/ dynacomp style compressor.  In combination with the fast recovery mod Mark Hammer has suggested, you can get a real interesting "pop" to accentuate a notes attack, just like you're talking about.  There seems to be a lot of variation possible in attack envelopes with the controls (sustain, blend, and the recovery switch).  The ross comp could use a bigger output cap (.47uF, maybe), if you were to go that route.  Maybe a bigger input cap too.

[hhelland]

I just assembled a breadboard What - without the LDR - and it seems promising so far.

But, the LED actually flickered quite a bit when I hit the low B on my bass. It'll be interesting to see if this gets smoothed out by the LDR.

This version is made only with TL072 dual op-amps: I discovered that I've no NE5532 chips left. If, or when, I veroboard it, I'll probably give the I/O section sockets for the op-amps so that I may try different ones. Looks like ELFA doesn't carry the MC3317B mentioned in the schematics.

Tomorrow I'll see if I can find an LDR in my drawers and make a DIY VTL5C9 thing. I imagine it should work all right.

[toneless]

Ldr's will do the job well . Another way is to try a simple compressor like "orange juice" and combine it with another fx -like an envelope fiter- in the same circuit board!

[hhelland]

So I found an old LDR, and put it together with an orange LED in a black rubber tube. Works all right.

But I found it was too "nice", so I removed R28 (the one that blends a bit of dry signal with the compressed signal) and I changed R25 from 680 to 27k, and I changed R41 from 1k to 330. Now it sounds pretty much how I want it. I may still do some tweaking to optimize it for my taste, but so far it is showing great promise indeed.

I just placed an order with ELFA for some parts including a small 2.3VA transformer - so it'll have a proper bipolar +-15V power supply, hopefully inside a Hammond 1590BB box.

[MR COFFEE]

Hi all,

On the choice of the VTL5C9 Vactrol...

Dave (Dee T) and I corresponded quite a bit several years ago due to our mutual interest in compressor circuits, just before he published the What circuit. The What design is optimized for April's vocals (his wife).

He chose the 'C9 because it has a very quick decay time (35ms IIRC), which he wanted because he was using electronic filtering to emulate the quick recovery-long recovery time characteristics of the ear. Most CdS cells aren't that quick, and he felt he needed the quick recovery time.

You-know-who over at Ma*ley says her chief engineer tried every vactrol made in one of their highly respected optical (recording) compressors - and they both picked the VTL5C2 as the "best-sounding." FWIW.

Compressors can be very simple or very complex and sophisticated: they can be specialized to work with a particular instrument\sound source (like drums, guitar, bass, etc.), or they can be designed to be a general-purpose device. Getting the sound you want may or may not require a sophisticated and specialized design. Depends on what you are going for and how picky you are about your sound.

To hhelland -

If you are happy with your design, don't do anything except keep on tweaking.

However, your description of LED flashing and your decision to remove the April resistor to get it to sound good to you strongly suggests to me that your inability to get a good sound out of the What design was due to a mismatch of your signal source level to the expected input level of the What circuit. Increasing the gain of the input buffer amp to match the lower signal level from your bass, plus changing your randomly-chosen photocell to the originally-selected model and replacing the April resistor will help significantly in getting the What to sound like it is supposed to. Which is quite good, particularly on vocals.

Or put the What in your amp's effects loop if you have one. And run the gain up to line level.

I agree with Mark H. that distortion from the gain control element is an issue in bass compressors. Some people actually want it to sound kinda wooly, other people don't like it. Faster cells, shorter sidechain time-constants, and less filtering increase the live dynamic feel... AND the distortion (which is *not* to the degree it sounds like a bass overdrive or anything ever close - it's pretty tame stuff, but it's distortion and you *can* hear it). But also keep in mind that some of the most respected opto-compressors don't even filter the rectified envelope electronically, and let the photocell characteristics do it all.

The choice of photocell and the amount of lowpass filtering in the sidechain has a powerful effect on the sound of a compressor - especially when you want to *hear* the sound of the compressor, as in your case, where you are using it for an effect. In my experience, most generic grab-bag photocells are floor sweepings and vary all over the place on all parameters. If you can't get the sound you want after tweaking your design, try selecting another photocell - and expect that you will also need to tweak all the other parameters (signal level, sidechain gain, sidechain filtering, etc.) to see what the new one can do.

[hhelland]

Thanks a lot Mr Coffee! Enlightening stuff.

It seems that ELFA (my electronics supplier) isn't carrying the VTL5C9, so at least for now I'll have to settle with what I've got. It'd be interesting to try it to see if there's any difference.... perhaps some other time.

It did occur to me that the What design was intended for higher input signal levels, but so far it seems to work quite well fed directly from my (passive) bass guitar. I do intend, however, at some time, to build a proper pre-amp for the bass; which may or may not change the behaviour of the compressor. In fact, perhaps I ought to do the preamp before I finish the compressor....

It has been very interesting to experiment with the What design though; and I'll keep at it for another while before I build it into a box.

For example, I'm thinking of putting a switch to select between two different values of R41 - it'll be a bit like a Ration switch, I expect.....


Thanks again

[hhelland]

Last night I tried to up the gain in the preamp a little bit, and guess what? It did make a difference.  Thanks for the suggestion. I could then un-do some of the changes I did earlier.
I think I'll fit a little jumper thing on the PCB so that I can change the preamp gain later, if needed (if for example I'll use it in an FX loop or something).

Thanks again mr Coffee!

[MR COFFEE]

Hi hhelland,

Thought that might make a difference.

The What sidechain needs a fairly good signal level to do more than briefly flash the LED on peaks. Of course, CdS cells are pretty forgiving. That's how some very well thought of opto-coupler designs get away without filtering the sidechain signal at all. Like the LA-4.

I didn't mention... I assumed, I should say, ... that you knew to change the input amplifier to non-inverting and therefore set the input impedance to the range a passive bass pickup expects to see. High-impedance guitar pickups should have a load of 250K or greater (1 Meg is pretty common), and the input impedance of the What is only 10K (since it is an inverting configuration with a 10K input resistor). Just in case...

Since you liked the sound you were getting with the lower input level where the photocell was doing most of the work of generating the recovery gain swell, you might want to try tweaking a simpler design like the DOD280A. Circuits don't have to be super-complicated to sound good, especially when you are *wanting* an effected sound. Tailoring the DOD circuit to your particular photocell, your bass and pickups, your playing style, and your desired final "compressed bass" sound - can yield quite impressive results.

You may want to try increasing the threshold on the DOD another diode drop (just a suggestion based on what it sounds like you like in bass compression sounds), and what I am guessing the photocell you are using acts like from your description. But let your ears tell you what is the right circuit design and values for the sound *you* want.

And it is a real plus that the DOD circuit will run off a 9 volt supply

The What is a great circuit design tailored for when you want a compressor that sounds like it isn't there, rather than a circuit design tailored for when you want a compressor that sounds like an effect. It's designed for only about 10 -13 db of compression; but you can push it to get a lot more compression than that if you want to. The sidechain may benefit quite a bit from a lot of tweaking if you are aiming for a 30 db compression range 'effect'-style compressor. But as an effect, you may well find that the modified What doesn't sound better to you than the simpler DOD circuit tweaked to your preferred sound, or other circuits designed more for effect use. FWIW. YMMV.

Happy tweakin'

[hhelland]

Thanks again mr Coffee!

I'm afraid I didn't think about imput impedance at all, I just swapped the resistor in the feedback loop (R2) to a bigger one, giving me about 4x amplification. It makes good sense to redo it to a non-inverting version though, as you suggest. I'll try that tonight.

The What compressor design did seem a bit sofisticated for my needs, yes; I've been thinking a bit about simplifying the sidechain somewhat. The dual release time, for example, may not be too important for my needs  . Perhaps a simple rectifier circuit will do just as nice.
However, the sidechain high pass filter is an interesting idea for bass guitar use. I'll do a bit of experimenting around that, too; perhaps try bypassing it and hear how that sounds.


I had a brief look at the DOD 280 schematic. I might try breadboarding it - but I see that it's a feedback design: the gain reduction is controlled by the output signal (rather than the input signal) - so I'm not sure it will be able to give all that radical compression effect.

Perhaps the DOD can be modified to feed-forward....


thanks again

[Johan]

..may I be so bold as to suggest...
http://aronnelson.com/gallery/johan/LA_Light_last_one_001
..look at the sidechain-filter...might be what you're looking for..the VU-section and the biasing op-amps arent really needed if you replace the biasing opamp with regular stompbox-type voltagedivider and cap. and you can use one of the LEDs through chassis for indicator..works with 9volts too..just put dc-blocking caps on input/output..

johan

[MR COFFEE]

Hi Johan,

I'd vote for your design over the DOD for hhelland's application. I remember a previous version of the LA light - perhaps by you or someone else - that didn't use the input divider trick to give a choice of limiting versus compression. This looks like a much improved design over the one I saw some time back.

NIce job!

hhelland,

The highpass filter does a lot of what the fancier 3 db per octave rise filter does in the What. You might want to try it first and see if it does what you want. The 3 db rise is nice though - I've used a similar type of filter on some instrument compressors. I'd expect the hpf cuts out a lot of the compressor distortions on bass notes, too.

I'd still hang a high impedance input stage on there for direct instrument input. Even 100K is too low for passive pickups to my way of thinking.

Have fun!

[hhelland]

Hi again,

I tried breadbording the DOD 280 last night, it didn't sound nearly as good as the (modified) What design.  But then again, I didn't really bother tweaking it much.

The LA-Light seems interesting, perhaps I'll try breadboarding that one tonight.

I got the tiny 2.3VA transformer in the mail today though, so it's no problem to build a circuit with +-15VDC supply.

Once I make up my mind on what to build (so far, modified What is most likely) I'll start thinking about how to cram it into a box. I also have some more 3PDT foot switches on the way.

What a life. 

[hhelland]

Well, at last I have some progress to share. I've built one channel of the What compressor circuit, including some simple modifications, into a Hammond BB size box, including a tiny power supply. And, here I surprised myself: It worked right away! amazing. Sounds quite good, if I may say so myself. The only thing I'm not totally satisfied with (besides the look - it was an old box I had used for a different project earlier) is that there is a very slight hum - which I am afraid is from the proximity of the transformer, although I took care to place it as far away from the signal chain as possible. But I think, for live use, the hum isn't going to be a problem.

Here are a few photos. Notice that the daughter board contains the entire sidechain, except the high pass filter.




I'll have a page about it up on my website sometime when I've recorded some sound clips.