Diamante Comp flutter problem [SOLVED]

[Andrekp]

Hey everyone,

I built the Briggs Diamante Comp based on the (stripped down) Diamond Comp using the stripboard layout over at Tagboard effects:
https://tagboardeffects.blogspot.com/2014/07/diamante-compressor.html?showComment=1529096095852#c715852899033865357

I have verified this layout against the schematic over at Briggs' site (the creator of this version):
https://revolutiondeux.blogspot.com/2012/06/diamond-compressor.html

Built it as suggested. Subbed in 1N914's for the 4148's, but that shouldn't matter. Used a VTL5C3 "work-alike" from Small Bear (MIR10CLF-R).  Also added a 1n5817 in series with the 100R power supply resistor for polarity protection, which should have no effect beyond a small voltage drop.  Otherwise it is as the layout at Tagboard.  I've triple checked every value and the component layout. 

It works generally and seems to be compressing.

HOWEVER, there is a lot of fluttering distortion going on. It's sort of like a fast trem. I assume the envelope detection is going wild. A few people in the tagboard thread have noted having the issue, but no analysis or solution was mentioned in that thread.

Voltages seem reasonable:
8.18V and GRND as expected on both IC's, with 3.8-4.1V in each of the other pins. (seems normal enough)
Q1 & Q2 have 6.75 on the collectors, a few mV on the bases, and GRND on the emitters, which also seems reasonable.
Vactrol LED is dropping about 1.43V.

Nothing obvious wrong, and it IS clearly working, aside from the flutter. Anybody ever have this issue with the Diamante and/or have some recommendation?

I did try it at 18V, since someone mentioned the flutter clearing up at higher voltage, but it had no effect.  Also tried subbing in 2N3904 trannies per a suggestion, but it had no appreciable effect on the distortion.  Also swapped the two IC's in case they were an issue - no change.  I have some LDR's on order to try rolling a different "vactrol," so maybe that fixes things, but the original uses a VTL5C3, and obviously a number of people have made clones of both the original Diamond and the Diamante using that device and I assume Small Bear's sub is close enough to work fine.

Since it IS working, I feel that the build is generally fine, I just suspect that there is some component tweak that needs to happen to bring it out of distortion.

Has anyone here built the Diamante, maybe had this very problem, and have any insight?

Thanks,
Andre

[Andrekp]

OK, so as the circuit seemed to be working, other than the fluttering distortion, I figured that the most likely problem was the vactrol work-alike from Small Bear that I used.

I made a "vactrol" out of a GL5539 which I ordered from Amazon (of all places) and a yellow LED I had on hand.  This was intended to be close to the VTL5C3 needed.

This worked and there was no flutter, but the compression, while there, was not all that obvious.  I figured that my subbing in the lower gain 2N3904's to try to solve the fluttering, probably were too low in gain to drive the vactrol.  So I put them back to being the original 2N5088's, and the pedal seems to work as it should have from the beginning.

So, FWIW, if you build one of these - or probably any other optical comp - and get odd, fluttering output, the suspect is the vactrol. 

[noisette]

Hey this could be ripple from the envelope detector, try upping the 2u2 in the vactrol driver.
Also vactrol workalike may be taken with a grain of salt, they are sometimes off a lot, could be,
they have much faster time response parameters...

Edit: just saw your additional post, well the transistor gain should be of no consequence, nice it´s working!

[Mark Hammer]

There are a number of places/nodes where envelope ripple can be reduced, but I'll focus on only one here.

In both the original Dynacomp - which the Diamond is similar to, but swaps an OTA for a photocell to minimize noise and clipping - the rectifier circuit uses a pair of diodes to "bleed" unwanted AC to ground.  In the Briggs schematic, this is D7 and D8.  The outputs of IC1a and IC1b are opposite in phase to each other.  Because of that, even though D7 and D8 both have the same orientation to ground, D7 theoretically "bleeds" off half the signal sitting on one half of the reference mid-point, while D8 theoretically bleeds off the other half cycle to ground.  When these two "pre-bled" outputs are combined, it is theoretically identical to "folding" the whole audio signal over.  IN other words, full-wave rectification.

But IS it?  What gets bled off to ground through D7 and D8 is only what can exceed the forward voltage of those respective diodes.  So if they were to have Vf of, let's say, 650mv each, then 650mv of envelope ripple would be preserved, and what gets stored in C21, to be discharged is not "pure" DC.

I had a Ross clone siting around and decided to replace the 1N4148s with Schottky diodes having a lower Vf.  I should probably scope the output of the FWR to confirm, but it sure sounds cleaner and less buzzy to me.

Of course, one of the advantages of using optoisolators/vactrols is that the slight lag in the photocell's recovery "smudges" any residual envelope ripple, the way one smudges oil pastels with your thumbs to eliminate borders/boundaries.  It's like an extra stage of filtering of the envelope signal.  In which case, I can't speak to whether the ripple you hear is something the photocell fails to scrub clean, and could be reduced further with a diode change.  However, if your build allows for it, try using diodes for D7 and D8 with the lowest possible Vf that you can get, such that as much of the unwanted half-cycle is dumped to ground.

Failing that, I will suggest raising the value of C21 from Briggs' indicated value of 2u2, to maybe 4u7 or greater.  As shown in his schematic, that's a mighty fast recovery time. Which is good, I suppose, for preserving attack, but responsiveness also can mean responsive to ripple.  So try the diode change first, and if that doesn't bring satisfaction, try tacking additional capacitance onto C21.  And on top of that - especially if it's a homebrew vactrol using an ultrabright LED - increase the value of R27.  You will note that the gain-recovery speed/time of Dynacomps and Rosses is set by a 150k resistor and 10uf cap to ground.  Granted that the illumination of an LED and control current fed to a 3080 are two different beasts, the time it takes to charge up C21 works the same way in the Diamond and Dynacomp contexts  If the LED you use and R27 allow the LED to remain pretty well illuminated most of the time, then you won't get much variation in compression amount.  IN some respects, you want it to take a bit of time to charge up C21, because that provides part of the smoothing of the envelope too.

[Andrekp]

Thanks for all the info, Mark.

As I said, it seems to be working acceptably with the home brew vactrol. Not quite as punchy as I'd expect, but that may be tweakable when I feel like diving back in.  It's really transparent, and compresses nicely.

Once I pulled it back out, I checked out the real vactrol and it's DOA.  Reads open circuit across the resistor leads no matter what. 

I'm thinking it might be reasonable to up the gain very slightly in the opamp with the compression pot.  I assume that would effectively allow for feeding more signal down to the compression circuit, maybe allowing more range of compression.  I was also considering upping the value of the cap and resistor you mentioned for the very reasons you suggest. Great minds, I guess.

I think I'll let it sit for the moment. I'm tired to messing with it for now and it works fine. 

Thanks for your input!
Andre

[PRR]

> the real vactrol and it's DOA.  Reads open circuit across the resistor leads no matter what.

Vactrol is a lamp and a photo-resistor, inside a light-tight case.

When the photo-resistor is in dead-dark, it should be >1Meg, often >100Meg. Most digi-meters won't read over a Meg or two. So this is the expected result.

You need light on the photo-resistor to see it conduct. You could chip the case open, let the sun shine in, but that would ruin it.

Put power to the internal light. These days almost certainly an LED. Connect 9V through 1K to the LED leads. Try both ways (polarity). The photo-resistor should come down below a few Kohms.

It may still be a real bad fit for the sidechain. But it may work in something else like a tremolo.

[Perfboard Patcher]

As I said, it seems to be working acceptably with the home brew vactrol. Not quite as punchy as I'd expect, but that may be tweakable when I feel like diving back in.  It's really transparent, and compresses nicely.

Hi, Andre.

Good to hear you got your compressor working.

The resistance the vactrol sees at pinouts 3 and 4 is low (R25||R23=100k||470k) compared to for instance the DOD280A compressor (3M) or the Dos Ochenta Compressor (1M). The modification I would consider first would be 470k for R25 and 2k2 for R19. Perhaps even increase the resistance of the voltage divider, 1M for R25 and R23.

[Andrekp]

Thanks guys, but I think I have it working fine with my homemade vactrol. 

And I do believe the issue was the vactrol for a couple of reasons: 

Replacing the vactrol, and only the vactrol, fixed the circuit.

In my homemade vactrol, I WAS able to read the dark resistance of the LDR, suggesting that it should have been a possibility with the manufactured one.

Putting some current across the LED side ( and I did try both polarities) made NO change on the LDR resistance.

So I'm pretty confident that this was simple an uncooperative vactrol.  I may tweak the circuit a little further now that it's working, so thanks for your suggestions.

[snk]

Hello,
I have built a simplified version of the Diamond Compressor, using Fred Briggs DIAMANTE design, and this layout on veroboard.
I really like what it does

However, there are two little things which I'd like to improve :

1- I'd like to be able to get a little bit more of output volume.
At high compression settings, i'm barely at unity gain, and I wish I could get 3 to 6 more dB at make-up gain. Is it possible (I mean, without adding a booster as a daughterboard) ?

2- The barebones Diamante design skips both the TONE and the INDICATOR LED. I don't need the tone, but now that i've built it, I wish I could add an indicator led. I don't need a fancy bicolor led, but it would be nice to have a led showing the amount of compression, without affecting the tone.
Is there any easy way to add this to the Diamante?
The Madbean Holy Roller has such a mod using a daughterboard involving a transistor : would such a design easy to adapt to the Diamante?


Thank you!

[PRR]

[noisette]

Just put the indicator LED in series with the vactrol LED! (it doesn´t get easier:D)