diode vca? yes!

zipzap · Last edited by zipzap on Thu Apr 12, 2007 3:13 am; edited 1 time in total

Hi People!
Visiting again, been a while...
I´m wondering how those so called diode limiters work...
Like this one:
http://www.uaudio.com/webzine/2006/august/index2.html

I have seen the diode ladder used in filters, i´m not quite sure how it´s working though. Apparently the resistance of a diode changes whith the current thats going through it, or something.
Well, the only information i could find on the diode ladder vca is that it exists, like in some of those classic compressors.
But no circuits or further descriptions...
Ok, when i take this filter as an example:
http://guitarfool.com/modules/images/Diode_mod_sch.jpg
What would happen if i kick out all the cabs in the ladde? A vca i guess. And if i kick out all the diodes and replace them with a resistor? Still a vca.
So that can´t be the Diode-vca.
Anyone know something about this stuff?

blue hell · Posted: Thu Feb 22, 2007 12:51 pm Post subject:

Look at a graph giving the current that flows through a component having two terminals vs the voltage that is over it as a result of that.

When you do so for a resistor you'll se that the graph is a straight line, a fixed relation between current and voltage. Make the current twice as large and the voltage will be twice as large as well.

When you do the same for a diode you will not see a straight line but an exponential curve instead. Because the line is not straight it means that the resistance varies with the current that flows through it. The resistance at any point on the curve can be determined by looking at the angle the curve at that point makes with the horizontal axis, or it's coefficient of direction (don't know the proper english word).

Now this effect is usable only in a small signal range, as it is essentially non linear. Large linear changes in the current would not result in linear changes in the voltage over the diode. This makes it a bit troublesome to use the principle, as it will always distort a bit or it will have to use very small signal levels resulting into relativly high noise levels.

For the ladder filter the distortion might actually be the thing that makes it's characteristic sound.

zipzap · Posted: Fri Mar 02, 2007 2:36 am Post subject:

and for the diode vca as well, i guess.
Thanks for your explanaition, that´s somehow what i thought it would work like.
Now i´ve been thinking of how to use this effect. might look something like this, right?

didevca.jpg

Description:

Filesize:

39.64 KB

Viewed:

21665 Time(s)

vtl5c3 · Posted: Fri Mar 02, 2007 9:19 am Post subject: Re: diode vca?

There's the Korg MS50 VCA:

http://machines.hyperreal.org/manufacturers/Korg/MS-synths/schematics/ms50filt.gif

Never heard one in action, but if it's like the VCF, it's got poor CV rejection (clicky clicky click!).

Romeo

blue hell · Posted: Fri Mar 02, 2007 3:40 pm Post subject:

zipzap · Posted: Wed Mar 07, 2007 1:35 pm Post subject:

I think our ideas might both work.
One thing about the korg vca:
there is the diode-ring formed inside the ca3019. where is pin 5 going to? open?
i´ll definetly check this thing out and post some samples... soon

vtl5c3 · Posted: Wed Mar 07, 2007 1:42 pm Post subject:

Pin 5 connects to Pin 2, which connects to a 100K resistor....

zipzap · Posted: Thu Mar 08, 2007 3:10 am Post subject:

that means the bottom left diode is shorted.
another thing i don´t get: R67 (100k) goes from + to a diode that goes to ground, and via R7 (360k) to the inv. in of IC10. Is that to generate a 0.6 voltage or is there something else going on?

zipzap · Posted: Thu Apr 12, 2007 3:12 am Post subject:

ok, the thing about the diode being shorted cannot work.

Anyway, here are two versions of the diode vca. haven´t decidet which is better yet.
the first makes the signal pass through the whole diode ring. i found this design in Jürgen Haibles Diode vca.
The cv is symmetrical.

The other version is from a compressor schematic.
The audio signal is symmetrical, the diode ring sits inbetween the + and - signal.

The diodes should be mashed (with steak and a glass of milk)
These vcas are very simple and produce an interesting distortion. The coulouring can be quite subtile and pleasant. Problem is that the output is very very low.
That´s why i imagine that the version with balanced in out might be better to handle.
Of course it could run unbalanced too, just place the ring between signal and ground. But balanced signals have one advantage, especially with low signal level.
A transformer (don´t have any right now) or instrumention amp could be used to amplify the signal and hopefully cancle out any hum.

Now i have a problem: I have no idea how to determine the impedance on the outputs. I have some INA103 chips. Low noise, balanced inputs. But they are specialised for low impedance sources (200r). That´s where they peform best.
Or maybe a difamp with SSM2210?

Ok, if someone can help me that would be cool. I want to turn this into a working compressor/limiter/vca schematic with lots of character and good performance, rather than a dirty distortion for my modular (ok ok, got me, of course i´ll do both Wink

diode vca core.jpg

Description:

Filesize:

143.69 KB

Viewed:

20182 Time(s)

zipzap · Posted: Thu Apr 12, 2007 4:54 am Post subject:

did some measuring, 2V pp sine input gives me 40mv output. CV set to max level without clipping, only some pleasant sounding deformation of the signal. Higher level sounds very ungly.
I also included dc levels and the current through the diode ring. Diodes are BAT43.
Ok, these balanced 40mv have to be brought back to 2v. No idea what the Rout of the signal is, or how i can calculate it.
With these small signal that propably does matter for low noise amplification.
Any ideas?

Posted Image, might have been reduced in size. Click Image to view fullscreen.

blue hell · Posted: Thu Apr 12, 2007 5:09 am Post subject:

zipzap · Posted: Thu Apr 12, 2007 8:52 am Post subject:

ok, in this design the diodes are always under current.
buffering with opamps works of course. on the other hand, i would like to take advantage of the fact that the signal is balanced. So an instrumention amp should be used (the inv and noninverting ins of an standard opamp are to different for real differential amplification.
On the breadboard this design is mainly suffering from hum and noise. When i look at my workshop i can imagine it´s not the design iteself though...
Think i´ll try the ina103 later. Don´t know how much the impedance matters, but the hifi people i know are sure that it matters a LOT... especially with certain "high-end" audio opamps.
A AD797 (0,9nv) is supposed to perform worse than most standard opamps when driven by high impedance sources.
don´t know, don´t know...

zipzap · Posted: Sun Apr 15, 2007 8:17 am Post subject:

Think we solved it. I use an instrumention amp with 2 OPA2132 as inputs. gain is 40. It works perfect. I´ll soon give you a schematic with the output amp and the scaling amp of the cv for conveniant range.
Here are some samples. For cv i used a compression setup.
1. bach is with feedforward, the others are feedbackward.
on the beginning of rosa the offset wasn´t adjusted correctly, it gets overdriven a little at the beginnings of the vocal phrases.

bach.mp3

Description:

Download (listen)

Filename:

bach.mp3

Filesize:

342.11 KB

Downloaded:

1087 Time(s)

elektro.mp3

Description:

Download (listen)

Filename:

elektro.mp3

Filesize:

160.44 KB

Downloaded:

986 Time(s)

reggae.mp3

Description:

Download (listen)

Filename:

reggae.mp3

Filesize:

237.42 KB

Downloaded:

994 Time(s)

rosa.mp3

Description:

Download (listen)

Filename:

rosa.mp3

Filesize:

417.87 KB

Downloaded:

972 Time(s)

zipzap · Posted: Sun Apr 15, 2007 12:02 pm Post subject:

here´s a little schematic and some questions. The upper circuit is the vca core.
Do i need the cabs 1-4? Do i need resistors from the output amp ins to ground?
Should i add any smaal cabs in the opamp feedback resistors? across Rg?
The lower part is for scaling the cv. In this case for use as an compressor. You can scale it for any type of reaction that fits your needs.
the offset trimpot is adjusted for max output at 0v cv without clipping. Then Rg is adjusted so that the output has the same level as the input.
CVs higher than 0 will attenuate the signal. for normal synth reaction the cv has to be inverted and shifted.

One addition:
In a compressor it´s possible that, with slow a/r, the cv is always quite a bit above 0v. The offsetpot can be used to shift the whole cv towards 0v. makes the output louder and improves sn ratio.
But if you do to much you get horrible distortion. So i have added a comparator that makes the led light every time the cv goes below the level set by the offset trimpot.

diodevca.jpg

Description:

Filesize:

135.73 KB

Viewed:

21240 Time(s)