tube VCA for digital volume control?

Hi guys,

I would like some input on the curcuit below.

The original paper and diagram are here:

http://www.cgs.synth.net/tube/vca.html

I have nowhere to host my own pictures yet, so I can't provide a link to my own schematic. It's basically identical to the one on the link above but I don't need the supressor grid so I tied it to the cathode.

According to the original designer, the gain should be linear up to a CV of 3v, making it compatible with solid state CV sources.

I am looking for something I can use as a digitally controlled volume control where CV is controlled through something like a 1024 step digital potentiometer, in turn controlled in 10bit resolution by a suitable microcontroller.

I want to keep the audio passing only through tube curcuitry.

Will this work? Anyone have any other better ideas. I think this is still going to need an additional gain stage as well, probably any pre amp would work.

This is about the simplest thing I could come up with for what I want to do. I have an Electro Harmonix EF86 to play with. They are a bit expensive though, I am not sure if there is another pentode that would also work and cost a bit less.

What do you think? Any suggestions?

Rich :grin:

This is a very difficult proposition unless you are very experienced with tube designs, and even then, still difficult as I dont know anyone thats done it yet.

I must say firstly that I dont know much about this topic and some here know much more than I and can explain it much more clearly. Hopefully they will come and comment (PRR!).

You have to make, essentially a variable gain (Mu) amp like whats in a Fairchild 660 or any other Mu comp, but instead of controling it with a sidechain run off the audio etc etc, you are controling it with a voltage source coming from a linear fader or a micro controller. That part isnt so hard in theory, the problem is that you can usually only reduce the gain of such a circuit by 10-20db before it starts to become very non linear. Many would argue that even a few db causes a noticable change in the sound, depending on the circuit.
The way to get around this is to have several such circuits in series, all doing their max of 20db reduction. So if you had 4 in a row youd get down to -80db, in theory. However Im sure there are major problems with this. One of them is that for four such stages youre looking at 8 single pentode/triodes. Add that to the other tubes youd need for the mic pre and eq and your desk just became impractical.

Now Ill step aside and let people who know what theyre talking about take over...hopefully!


M@

I am probably going to completely show off my ignorance now of how these curcuits work, so please be kind, I am here to learn.

I was under the impression that the particular curcuit I linked to was not reducing gain of the input signal but increasing it from zero. With -1v at the screen the tube is apparently "completely shut off" and a voltage above +1v will begin to amplify the input signal.

"Figure 1 shows the VCA design, implemented with an EF86 miniature pentode. This circuit has been extensively tested, and it has unique electrical characteristics and sound which make it musically valid. In order to completely cut off the tube when zero volts is applied to the control input, a voltage divider applies about -1 volts to the screen. Then, when the CV rises above about +1v, the tube will start to amplify the signal at the input (pin 9)."

The claim is "that this can be used as a linear VCA if the CV is kept below approximately 3 volts".

So I thought if I vary cv between 0 (-1v at the screen thanks to the voltage divider) and +3v I will get a linear amplification of the input signal.

I am probably interpreting this entirely wrong and / or backwards, but I hope someone will explain it to me.

Rich :grin:

If a 100-mV signal is at the grid, distortion at the output will rise steadily with CV until it tops out at about 2% at a CV of about 3 volts; then it will drop down to about 1.0%, remaining fixed until the screen voltage reaches about 50 volts dc, then producing peak gain of roughly 150.

Click to expand...

I think that sentance is the key to the problem with that design. Someone with more knowledge will have to explain it properly though, as Im already out of my depth.
Personally, I wouldnt find a design that adds up to 2% THD to the console's audio path (even if it is mostly 2nd harmonic as that document states) an acceptable solution to the problem of how to have computer control of the channel fader.


M@

Okay, I'm with you now!

I see the problem.

Thanks for your patience with me!

Rich

I have an idea now that I want to simulate in SPICE later today when I get home.

If I had two of these curcuits with the same input source and the minimum CV voltage for both set at a gain level where THD is at a minimum (it just says THD drops to 1% when CV is above 3v up to 50v), and fix one at this minimum level as a reference, using the other one to actually vary gain above this level.

Then could the output from the fixed reference amp be phase inverted, and the outputs of both amplifiers multiplied through another tube stage so that if CV were equal on both the reference and and variable VCA's the two signals would cancel each other out entirely, resulting in no signal at all passing through, but with an increase in CV on the VCA that is variable, the output signal would increase accordingly.

It's just a theoretical idea using perhaps only three tubes, though even 1%THD is still too high. I wonder if this has anything to do with the EF86 specifically, or if any other tube would have the same distortion characteristics.

I would also have to drive the control voltage up to 50V rather than the 3V I was initially contemplating.

Does this sound feasible even in theory or am I smoking something?

Rich :grin:

Yeah, it would work after a fashion over a limited dB range. But what are you trying to achieve? Are you trying to build a dB-linear attenuator/gain circuit? Do you really want an output high in second-harmonic distortion, the percentage of which varies with attenuation setting?

Sorry to be cynical, but quite often recently I've seen valve (tube) circuits pressed into unsuitable use just because they are valves and so they must sound good.

You would get a better dB range out of something like a Vactrol, but without the distortion. Assuming you don't want dynamic attenuation (as in a compressor, for example), you could drive the Vactrol LED from the PWM circuits present in almost any microcontroller and use the inherent lag in the Vactrol do do the averaging. The problem of linearity and repeatability between units takes a bit more design skill, but would be no different for the pentode design in the reference you posted.

Thanks for the input! At this stage all of this is just theoretical, the problem is however real, for my application at least, and I believe that a solution must exist, if I am wrong and there is no real way to make this work, or the "solution" proves to in fact compromise the real quality of the audio passing through it to an unacceptable degree then I will try another approach entirely.

I would like to get rid of the second harmonic distortion (well all distortion) as much as possible, and I have been studying some papers written on the subject of introducing "inverse pre-distortion" which in theory and seemingly in practice (depending on how it's implimented) cancels out not only 2nd harmonic distortion, but higher harmonics as well.

To a certain extent, I believe that I can compensate for any non linearity of the VCA itself, and in fact even tweak the curve to be as linear or non linear as I want entirely in programming the software that drives the control voltage of the VCA to compensate.

Distortion however, is another issue altogether, and may actually be the thing I need to concentrate on resolving.

It's early days for this one, I can't wait to actually build something a test it though.

Rich :grin:

The pwm driven vactrol idea is interesting instead of using a serial interfaced digital potentiometer. It doesn't change the need for a VCA type curcuit to actually attenuate the audio signal though.

Thanks again for your advice and suggestions!

Rich

[quote author="Boswell"] Assuming you don't want dynamic attenuation (as in a compressor, for example), you could drive the Vactrol LED from the PWM circuits present in almost any microcontroller and use the inherent lag in the Vactrol do do the averaging.[/quote]

Hey Boswell
Reading what you wrote about driving the vactrol with PWM makes me remember about the optical comp from Pendul*m where they say they use a trick to make optos react faster. Do you think a PWM solution would do anything in this direction? Any thoughts related to this subject?

chrissugar

Boswell, I think I know what you were getting at with the vactrol, as the photocell is a resistive element, it could be used to control gain just as a normal potentiometer could in any other tube amplifier, only it's resistance varies with the pwm of the LED rather than a hand operated pot. As you said, the inherent lag in the photocell would average the high frequency pulses into a continuously variable resistor.

This is great as it would not suffer the stepping associated with the fixed network of resistive elements in a digital potentiometer (although at 1024 steps, it's hardly going to be noticeable) or the high tolerances either.

This means the particular VCA curcuit in question is no longer needed. This also should get rid of the distortion problem (over the dB range expected from any mic or line pre amp at least).

I would still have to use the same principle discussed before of phase inverting the fixed gain output of one amp and using it to cancel out the output from the second variable gain amp to leave only the difference between them at the output.

Unless of course there is a better way to do that too.

Rich

6SJ7 works good inplace of EF86s, might need a little tweaking, but in that circuit it will probably work pretty close tot he same as is. Its cheap and plentiful, do not seems to be anywhere near as microphonic as EF86s either. I like their sound better as well.

Why not use something like an OTA as a voltage controled resistor shunting the signal to ground. It would still keep the audio path tube, it is not to complex. EDN (http://www.edn.com/) had an article on useing the 13700 as a VCR a couple years ago if I remember correctly. it worked pretty good, probably more then good enough to just shunt audio to ground. More complex then vactrols or LEDs/LDRs but you do not need to deal with matching them.

There is also a post just like this in the drawing board right now.

http://www.groupdiy.com/index.php?topic=16760

adam

chrissugar wrote:

Reading what you wrote about driving the vactrol with PWM makes me remember about the optical comp from Pendul*m where they say they use a trick to make optos react faster. Do you think a PWM solution would do anything in this direction? Any thoughts related to this subject?

Click to expand...

My memory is that the "fast Vactrol trick" was nothing more than splitting the LED series resistor into two parts and putting a capacitor across one of the parts. In this way, it gets a good kick of current when there's a sudden change, and the time constant can be fudged to mimic the Vactrol. There's possibly a diode as well so that the attack is quick but the release is slow, but the Vactrol has that characteristic anyway.

rlackey wrote:

I would still have to use the same principle discussed before of phase inverting the fixed gain output of one amp and using it to cancel out the output from the second variable gain amp to leave only the difference between them at the output.

Click to expand...

I think you're going to have to run this past me again, as I didn't understand the reasoning behind it the first time round. If it's for distortion correction, you can't cancel distortion with anti-distortion in a variable-distortion circuit. Better not to use a distorting circuit in the first place.

Wow, that thread is almost exactly the same thing I am trying to do only mines for a multichannel mixer not a 5.1 surround amp, and mine needs to be microcontroller controlled.

In his case, I would agree that a stepped attenuator is the best route because he doesn't require fully automated digital control of the "fader". What I am trying to implement is essentially digital "fly by wire" attenuation of the audio level to replace a traditional resistive fader, and maintain the all valve topography of the console.

I'm a bit confused now though, do you have any sketches to illustrate your suggestion of using an OTA as a voltage controlled resistor shunting the signal to ground?

Thanks for the ideas Adam!

Rich :grin:

[quote author="Boswell"]I think you're going to have to run this past me again, as I didn't understand the reasoning behind it the first time round. If it's for distortion correction, you can't cancel distortion with anti-distortion in a variable-distortion circuit. Better not to use a distorting circuit in the first place.[/quote]

No, you're right, I think I was a bit confused still at that stage. :razz:

Rich

I have not really played around with voltage controlled resistors much, abit years ago, I was just throwing out an idea. But there is alot of information out there the LM13700 data sheet has some info, and a google search for "OTA Voltage Controlled Resistor" will bring up alot info.

adam

Greatings from the guy thats doing a 5.1 preamp

It wouldn't be hard to do an led/ldr as was recomended to me, and you wouldn't have the matching problems i would (you can compensate on the digital end)
if you pwm or just some how adjust the brightness of the led, you get an isolated resistance that is digitaly controlled.
good luck!
Bill

[quote author="chrissugar"]
Reading what you wrote about driving the vactrol with PWM makes me remember about the optical comp from Pendul*m where they say they use a trick to make optos react faster. [/quote]
Isn't it Manlee who says they have a "trick" with the vactrol. I believe they've a patent on the term "trick" being used in that app.

[quote author="WJS"]Greatings from the guy thats doing a 5.1 preamp

It wouldn't be hard to do an led/ldr as was recomended to me, and you wouldn't have the matching problems i would (you can compensate on the digital end)
if you pwm or just some how adjust the brightness of the led, you get an isolated resistance that is digitaly controlled.
good luck!
Bill[/quote]

Hi!

Yep, I think the vactrol is the front runner right now. It won't be that hard to tweak the resulting output in software.

Can I patent the term "tweak" in this app? :razz:

I had an idea last night that came to me in the twilight zone somewhere between awake and sleep, that could make matching outputs easier.

First some background on the motor faders

For the motor faders, the motors are pwm controlled by the microcontroller, and the resistance track of the fader is only used to provide feedback to the microcontroller so that the microcontroller "knows" the position of the fader. This is what provides the midi output to whatever multitrack recording software you are using.

This feedback also makes remote control of the faders possible, as the microcontroller recieves a midi value for the particular channel (either greater than or less than the current value) and begins to move the fader, constantly aware of the closing gap between where the fader has been told to be, and where it is. As the gap closes the motor slows until a "deadband" is reached close enough to the desired position where the motor stops. This deadband keeps the fader from constantly jittering around it's desired position.

The fact that the microcontroller knows exactly where the fader is at any given time, and where it should be when being addressed by software, it will also (through a Vactrol or whatever method is chosen) then attenuate the actual audio level for that channel accordingly.

That's the "fly by wire" part. The idea I had last night, is to use a similar feedback from the output of the vactrol back into the microcontroller, so that the microcontroller knows exactly what the level is on the resistor side of the vactrol. Then matching levels and response across multiple channels will be much easier, and the tweaking can largely just be left to software.

I think this will work. This console is really starting to come together now.

Rich :grin:

Oops, I somehow managed to post that one twice. This edit is to delete the second identical post.