Using Arduino to "linearize" a vactrol

cslammy

I have been working on a lag circuit and would like to make it voltage controlled, for the lag. I could use OTA I guess (I have seen that before) but wanted to see if I could use a vatrol or other opto.

The problem of course is that no 2 vactrols are alike, and the thing is crazy non musical for this sort of application.

OK let's tame that beast using our best, brightest buddies from Arduino and atmel?

Idea one:

I figure an arduino and lookup table could be used to force the vactrol to behave in whatever way we want.

I also think you'd just use PWM out to "light up" the vactrol's LED so it's a low parts count circuit fragment--buffer the CV input and then hook the PWM to a resistor then to the LED anode of the vactrol.

With that in hand you can control all sorts of things in a more musical way....make it more linear, make it more log--yes, not good for v octave, but also not the odd non linear stuff we get forced into using out of an untamed vactrol.

use a lookup table?

for 100 ohms, use PWM frequency x.
For 150 ohms, use x1
for 250 ohms, use freq x2
for 1K ohms, use freq. x10
for 10K ohms, use freq x15
etc.

have the incoming CV sweep through this lookup table.

No, this still won't be super accurate, will probably be slow (Vactrols are slow by nature right) and I figure the lookup table will change based on the vactrol.

But it can be done, esp for things like portamento, attenunators, filters, etc.

Idea two:

Automate the lookup table creation!

Write some code that allows one arduino to blast these values into another arduino's lookup table.

So have a ohmmeter on arduino 2, sweep a vactrol with arduino one, then create the lookup table values based on what arduino #2 sees at the ohmmeter

Copy that output into the arduino's memory, and use Arduino 2 (attiny85 I figure? not sure it has enough ram, but something small. Pro mini?) to control the vactrol?

Has anyone already done either of these? I figure idea one should be pretty easy. But I don't think I've seen it before, which maybe means I am missing something obvious here.

Anyone (Grumble? Phobos?) want to weigh in and tell me I'm crazy? Unless I can be dissuaded I will experiment with idea 1. Idea 2 is a bit more out there but I figure it can be done.

Why not dudes?

Grumble · Posted: Tue Dec 10, 2019 8:44 am Post subject:

First of all: I have never worked with vactrols, but I do know they are relatively slow.
If you wish to control the brightness of an LED you use PWM (Pulse Width Modulation), something I have never done as well Rolling Eyes

but I can see two problems:
1 - From what I read, the frequency used for controlling LED brightness is around 50Hz, and I'm afraid that even the slow vactrol will follow that frequency in some degree and in that case it will be audible.
2 - The usual bit-depth is 8 bits, so 256 degrees of brightness, is that enough?

Why not use a digital controllable potentiometer, if you still wish to use an Arduino?
They come even with 1024 steps...

cslammy · Posted: Tue Dec 10, 2019 8:54 am Post subject:

Grumble · Posted: Tue Dec 10, 2019 10:40 am Post subject:

cslammy · Posted: Tue Dec 10, 2019 4:41 pm Post subject:

Grumble · Posted: Wed Dec 11, 2019 12:11 am Post subject:

cslammy · Posted: Wed Dec 11, 2019 8:03 am Post subject:

[quote="Grumble"]

Grumble · Posted: Wed Dec 11, 2019 9:50 am Post subject:

MapacheRaper · Joined: Feb 15, 2018 Posts: 166 Location: Spain

I like the idea: an arduino app that "scans" the vactrol, writes a table of results and then tries to linearize log or expo the result. As Im imagining it it will have a lot of developing work but I dont see why it shouldnt work (with the natural inconsistencies and ilinearities of the vactrols), but It should work.

And it that works I can imagine a lot of people using the library after.

If I where going to do so I would try to do the PWM in the Khz or Mhz, not at 50Hz. Also, a RC filter is just 2 components or 4 if you want to do second order.

Anyway, cool idea. I hope we see the results in a future post on audiowhy

cslammy · Posted: Sun Dec 15, 2019 11:22 am Post subject:

MapacheRaper · Joined: Feb 15, 2018 Posts: 166 Location: Spain

Delicious post, Mr Lamm. I enjoy inmensely every new pots.

In fact I have an esp8266 looking at me. I know I have pretty much exploit the venerable atmega 328p till the limit and the ESP is the next one in been explored and tinkered... Twisted Evil

Regarding the PWM filtering (for the nano at least) Kassutronics has a pretty much perfect filter in this design:

https://github.com/kassu/KassutronicsQuantizer

The filtering is so good that you can perfectly quantize CVs for the modular with it and only takes 4 resistors and 4 caps. In fact I use just half of the poles it and keeps working wonders

MapacheRaper · Joined: Feb 15, 2018 Posts: 166 Location: Spain

The filter . I just use r2, r3, c5, c1, but you can make it double to even clean more the pwm signal

kassufilter.jpg

Description:

Filesize:

64.29 KB

Viewed:

301 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

PHOBoS · Posted: Mon Dec 16, 2019 3:29 am Post subject:

cslammy · Posted: Mon Dec 16, 2019 8:21 am Post subject:

Electric Druid · Joined: Mar 13, 2012 Posts: 44 Location: UK

If you're using the PWM to drive a Vactrol's LED, you don't need any filtering. In fact, you're definitely better off without it. You said in your blog post "could it be that easy?" - the answer's "Yes!" and that's one of the best reasons for doing it! The LED's response is more linear given a PWM input than a variable-voltage input (which is what you get if you filter it. So don't filter!

Example of a multimode filter like yours, driven by a PWM signal..ok, actually it's PDM in this case, but it's a very similar thing:

https://electricdruid.net/filterfx-lp-bp-hp-lfo-filter/

Just run the PWM direct (one series resistor) to the LED. Then you need some way to feed the LDR response back to the uP for measurement. Once you've got that you can determine the vactrols' response.

The classic way would be like the 1980's synth tuning routines: You're trying to achieve a given value X (in this case an LDR resistance) so you set the highest bit of your output. If the LDR's response is too low, you leave it set and move onto the next bit down. If it's too high, you unset it and move onto the next bit down. When you get to the bottom of however many bits you have, you've got a digital value that produces as close to the value you wanted as you can get.

You could also generate a table of log or expo values in the same way. Have an LDR value you're aiming at, and then reach the required PWM output value by successive approximation. In fact, why stop there? - you could do "s" curves too and whatever else you can dream up. The problem on old synths was that doing this for bass notes was pretty slow, and that could potentially be an issue here if the LDRs reposes is very slow. It is on some vactrols, and light-to-dark is different from dark-to-light, so there might be a better direction to run the tests, either bottom to top or top to bottom.

Finally, 8 bits might be the typical PWM resolution on an Arduino (I dunno - not my thing), but the underlying processors can often do 10-bit or even 12-bit without any difficulty, and at many-KHz speeds. I use PICs with 10-bit PWM output at 39KHz.

cslammy · Posted: Thu Dec 19, 2019 5:11 pm Post subject:

Thanks.

I just got my first 2 ESP32 dev boards which do 15 bit PWM out of the box. probably more than I need. Also 8 bit D to A x2 on the board. They were really easy to get working on a Fedora Linux/Arduino IDE setup.

I will continue to experiment with this and post.

But, I have been thinking about this a lot. it would be nice to do some sort of feedback to read the resistance of the Vac as it does its thing, but that could get pretty complicated fast.

I was thinking to do that accurately I'd need to read voltage drop across the resistor part of vac (easy if you buffer it), so that's the V in ohms law, but also the current (which means adding some sort of shunt setup) to calculate the VAC behavior "on the fly". The current measurement would need more buffering so now I think you're looking at a TL074.

With V and i it's easy to get at what the vac is really doing using ohms law.

then goose up or down the PWM using the ESP32 to get the incoming CV to do what you really want it to do.

And of course we are often dealing with AC flowing through the resistor side of the vac, ok now what? RMS conversion of V?

So I don't know how much trouble to go to here. I guess that's always the question? Very Happy

Anyone want to weigh in have at it. How much work is it worth to have a VAC behave a bit more like a digital pot? Maybe not that much?

MapacheRaper · Joined: Feb 15, 2018 Posts: 166 Location: Spain

Doubth sees the obstacles, faith sees the way.

Go for it or you couldnt sleep in peace and will have vactrol nightmares!

Electricdruid... They call him electricdruid for something... Vactrols doesnt need PWM filtering... so true! What we were thinking about?!?!?

Twisted Evil

Grumble · Posted: Fri Dec 20, 2019 3:12 am Post subject:

Electric Druid · Joined: Mar 13, 2012 Posts: 44 Location: UK

cslammy · Posted: Fri Dec 20, 2019 7:20 am Post subject:

cslammy · Posted: Fri Dec 20, 2019 7:31 am Post subject:

cslammy

https://audiodiwhy.blogspot.com/2019/12/vactrols-controlled-by-arduinos-part-ii.html

Spent a lot of time on this, got some interesting results but at the end of the day couldn't get this working reliably and couldn't get it to work at all on ESP32.

Sad

Whatever. In the midst of that I started to experiment with using Vinyl for front panel art which is working better than I thought. Onward.

MapacheRaper · Joined: Feb 15, 2018 Posts: 166 Location: Spain

Even if 2-3 of these days on the workbench sucked, the article/post rocks. You found the workable range of the vacs to share with the community and write a script and possible ways to taking it further.

The esp chips are so powerful and cheap, but seems like bitches to program, isnt?. When I read about them I thought they were to produce next generation synth delicatessens, like an ornament and crimes interfaced with the smartphone, so you can use the screen to control things in the modular. However it seems noone is getting to the deep guts of the ESP, maybe because they are not well docummented?.

In any case ,a delicious post, Charlie. For a 2020 full of electronic joys and pleasures. Cheers