MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

OTA circuitry - which "topology" ?
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author OTA circuitry - which "topology" ?
hox3d
Hey people,

Just had a dumb question.
I know that answers are lying all around here, but couldn't find any relevant case for the comparison I try to achieve here.

So. Since I'm working on an existing design I'd like to modify, this is the first time I actually had to confront an OTA.
And I don't understand the thing quite well (I'm more about digital electronics).
Well, I could use an existing circuitry, do it without understanding, and we're set.

However, while I was looking for different examples, I discovered few circuits did the same thing but in a slightly different way. If I'm not mistaken, these are used to produce the current to "close/open" the OTA.
Here they are, in the following order: Befaco, Thomas Henry VCA-1, MI's Ripples.

So basically all these "drive" an OTA in order to make a VCA.

The actual values of the resistors don't really matter: I omitted the mixing circuitry (CV in) to get only the pot control over VCA (which is what I try to do, a simple pot-controlled VCA, when I told you it was a stupid question, I wasn't kidding).
These values have to change, of course, but I was just wondering about the topology.

So, is there one which is better? I believe that the VCA-1 has bipolar control (attenuverter).
Since the part count really is different, do you think I could swap (almost seamlessly) one circuitry for another?
For instance Befaco used a lot of resistors, which MI didn't.
How do you explain these differences between the three of them?

I might have few additionnal questions though:

- The Darligton buffers in the Ripples design are not used, however one of the pins is connected to Vcc. Why for?
- I thought a LM13700 was a VCCS (voltage controlled current source). However, here, we seem to be controlling the amplitude with the Iabc input. So in fact we're actually controlling the amplitude with current, right?

If you can enlighten me a little bit about this, that would be nice.

P.

NANO-Jorge
I use the MI Ripples topology and works fine, but I can't tell you more because I don't know so much about LM13700.
ucacjbs
The topology used in ripples looks like Mike Sims’ VCA, described in some detail here:

http://www.teaser.fr/~amajorel/sims/

which might make a good starting point for understanding the TH design too.

This is a great topic for discussion BTW - hope that we get some comments from more experienced people than me!
whyfarer
interested and watching but unfortunately don't have anything to add
J3RK
Here's my version of the Sims style VCA. I like this CV setup. It rectifies the incoming CV signal to keep it all positive going, it's low parts count, and with the recently added CV trim, you can zero out any offset. (one could also just add a 20-22M resistor to +V to add a tiny negative offset (few microvolts or so) which ensures that it closes all the way.) The CV Input summing resistor can be changed to 39K for 0-10V operation.

As to the question of being able to swap them, I don't see any reason you couldn't do that. You might need to tweak the final output buffer level a little bit, but anything before the OTA bias pin you should be able to swap out.

Dr. Sketch-n-Etch
The OTA is an "Operational Transconductance Amplifier."

Transconductance is simply the conversion of a voltage into a current. This is what an OTA does: it converts the voltage difference at the input terminals into a current at the output. The gain of this transfer is determined by the current on the Iabc pin.

The relevant math is as follows:

I_Out = (V_In+ - V_In-)*I_abc*(q/2kT)

kT/q ~ 26 mV, so q/kT ~ 0.0385/mV and q/2kT ~ 0.0192/mV ~ 19.2/V

Hence,

I_Out ~ 19.2*(V_In+ - V_In-)*I_abc for V in volts, or

I_Out ~ 0.0192*(V_In+ - V_In-)*I_abc for V in mV.

Note that OTAs are horribly non-linear if the input voltages exceed just a few mV. Hence, the voltages in synth circuits are typically divided down about 500 times before being fed into OTAs. That's why you will often find a resistor divider network of something like 100k-220R at the inputs to an OTA. Also, one side of an OTA is often grounded, so this makes the OTA a direct voltage-to-current converter.

Let's take a quick look at a simple OTA LPF stage:


Here, the signal is coming into the + input. The LPF requires negative feedback, which is taken to the - input of the OTA. The voltage gain to the inputs is roughly 0.0022 -- hence, a 10Vpp signal will enter the OTA as only 22mVpp. The advantage of an OTA (over a 2164 VCA) is that it needn't terminate to virtual ground, so it can go straight to a cap-and-buffer configuration where the output voltage of the filter stage is exerted at the output of the OTA. This makes it useful for Sallen-Key filters and other configurations where a 2164 VCA simply cannot be used. Nodal analysis of this circuit reveals the classic LPF transfer function:

V_Out/V_In = 1/(sRC + 1)

where R = 1/ag

where a = 0.0022 and g = the gain of the OTA (= 19.2 I_abc)

(Pop-quiz question: Given the component values shown, what is the corner frequency of this filter stage at an I_abc current of 1 mA?)

I hope this makes it a bit more clear how OTAs work and how to include them in designs.
ucacjbs
Ha! My memory is bad. We had a big discussion on this a while back (a few years I think), and I ended up using J3rk's version in a wavefolder I cobbled together.

From what I can remember of that discussion, and having a quick look at the three circuits, I'm somewhat confident that the befaco circuit offers exponential control over the current to the OTA control pin, and that the J3rk/Sims VCA offers linear control of the current, as does the TH one.

J3rk's version of the ripples CV circuit (and the TH design) offer protection against negative input voltages via the diode and path to ground in the feedback loop. This prevents the transistor junction becoming reverse biased to the degree that it breaks down.
J3RK
Putting that in the feedback path of the op amp linearizes the control. This is actually what I do on my linear slope module. I use an inverting integrator, (cap in feedback path like a VCO) so that the output is linear.

(not to mention allowing the summing of the CV signals at the same time)
ucacjbs
J3RK wrote:
Putting that in the feedback path of the op amp linearizes the control. This is actually what I do on my linear slope module. I use an inverting integrator, (cap in feedback path like a VCO) so that the output is linear.

(not to mention allowing the summing of the CV signals at the same time)


I'm having context failure d'oh! hihi -- what's the 'that' in the feedback path? The transistor?
J3RK
ucacjbs wrote:
J3RK wrote:
Putting that in the feedback path of the op amp linearizes the control. This is actually what I do on my linear slope module. I use an inverting integrator, (cap in feedback path like a VCO) so that the output is linear.

(not to mention allowing the summing of the CV signals at the same time)


I'm having context failure d'oh! hihi -- what's the 'that' in the feedback path? The transistor?


I was just referring to the whole CV block (cap, diode, transistor).
ucacjbs
J3RK wrote:


I was just referring to the whole CV block (cap, diode, transistor).


Ta thumbs up
cygmu
Thomas Henry's book "Making Music with the 3080 OTA" contains some nice circuits to drive OTAs with clear explanations.

As others have said, the Befaco circuit is an exponential driver while the other two are linear. I prefer the Thomas Henry one because it gives you a lot of control over the current that is delivered per volt of CV input, by varying the resistors. In the Mutable design the current out is whatever flows from the input CV to the virtual ground (1/62 mA per volt of CV with the values you have). To change this you need to change the input resistor which alters the input impedance of the op amp stage.

Thomas Henry's circuit delivers additional current from ground via the 11k resistor so you can vary this independently by altering either that 11k or the one you currently have at 100k.

But, for your simple application, if you only want to control the OTA via a pot, you can just string a pot from 0V to the negative supply, or if you want, from positive supply to negative supply, and hang a resistor off the wiper. Dr Sketch-n-Etch has given you the formula to calculate what current you want to deliver. The OTA Iabc pin is at two diode drops above the negative supply. Armed with this you can calculate the size of the resistor you need to use to get the current range you are looking for.
hox3d
Thanks for your answers.

cygmu, that's why I thought as well. This is for audio, so I guess a log pot would do.

Dr. Sketch-n-Etch, J3RK, these are quite precise answers.

For those interested, that's a 4U-format Befaco Crush Delay with some tweaks.

For other inputs (those mixing an input, its attenuator, and a control pot), I wanted to know how this worked.
I actually simulated the Befaco circuit shown above, and with an LFO input, it wasn't linear at all. But it's on the feedback control, so I guess it has to be log.

The current front panel concept does not allow CV control over amount sent to delay, only an attenuator (that's a choice I made when I adapted the circuit).
So I just wondered if I'd put some optional pins allowing one to add the CV-controlled amount, where I could use such circuitry.

Thanks to all anyway, that did clear things up.
trip
The Mike Sims page contains a lot of the relevant information for calculating the gain of his dual ota vca and is very well explained:

http://www.teaser.fr/~amajorel/sims/

I found it useful in understanding how the lm13700 works
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group