Operational transconductance amplifiers: Demystified

[R.G.]

OTAs are another place where the voltage-centric state of electronics learning can mystify things. OTAs are current-centric, best understood as manipulating currents.

1. The input is a differential amplifier. On a diffamp, a difference of +/-25mV steers the entire emitter bias current to one or the other of the collectors.
2. The collectors are manipulated by current mirrors to reflect the currents from the collectors of the diffamp to the output pin. If you force the + input pin more than 50mV above the (-) pin, then the total diffamp current tries to flow OUT of the output pin. The other way on the inputs and the output pin tries to suck the entire diffamp current into the output pin.
3. For less-than-total imbalances in the input pin voltages, the output pin does a proportional current in/out.
4. The diffamp emitter current is equal to the bias current (another current mirror).

So the output current is the input voltage imbalance ratio times the bias current. The output may be converted to a voltage by running it through a resistor, cause of that Ohm's Law thing.

OTAs act like current controlled voltage to current converters. If you put a resistor on the output, they act like current controlled voltage amplifiers. If you convert a control voltage to a bias voltage, they act like voltage controlled voltage amplifiers.

OTAs saturate ("that's all there is and there ain't no more") when the inputs are more than +/-25mV apart. So the inputs have to be either fed a current (those circuits which have big-value resistors to the inputs) or fed a small voltage with voltage dividers to keep the change in voltage below about 10mV for linearity or 25mV absolute for low distortion.

The rest is all math to get the numbers to come out right.

[chicago_mike]

Theres excellent reading on the CA3280. I have a bunch of those for projects and it really beats the LM13600 / LM13700.

[soggybag]

Funny this came up. I had just been experimenting with a 13600 on the breadboard. Last night I tried to get this Ring Mod to work. Without much luck.


http://books.google.com/books?id=9_tkus_y8BQC&lpg=PA123&ots=QxhKKE2Y8Z&dq=One%20OTA%20ring%20mod&pg=PA124#v=onepage&q=&f=false

See figure 7.13

I thought I'd backtrack a bit and start with R.G.s OTA article. I built the first example on the breadboard. This one didn't work either. Just as I wrapped up I realized I had forgotten to connect the 13600 to V+. Tomorrow I'll give it another try.

[TELEFUNKON]

Funny this came up. I had just been experimenting with a 13600 on the breadboard. Last night I tried to get this Ring Mod to work. Without much luck.


http://books.google.com/books?id=9_tkus_y8BQC&lpg=PA123&ots=QxhKKE2Y8Z&dq=One%20OTA%20ring%20mod&pg=PA124#v=onepage&q=&f=false

See figure 7.13

I thought I'd backtrack a bit and start with R.G.s OTA article. I built the first example on the breadboard. This one didn't work either. Just as I wrapped up I realized I had forgotten to connect the 13600 to V+. Tomorrow I'll give it another try.

Not funny at all: http://www.diystompboxes.com/smfforum/index.php?topic=83661.0 is the related topic!

Don`t forget to connect the IC in Ray Marston`s version of the original datasheet-circuit to V- also!

[soggybag]

Pretty weird stuff.

I connected everything up as per R.G.s OTA article. I have one side of a 13600 set up as a non inverting amp, with a non-inverting op-amp follower acting as a buffer. I have another op-amp set up as a non-inverting follower. For this op-amp I have + connected to the wiper of a pot whose other legs are tied to V+ and V-. The output of this second op-amp is connected to the Iabc input of the OTA through a 7.5K resistor.

So, I turn the knob and the volume goes up and down. But wait! The volume goes up, but, at some point in the middle the volume bumps up. It sort of gets all the way quiet, then it jumps up to all the way on again! Pretty strange.

So then I think, I'll get out my multimeter and measure the voltage at pin 1, Iabc, and see at what voltage the volume does it's little jump. And here's where it gets stranger, Pin 1 shows about 1.2v to 1.4v. It never seems to change much. Very weird.

This sort of jives with some of the info in the R.G. article. There is a note in there about how the Iabc input looks like two diodes from outside so it should never get above 1.4v.

Pretty strange, I'm curious about this what's happening with the jump in volume?

Note: I built everything as per the R.G. diagram, with the exception that I used two 470  ohm resistors in place of the the two 240 and 1K trimmer. I figured for testing I didn't really need the trimmer.

[soggybag]

Maybe someone can explain this. I have a 13600 set up as a non-inverting amp. With an op-amp buffer. I have another op-amp set up as a non-inverting follower with a pot connected to + input. The output of this amp is connected to the Iabc input of the OTA through a 7.5K resistor.

When the voltage feeding into the 7.5K is at +9v I get the max volume. The volume goes down as I lower the voltage. Nothing to really explain here. When the voltage gets down to about .8v sound volume off. But, if the voltage goes any lower, all of a sudden the output of the OTA jumps to full volume! What's going on here?

[TELEFUNKON]

soggybag: are you talking about the first circuit here: http://www.geofex.com/Article_Folders/VCA%20Applications.pdf ? That one is for single power supply voltage, while the datasheet application notes use dual.

[R.G.]

Maybe someone can explain this. I have a 13600 set up as a non-inverting amp. With an op-amp buffer. I have another op-amp set up as a non-inverting follower with a pot connected to + input. The output of this amp is connected to the Iabc input of the OTA through a 7.5K resistor.

When the voltage feeding into the 7.5K is at +9v I get the max volume. The volume goes down as I lower the voltage. Nothing to really explain here. When the voltage gets down to about .8v sound volume off. But, if the voltage goes any lower, all of a sudden the output of the OTA jumps to full volume! What's going on here?

I think it may be this:
I have another op-amp set up as a non-inverting follower with a pot connected to + input. The output of this amp is connected to the Iabc input of the OTA through a 7.5K resistor.
Depending on exactly what opamp you have there, the imperfections of your opamp may be making the output go to full + voltage when you try to force the + input to the minus power supply. Many opamps do this unless they are specifically specified for the input common mode range including the negative power supply. The LM324 does this last correctly. More normal opamps do not.

The limits on common mode input range of opamps is an always-present issue with single power supply LFOs. What you can do is force 0V on the input and output to actually be some voltage that's not right at the negative power supply by using level shifting.

Level shifting is an old, old technique. What you do is bias the negative and positive inputs some distance above the negative power supply with a fixed voltage, like a zener, a string of diodes, or something similar. And you offset the output above the negative power supply by putting a similar offset device(s) on the output. This lets the input go to a non-zero voltage, but the string of offsets on the output corrects this back to zero on the output.

Here's how this works for your setup. You put , say, three silicon diodes or an LED in series with the side of your pot connecting to the minus supply. You put the same number/kind of diodes in series with the output pin. Now the + input can only go down to Voffset above ground, and with the amp set up as a follower, so does the output. But the output offset "corrects" this back down to 0V. The opamp is happy and does not go into output inversion.

Of course, there may be something else going on entirely, so this may not be all of it.

[soggybag]

Yes, I built the first example in R.G.s article. This is on the breadboard right now. And this is where I noticed the volume jump when voltage going into the Iabc dipped below .8v.

I also built the Ring Mod from the  R. M. Marston book. When I built this I set up a 1/2V+ thing for a ground using two resistors and some caps. I substituted this for all of the Ground connections and connected all of the -15v to 0v. This almost worked.

The Ring Mod looked very similar to the Frobnicator so I thought it might work with a 13600 at 9V. The Frobnicator uses the non-inverting input and the Marsten Ring mod uses the inverting input with a feedback resistor. I think there is a note about this reducing feed through.

[soggybag]

I'm still "mystified". Maybe I can measure this phenomenon. I had been suspecting that it had something to do with the Iabc pin. It seemed the .8v was so close to the two diode drops at this input it had to have something to do with it.

[soggybag]

R.G. is correct! I did a quick measurement before leaving for work this morning. It turns out my TL072 wants to jump to ~8v when the voltage at + becomes lower than something like ~1.4v.

OTA's are getting less mysterious. Have to experiment some more. Will try to add an LFO to modulate the output next...

[TELEFUNKON]

When I was working on the accounts at EH UK back in 1980 or so, I remember seeing two or three sheets of information and tricks on this OTA. Pity I didn't take a copy when I looked through the circuits draw! Just thought I'd tell you that - bloody annoying really.

http://www.diystompboxes.com/smfforum/index.php?topic=20359.msg705744#msg705744

[PRR]

> my TL072 wants to jump to ~8v when the voltage at + becomes lower than something like ~1.4v.

That's "THE" major flaw with the TL071 series. They work great to a point, enuff for any sane audio. But go a little further and they flip-out. The problem is that the input stage "turns off" when an input pin (either one) gets too close to the negative power pin.

For what you are doing, the LM324 will work. Both inputs may be swing all the way to the negative power rail (and a bit beyond). The output will swing very-very close to the negative power rail, "all the way" with a pull-down resistor.