Discrete OTA?

[Earthscum]

I was wondering if anyone had any links or designs for a discrete version of an OTA. I wanna play around and learn how it works and maybe incorporate it into a phaser or something... learning process. Basically, kinda how I played with long tailed pairs and the OP compression amp to decide that OP Amps weren't all that magical... getting a transistor to do backflips was more rewarding, lol.

Even if someone has a link or just a good breakdown, may help a bit... I don't quite know why, but it makes me feel better knowing a bit about what's under the hood, ya know?

[merlinb]

Don't know if this is the sort of thing you're looking for. It's an OTA I put together for use in various projects. With a 7.5V control voltage the overall gain is unity, but this can be changed by altering the gain of the opamp, of course. Maximum input is 8Vp-p, which is more than most stompboxes will deliver.

[Mark Hammer]

Look up VCAs for synths.  There are tons of discrete designs.  Jurgen Haible has an all-discrete replacement for the long-out-of-production SSM2040 (a highly-prized quad OTA designed for synth applications) on his site.  http://home.debitel.net/user/jhaible/hj_sch.html

[Earthscum]

Thanks a bunch... I'll look into it. So, OTA and VCA essentially work the same? Or is there a difference? I've built a couple VCA's from schems on the web. None really worked all that great (may have been user error, lol...)

[merlinb]

So, OTA and VCA essentially work the same?

Not really. An OTA is a transconductance device- it produces an output current proportional to an input voltage. (But the output current can easily be turned into a voltage with a resistor). They usually use a fancy LTP topology or 'gain cell'. The schem I posted is a bare-bones version that does nothing more than provide gain that is proportional to the control voltage.

VCAs produce an output current for an input current and are rather more elaborate than OTAs, as they ideally produce a logarithmic gain control .
http://www.thatcorp.com/History_of_VCAs.shtml

[Earthscum]

What would application examples for the two, where you can use an OTA in "X" design, but a VCA won't work, and in "Y" design, a VCA will work, but OTA won't?

Like the Small Stone... it uses the OTA because of the variable current output to drive the frequency of the circuit, like a fet to ground does in the Phase 90/180. Is that correct?

I see VCA's in almost all synth gear. I've noticed some are basically some form of compressor/expander with the follower circuit replaced by a voltage controlled element. They definitely appear to be simpler circuits to implement than OTA's. The OTA data sheet's discrete diagram was a bit hard for me to follow, even after marking out all the current sources.

[Earthscum]

Aha! This one makes sense to me, lol...

http://webpages.charter.net/dawill/tmoranwms/Circuits_2010/Power_OTA.png (from here
"Power_OTA.png

The complete circuit, showing two power mirrors (as above), the mirrors to drive them, and the bootstrap supplies as required. A differential pair provides input gain, while the tail current is adjustable to set total current, thus making this an operational transconductance amplifier (OTA). Unresolved are potential oscillation/compensation issues, and how to set class A bias current independently of peak output current."

Not sure about the issues they speak of, but as far as the basic circuit goes, if this is a good example, then I understand a bit more. I just don't see how they actually adjust gain any differently than a normal differential OP Amp, other than I can see how many apps it opens up to have bias set separately from your gain. Is this the key? or are there other things at work that make them so useful? (besides the output).

Quick question... are the buffers in LM13600/13700's there as current to voltage converters?

[merlinb]

What would application examples for the two, where you can use an OTA in "X" design, but a VCA won't work, and in "Y" design, a VCA will work, but OTA won't?

VCA are almost exclusively used for audio gain control- that's what they were designed for. I've not seen them used for anything else.

You can use an OTA to do the same thing, but it's annoying trying to get a logarithmic response, and the noise/distortion will be worse than a VCA doing the same job. That's one of the reasons I designed my own- to get lower noise. However, a real OTA can do all sorts of other tricks, like voltage multiplication- just look at the examples in the LM13700 datasheet. http://www.media.mit.edu/resenv/classes/MAS836/ToPost/LM13700.pdf

Like the Small Stone... it uses the OTA because of the variable current output to drive the frequency of the circuit, like a fet to ground does in the Phase 90/180. Is that correct?

Note sure what you mean. Check out the datasheet for examples of voltage controlled filters.

[merlinb]

I just don't see how they actually adjust gain any differently than a normal differential OP Amp, other than I can see how many apps it opens up to have bias set separately from your gain. Is this the key? or are there other things at work that make them so useful? (besides the output).

I think that project is supposed to be a standard MOSFET power amp whose gain can be conveniently controlled via a single control input (the LTP tail). It is the LTP that does the variable gain part, the rest is just a differential amp, and performs the same task as the opamp in the schem I posed (only on a bigger scale). An opamp has its gain fixed with feedback resistors. An OTA has gain controlled by current flowing into a control pin. To me they don't seem very similar at all!

Quick question... are the buffers in LM13600/13700's there as current to voltage converters?

No, you still need a resistor to ground for the current to dump into (or some other impedance if you're making something more interesting). The transistors then just buffer the voltage developed across the resistor.

[mrslunk]

OP:
have a read of http://synth.stromeko.net/diy/OTA.pdf
bit math heavy, but that's my thing so...
however, if you crossreference it to the lm13700 datasheet you should be able to get the gist

[slacker]

Another snippet here http://electronotes.netfirms.com/EN62pg12.pdf

[R.G.]

If you'll do a search, I posted a design for a discrete OTA here (I think).

[R.G.]

on the original post:

Here's what you're looking for:
http://www.uni-bonn.de/~uzs159/ota3080.html

[slacker]

If you'll do a search, I posted a design for a discrete OTA here (I think).

You did, but I've tried to find it since and failed 

[Earthscum]

Aha! Thanks RG. Also, thanks merlin, slacker, and Mark. Got some reading to do this afternoon  (it's slow... lots of time between cleaning screens, printing, and making new screens).

[Mark Hammer]

Apologies for the misdirection.  Clearly, while OTAs are used as a part of a VCA, an OTA is not the same thing as a VCA.  As long as we're all clear on that.

That being said, as long as you can separate the OTA part in any of the VCA designs you stumble onto, you'll be in good shape. 

[Lurco]

follow the links! http://www.diystompboxes.com/smfforum/index.php?topic=38111.0

[Cliff Schecht]

There is a problem with discrete OTA's that you'll be hard pressed to get around. They drift pretty badly usually, even with 1-2% matched transistor pairs. The matching that manufacturers can achieve on an IC is ridiculous, plus the fact that everything is on one die means that everything in the IC will drift pretty equally. The discrete versions tend to be much less predictable and are hard to compensate for. I spent a good 6 months developing a new VCO for PAiA that used a triangle-wave at the heart instead of a ramp wave. The problem I eventually ran into was that any of the commercially available OTA's are either too expensive or don't have a wide dynamic range. With the LM13700, I ran out of notes after about an octave and a half. I figured I could come up with an OTA solution myself using a bunch of 2% matched transistor pairs but ran into nothing but problems with these designs. If they worked at all, they were just too drifty to be used as the heart of a VCO.

With all of that said, there is a lot to learn about the different available OTA type topologies. You'll learn a bit of basic IC design too which is never bad. I'll try to dig up some of my favorite links to discrete OTA stuff and post them later. There is a lot to learn here, that is for sure. I would also recommend studying different op amp topologies, it's usually fairly easy to modify them for voltage controlled operation (although actually building a discrete op amp is no meager task).

[R.G.]

There is a problem with discrete OTA's that you'll be hard pressed to get around. They drift pretty badly usually, even with 1-2% matched transistor pairs. The matching that manufacturers can achieve on an IC is ridiculous, plus the fact that everything is on one die means that everything in the IC will drift pretty equally. The discrete versions tend to be much less predictable and are hard to compensate for...

Oh, sure, throw the cold water of reality in our faces... 

That's exactly what I found. You can match all you want, but they still drift.

[Earthscum]

There is a problem with discrete OTA's that you'll be hard pressed to get around. They drift pretty badly usually, even with 1-2% matched transistor pairs. The matching that manufacturers can achieve on an IC is ridiculous, plus the fact that everything is on one die means that everything in the IC will drift pretty equally. The discrete versions tend to be much less predictable and are hard to compensate for...

Oh, sure, throw the cold water of reality in our faces... 

That's exactly what I found. You can match all you want, but they still drift.

Lol! I think I prepared myself for that shower...

I kinda figured there would be some problems after reading all the links (and links from those links). One thing that came up in your thread, RG, was thermal compensation, which I've experimented with myself (lighter and a supposedly matched pair exposes thermal differences). Oh well, still learning a huge amount of info today! That makes any inquiry worth it, IMHO.