Operational Transconductance Amplifier Input Configurations

Question

I recently purchased several OTAs (Lm13700) but have had quite a bit of trouble using them. Going through the TI datasheet and a Nuts & Volts tutorial it's occured to me that transconductance amps don't really function in the same manner as "normal" op amps. Excluding the obvious currrent output in place of a voltage output the input configurations for many of the example circuits I've seen don't seem to follow normal op amp rules. Most pointedly, while I understand that hysteresis isn't a problem thanks to controllable gain I don't understand why so many circuits feature resistors attached to both input pins. It would seem to me that to achieve differential voltage gain one simple had to apply a voltage to the non-inverting input (perhaps with a voltage divider) and tie the other to ground. However, I tried this on a breadboard without success and only one circuit I found actually suggested this. Would someone mind explaning why OTA amplifier circuits differ so radically from ordinary op amps?

Answer 1

I didn't really want to provide this as an answer, but haven't figured out on the iPad app how to add it as a comment to your question. I am learning about OTAs right now and only recently within the past few months got the hang of regular op amps. Here is a YouTube video that I found helpful regarding the OTA and specifically how it differs from op amps. I've watched this video a bunch of times now and between that, playing around with the circuits on the data sheet, and reading what I can on the chip, I am getting the hang of basic functionality as a current controlled amplifier.

https://youtu.be/I8v0VGsxElA

As I understand it, here are a couple things that make the OTA different from a traditional op amp.

• It isn't generally used with feedback configurations. The output doesn't try to set itself so that the two inputs become equal
• The inverting and non inverting inputs should be no more than a few millivolts. You can set one of them to ground. Exceeding that limit on the input pins will cause the OTA to lose its linear gain properties. They have linearizing diodes to address this, but I haven't used them yet and don't really know how the work. One thing that I have seen people do is to apply the same signal to both inputs, but adding a really small divider resistor between the two inputs (like 100ohms) to make the signal at both inputs ever so slightly different.
• Different from a traditional op amp, the OTA uses a control current to set the gain (it does also use the difference of the input pins to a certain extent).
• not a difference but a word of caution: don't drive the I(abc) pin with more than 2mA! That is the max rating and if you look at the graphs on the data sheet that relate to I(abc) they keep the current below 1mA. Use an ammeter to be safe.

Another resource I would recommend is Ray Wilson's book on DIY analog synthesis, "Make: Analog Synthesizers". There is a section dedicated to the LM13700 that is very approachable and has practical circuits for anyone interested in building their own synth. I found it a lot easier to read than the Nuts and Volts magazine article on the LM13700 - which is still not bad considering that it pretty much goes through the datasheet schematics in more detail than the data sheet does.

I like the design where you can set the gain from a pin, but I need to better understand voltage to current conversions, and current sources before really using it for analog applications.

I wish you good luck with the OTA!

Answer 2

As mentioned already in Dave G`s contribution, the input diodes are for linearizing purposes. This is necessary because - in contrast to voltage opamps - the open-loop "gain" (which, in fact is a transconductance gm) can never be neglected in designing OTA based circuits (with or without feedback).

In classical voltage opamps we have very large open-loop gain values, which means that for linear operation we use only a very small region of the input diff. amplifiers non-linear transfer characteristic (tanh curve) around the origin. This leads to the common rule to assume a zero input diff. voltage.

In contrast, this does not apply to OTAs - even if the input diff. amplifier stage is similar to the opamp case. The operation of OTAs is not restricted to the small "linear" part of the input characteristics (tanh) because we always have a finite input difference that must not be neglected: OTA based circuits must not be designed based on the assumption gm approacing infinite and Vdiff=0. Therefore, some linearization of the input characteristics is necessary - provided by the diodes.

UPDATE: "I don't understand why so many circuits feature resistors attached to both input pins."

OTAs do not "need" any resistors at the input terminals to work properly. However, some applications may require such resistors: (1)Each input terminal requires a dc connection to ground DC bias current) if this is not provided by the signal source; (2) for high frequency applications an input matching may be necessary; (3) OTAs are often operated with negative feedback.