Talk:Operational transconductance amplifier

From Wikipedia, the free encyclopedia

[edit] General Comments

Sorry, but I just came across this article. I do not have time to make these edits, but I hope someone else does. You can find almost all of this information in Gray & Meyer (Analysis and Design of Analog Integrated Circuits).

1) It seems that this article focuses on some older applications using discrete OTAs as opposed to highly-integrated applications (such as an ADC) where many OTAs are used internally. Today, OTAs are primarily used in integrated, switched-capacitor (SC) applications. When you consider the number of ADCs in use in modern electronics, it would not be surprising if CMOS OTAs are the most numerous analog amplifiers in existence. I don't have a reference for that unfortunately. But my point is that the majority use-case is being ignored.

2) OTAs are poor for driving resistive loads, but they are just fine for capacitive loads (hence their use in SC ckts). SC circuits are generally understood as charge-mode devices. CMOS OTAs, because they have such little gate leakage, are perfect for this.

3) The article says that OTA are not used with feedback, but this is not true. In the most common applications, they are almost universally used with feedback.

4) The article says that OTAs have no internal passive components. In modern CMOS, it is very difficult to get high enough gain without using a 2-stage OTA. Once you introdce this second stage, the OTA must be internally compensated in order to be stable in any sort of feedback loop, and this is done with (explicit) Miller capacitance (you might argue that you can make the dominant pole at the OTA output and that this may make it possible to not use any internal compensation, but then you will have a situation where the first stage is too fast for the second, and this causes large output common-mode errors during slewing). In addition to this Miller cap, you will also need lots of other little passives here and there for various reasons (e.g. nulling resistors, neutralization caps, etc).

5) I do not like that the OTA pictured has these extra terminals for biasing. These are implementation specific items that are not really part of the most basic function of an OTA.

[edit] Voltage gain

The article says:

The amplifier's voltage gain is the product of its output current and its load resistance:
G_\mathrm{voltage} = I_\mathrm{out} \cdot R_\mathrm{load}

Shouldn't that be the output voltage?

V_\mathrm{out} = I_\mathrm{out} \cdot R_\mathrm{load}

The voltage gain would then be the output voltage divided by the differential input voltage:

G_\mathrm{voltage} = {V_\mathrm{out} \over (V_\mathrm{in+} - V_\mathrm{in-})} = {I_\mathrm{out} \cdot R_\mathrm{load} \over {I_\mathrm{out} \over g_\mathrm{m}}} = R_\mathrm{load} \cdot g_\mathrm{m}

The units match, too. — Omegatron 15:37, 24 July 2006 (UTC)

Omegatron, you're correct. I don't know what I was thinking of but it was probably a result of being up too late at night when I wrote this. Thanks for catching the error. Do you want to make the substitution or shall I? Anoneditor 19:16, 24 July 2006 (UTC)

[edit] Diagram wrong?

Schematic symbol for the OTA — Like the standard operational amplifier, it has both inverting (−) and noninverting (+) inputs; power supply lines (V+ and V−); and a single output. Unlike the traditional op-amp, it has an additional input Iabc, explained below.
Schematic symbol for the OTA — Like the standard operational amplifier, it has both inverting (−) and noninverting (+) inputs; power supply lines (V+ and V−); and a single output. Unlike the traditional op-amp, it has an additional input Iabc, explained below.

I don't think the pin between the two inputs is the gm-adjusting pin. That pin is called the "diode bias" in datasheets. The "amp bias", which I believe is the variable-gm control, is a separate pin, so there are actually two more pins than an op-amp. See the LM13700, for example. — Omegatron 17:26, 24 July 2006 (UTC)

Omegatron. This is really embarrassing. As I look at it again, the diagram is quite wrong, as you suggest. I was intending to show a diagram for the basic OTA in the style of the original CA3080, which had no input linearizing diodes to bias but only a variable gm control input. However, when I was making the diagram, I was looking at the later LM13700, which does. So, my diagram is a hybrid that should never have been created. I'll get a new one in there soon. Do you want to write something on the linearizing diodes in later generation OTAs? Anoneditor 19:30, 24 July 2006 (UTC)
It's alright; I've done worse. The wiki always corrects it, though.  :-)
Sorry, I don't know any more about OTAs than is already in this article; you'll have to add it. It should definitely be in there, though.
I was already thinking about making a better SVG symbol in Inkscape, though. How about I handle the image and you handle the diodes?
Where does the abbreviation Iabc come from? — Omegatron 20:06, 24 July 2006 (UTC)
My computer is screwed up, so I'll be working on that tonight, but I'll do a good SVG image when I have time. I scribbled on another leg for now. — Omegatron 20:41, 24 July 2006 (UTC)
I'll be happy to write something on the later generation OTA's, but I can't do it until this weekend.
I corrected the diagram before I read your gracious offer of help. If you want to further upgrade it, please do. If you are planning to substitute an image showing the biasing diodes a la the LM13700, perhaps you should wait until I write the piece on them. That way, there won't be any confusion as to the additional input on the diagram. What do you think?
Iabc stands for I amplifier bias current and comes from a book (now out of print) entitled, "IC Op-Amp Cookbook" by an electrical engineer named Walter Jung. It was first published in the U.S. by Howard W. Sams & Co., Inc. in 1974. He uses the term in his discussion of the RCA OTAs that existed at the time. Anoneditor 20:58, 24 July 2006 (UTC)
The diagram is almost correct for half of a CA3280, which is a higher performance dual OTA which did not contain an output buffer. The linearizing diodes in the 3280 are driven by a current mirror and it also includes an emitter pin (for the differential pair driven by the control current) which normally goes unused. So, I think the diagram is ok, though if the input diodes were omitted for clarity that would be fine, too. The CA3080 and CA3280 have been discontinued, sadly. I'd like to say more about these devices, and discuss some of their applications especially their use in audio and electronic music, but I haven't the time to do it properly at the moment. Zzombie 08:17, 28 February 2007 (UTC)
Zzombie, it's actually the diagram for one of the core amplifiers (i.e., sans output buffer) in the dual LM 13600 or 13700 OTA. Anoneditor 18:44, 28 February 2007 (UTC)