Talk:Transimpedance amplifier

Initial comment
Hi, I don't understand the use of the original research or unverified claims tag here. This is everyday electronic engineering stuff that has been very well presented. Sfrahm 11:10, 26 October 2007 (UTC)

Good article. I hope you wont mind me trying to improve it by Wikifying and generally tidying it!--Light current 00:33, 11 August 2006 (UTC)


 * Hi Light current! Thank you for the willingness of cooperation. I had just begun thinking about your last discussions on the page of negative resistance when I noted your changes in the page of transimpedance amplifier.


 * Of course, my insertions need improving by a native English speaker(s). I have realized that I have only roughly exposed the topic. I promise you that I will assist you (in return for your editorial help), if you ever decide to join the BG Wikipedia :) Circuit-fantasist 08:06, 12 August 2006 (UTC)

Diagrams
I know it's a pain, but the word 'harmful' in your diagrams gives the wrong impression to readers. Maybe you could change it to 'unwanted' 8-)--Light current 01:06, 11 August 2006 (UTC)

Impede, embarrass, hamper, enervate, decrease, diminish...
I would like to say that something is bad as it impedes, embarrasses, hampers, enervates the excitation input voltage when it strives to create a current. Also, I would like to say that the "bad thing" decreases, diminishes the effective (actual) current-creating voltage VRi = VIN - VR (in this arrangement, the resistor Ri actually acts as a voltage-to-current converter).


 * 'Acts in opposition to'; or 'opposes'.
 * The "bad thing" REDUCES (or diminishes or cancels) the current-creating voltage VRi = VIN - VR --Light current 20:01, 20 August 2006 (UTC)

"...effectively modifies the excitation voltage VIN..."
According to the considerations abov, we have to write, "...effectively modifies the actual current-creating voltage VRi..." By the way, can you "unearth" some simpler single word (like decrease, diminish etc.) instead the combination "effectively modify"?

What is 'harmful' - resistance, voltage or the both?

 * Only, beginning to think about your word substitutions (I agree them) I have gradually realized that actually only the resistance is the "bad thing" in this arrangement (discuss). The voltage drop VR is not a 'harmful' quantity; instead, it is a useful quantity as it is created by the input voltage source, in order to overcome the resistance! In other words, the voltage drop VR is the reaction of the excitation source to the impeding resistance; it is its voltage, not resistance's voltage! So, if you allow me, I will apply the adjective 'undesired' only to the resistance R.


 * Maybe, this discussion is also suitable for Ohm's law where it is more precisely to say, "...the voltage source develops a voltage drop across the terminal conductors of the resistors..." instead, "...resistors develop a voltage across their terminal conductors..."

The basic idea behind the passive current-to-voltage converter
I have inserted a para about the basic (non-electrical) idea of the passive version (similarly the active version). The idea is simple and well-known from our life: the impidements cause a pressure; so, in order to create a pressure, an impediment has to be applied.

A new page about the passive current-to-voltage converter?
I suppose to open a new page about passive current-to-voltage converter; IMO, it deserves attention. Then, we may remove the part about the passive version from this page to the new one.

Swapping Transimpedance amplifier and Current-to-voltage converter?
I would like to come up for discussion the title. Actually, transimpedance amplifier and current-to-voltage converter are the same device. Only, I don't know why the first name is more frequently used although
 * transimpedance amplifier is more meaningless than current-to-voltage converter
 * current-to-voltage converter is more general than transimpedance amplifier as there are two versions (passive and active one) for it.

So, I suggest to make 'Current-to-voltage converter' main page and 'Transimpedance amplifier' - redirected page. Circuit-fantasist 14:47, 12 August 2006 (UTC)

Actually, I would suggest going with the one that is more frequently used which is transimpedance. Also, to say that transimpedance is more meaningless would show a lack of understanding of an transimpedance amplifier. The gain of a transimpedance amplifier is ohms. Hence the name transimpedance. I can't tell if the trans is like transconductance in which that article says it stands for transfer but the same concept. Transconductance means the gain is going to be in Siemens. 155.33.109.148 22:47, 11 August 2007 (UTC)


 * I would like to say that the word transimpedance amplifier is more meaningless than the word current-to-voltage converter. Circuit-fantasist 16:06, 14 August 2007 (UTC)

Cleanup
I added the cleanup tag because of formatting and layout. It's rather choppy with very short paragraphs each with a heading. (Information looks good during a cursory look but the layout just makes it hard to follow). RJFJR 21:07, 21 September 2006 (UTC)


 * Is it OK enough already? Circuit-fantasist 10:23, 22 September 2006 (UTC)

Transimpedance Amplifier vs. Current to Voltage Converter
A C2VA is not always a TIA but a TIA is always a C2VA. Example: A resistor converts current to voltage, but its not a TIA. The TIA is an idealized model, just like the voltage opamp is an idealized model for simplified calculations. Steve110 20:31, 28 March 2007 (UTC)


 * [[Image:Classic_300.jpg|right|frame| Current-to-voltage converter (the most popular circuit implementation) ]]Current-to-voltage converter is the most general term including both the passive current-to-voltage converter (a bare resistor) and active current-to-voltage converter (transimpedance amplifier). There is a close interrelation between the two versions - the active version is come from the passive one; the more complex active version consists of a passive version and a compensating op-amp. The passive version is reversible (it may act as the reverse voltage-to-current converter); the active version is not reversible. Current-to-voltage converter is more natural, clear and meaningful word than transimpedance maplifier. Really, the transimpedance amplifier is an idealized model but the circuit showed on the right is one of the possible practical circuit implementations (maybe, the most popular one). Circuit-fantasist 17:04, 29 March 2007 (UTC)

I have swapped the two pages as the current-to-voltage converter name is more general than transimpedance amplifier one. Now, Current-to-voltage converter begins with the passive version and ends with the active one (see also Voltage-to-current converter).

I have moved also the discussion from this page to Current-to-voltage converter talk page.

Circuit-fantasist 12:06, 17 March 2007 (UTC)

I removed the redirection to Voltage-to-current converter and copied the most recent edits. Zen-in (talk) 02:50, 1 December 2013 (UTC)

Confusing pictures
The pictures in this article have so many layers of annotation in multiple colors that they are extremely confusing. They look like the end result of a lecture, rather than an illustration for an article in an encyclopedia. A simpler set of schematics would be a valuable replacement. Kevin k (talk) 17:08, 11 December 2012 (UTC)

Agreed. The pictures are of low quality, they are too verbose and they don't clearly explain the matter at hand. If someone can simplify the article & pictures - please do it!

Vaxquis (talk) 21:04, 17 September 2013 (UTC)

+1 I am also in agreement. This definitely looks like this is notes from a lecture instead of an encyclopedia page. Reportingsjr (talk) 18:40, 20 October 2013 (UTC)

Page re-write and rename as Transimpedance Amplifier (TIA)
I propose this page get a complete re-write and be renamed Transimpedance Amplifier. In Robert Pease's article he states "we used to call this a current-to-voltage converter,". The technical details on this page are weak and not well presented, despite the excessive verbosity and colorful plots. There is no discussion of compensation or the gain/BW tradeoff. Zen-in (talk) 16:49, 2 November 2013 (UTC)
 * I agree with the name change. "Current-to-voltage converter" should redirect to newly named page.Constant314 (talk) 14:52, 22 November 2013 (UTC)


 * Both Transimpedance Amplifier (TIA) and Transimpedance Amplifier are non-conforming names. Transimpedance amplifier is ok though.  Spinning  Spark  16:40, 22 November 2013 (UTC)


 * I moved the most recent version of this pages content (after removing CD's well-intentioned graphi-analytic discussion).   What should we do with this page now?   Maybe describe transistor and other simple current to voltage circuits?   Or is that too trivial?   Zen-in (talk) 02:56, 1 December 2013 (UTC)


 * Oops, I remember a discussion about forks a while back so I replaced the content of this page with a redirect to Transimpedance amplifier. Anyone object?   Zen-in (talk) 03:09, 1 December 2013 (UTC)
 * You have also forgotten that cut and paste moves are not desirable for licencing reasons. I have deleted the forked transimpedance amplifier and reverted the redirect.  Please move the article using the move button (if you still desire to do so) and also move this talk page.  I will restore your most recent edits once you have done that (as long as there are no intervening edits while the page is still here).  Next time you should request deletion of the target page by an admin if the wikimedia system stops you from moving via the move button.  Spinning  Spark  09:50, 1 December 2013 (UTC)
 * Yes, I guess I'm not the only one who is out of practice.  I hope this page is more useful now.   I will try to make the technical parts more descriptive.   Transimpedance amplifiers are somewhat complicated beasts so simplifying the analysis would not be useful to anyone.   A section showing why transimpedance amplifiers tend to oscillate and have gain peaking will be added. Zen-in (talk) 18:20, 1 December 2013 (UTC)
 * I have now restored your deleted edits as promised, but since the article has been edited in the intervening time, they are no longer in the current article. You can retrieve them from the history if you so wish.  Spinning  Spark  18:51, 1 December 2013 (UTC)
 * That's ok. I added the deleted edits and then some.   The current version has more corrections. Zen-in (talk) 21:52, 1 December 2013 (UTC)

Overly general statements in Bandwidth and stability section
You are doing a great job. Consider this as a minor suggestion. I think a couple of your sentences are overly broad. First, “There is no cut and dry formula for calculating the capacitor value that works for all cases”. If all cases includes all types of amplifiers, loads, sources, applications, stability and bandwidth requirements then the statement is trivially true and applies to all amplifier design. Second “the iterative method required to optimize the value”. An iterative method can be used but there are many cases where those sufficiently skilled in algebra do not need to iterate. I have not looked at St Bob’s article, but I suspect he formed his opinions in the bad old days when opamps would go crazy with even a slight capacitive load and the spec sheet didn’t mention it. You were pretty much stuck with iterating the hardware (soldering real capacitors onto the breadboard to see what would happen). But in the modern era, opamp designers have recognized that there is often some capacitance hanging on the output. They have built in some tolerance to capacitive loads and they even spec it. If the series combination of Ci and Cf is less than the opamp’s specified capacitance driving capability then you can probably optimize the design on paper. I suppose you might iterate the paper design. Maybe that is what you meant. Still, I would replace “required” with something less compelling.Constant314 (talk) 05:18, 5 December 2013 (UTC)
 * Thanks for the support and compliments.  The last few sentences need some work.  I have slowed down in the last couple of days since I'm very busy with other things on weekdays.  I will be adding something about the bandwidth and will try to re-word that last section.  Graeme's analysis of tia instability and their compensation are very well presented.   I have just skimmed over his analysis here and recommend others read his book for the full story.  My limited experience designing and optimizing tias, for quantum dot sensors, has taught me iteration is needed and any textbook calculation of Cf just keeps you on the fairway.   There are no hole in ones in this game.    The gain-setting resistor and even the kind of load the tia is driving affect the final value.  Even the placement of the cap makes a big difference.   While this gets beyond what an encyclopedia should present, it would be incorrect to write that a paper design can be iterated or will ever be optimal.   Zen-in (talk) 15:11, 5 December 2013 (UTC)
 * I’ve done it twice. Depending on what you mean by iterating, I may have iterated the paper design but not the hardware.  I suppose it was optimal in the sense that it was not noisier than the paper analysis said it would be.Constant314 (talk) 04:17, 6 December 2013 (UTC)
 * Yes I guess you can design it on paper and it won't oscillate after the solder has frozen.  I think providing links to something St. Robert wrote or other sources is sufficient.   The level of detail required to describe the design and iteration of compensation may not belong here and could just disappear even quicker.   There are a lot of other aspects to tias that should be covered.   If you have the time to add something about the bandwidth of tias go for it.   I have left that part out for now.  Maybe it should be covered on the opamp page but if it is well done it will likely get moved and live on somewhere else.  Zen-in (talk) 07:20, 6 December 2013 (UTC)
 * I'll just add the word often in front of required.Constant314 (talk) 22:48, 6 December 2013 (UTC)

Transimpedance amplifier gain
Copied from User talk:Constant314
 * I reverted the first section back to a discussion about amplification and gain since it was starting to get muddled. Gain, as described in a Wikipedia page is not tied to any units.  The gain, or amplification of a TIA is measured in Ohms.   I don't believe there is any ambiguity in stating the current is amplified anymore than it would to say what comes before a speaker is an amplifier.  Zen-in (talk) 05:53, 10 July 2014 (UTC)


 * They are called trans-impedance amplifiers because their gain is an impedance. The input is current and the output is a voltage. The only gain that you can multiple a current by and get a voltage has the units of impedance or ohms.  In fact, the gain of the circuit in the first picture is RF.  Assuming an ideal (zero output impedance) amplifier, the output current depends on the amplifier load and can range from zero to arbitrarily large.  Current is not amplified by a trans-impedance amplifier.  For that you would need a current amplifier.Constant314 (talk) 21:08, 10 July 2014 (UTC)


 * And by the way, Wikipedia pages are not considered reliable sources.Constant314 (talk) 21:09, 10 July 2014 (UTC)


 * No you are wrong about this.  Amplification does not require the same units at the input and output.  There is no ambiguity in how this is stated in the article.   Gain is an important property of TIAs and it is always stated in Ohms.   Open any good book on electronics, like the ones I consulted in writing this page, and you will see the terms gain and amplification used the same way I have used them here.  "Assuming an ideal (zero output impedance) ampli..."  It appears you don't understand how op-amp circuits operate and very little about TIAs.   "Current is not amplified by a trans-impedan..."   Current is amplified by a TIA, but the output is a voltage.  The gain has units of Ohms.   Since you can't comprehend this concept I suggest that you don't try editing electronics pages.  Zen-in (talk)
 * Even if people knew what you meant when you say the wrong thing, it is not an excuse to say the wrong thing. TIA's do not amplify current. They convert current to voltage.Constant314 (talk) 04:18, 11 July 2014 (UTC)


 * I read your reference. As long as the sentence says "amplify the current output ... to a voltage" it is acceptable.Constant314 (talk) 22:28, 11 July 2014 (UTC)

Inverting configuration
I removed all this today because it is absolute unmitigated nonsense. There is no reason to ever want a "none (sic) inverting configuration" of a TIA since an inverting or non inverting follower is usually added. This was not an example of a non inverting amplifier anyway. Incrementally the circuit is still an inverting amplifier because the feedback is to the inverting terminal. So all that nonsense has been reverted.Zen-in (talk) 16:51, 5 July 2015 (UTC)
 * I agree with you. The "none inverting configuration" was non-sense.Constant314 (talk) 18:27, 5 July 2015 (UTC)

DC Operation load
It says, "In the circuit shown in Fig. 1 the photodiode is connected between ground and the inverting input of the op-amp. The other input of the op-amp is also connected to ground. This provides a low impedance load for the photodiode, which keeps the photodiode voltage low." As an ideal op-amp has infinite input impedance, how can connecting the photodiode between ground and the inverting input of the op-amp provide a low impedance load for the photodiode? Seems that it would provide a very high impedance load. The actual current path would be through the feedback resistor to the (ideally) zero impedance output, but the impedance of that path depends on Rf. Stevensrmiller (talk) 19:55, 1 June 2016 (UTC)


 * It is correct. Feedback action holds the inverting node voltage close to ground, regardless of the photo current.  You can think of dynamic resistance, a large change in photo current produces a small change in voltage at the inverting input, hence a low impedance.  Actual impedance would be on the order of Rf/A where A is the voltage gain of the opamp.  So if A= 1,000,000 and Rf = 100k then the input impedance would be on the order of 0.1 ohm. Constant314 (talk) 20:11, 1 June 2016 (UTC)

MOSFET and Bipolar TIA additions
This appears to be largely theoretical. There are no references for this idea. A single transistor stage does not provide enough gain for transimpedance to be realized. This recent addition should be removed. There are a few other issues with this page I want to address shortly. Zen-in (talk) 14:55, 21 August 2018 (UTC)
 * I agree. Wikipedia is not a text book.  The sections have a lot of tedious math and no citations.  They are poor trans-impedance amps.  They don't bring anything to the article that is not covered elsewhere.  Constant314 (talk) 17:12, 21 August 2018 (UTC)
 * The intention of the edit was that the article explains in general what a transimpedance amplifier is and shows some circuits how to build one. Before the edit the article was very restricted to one particular circuit (opamp with single resistor as feedback) and one particular application (fibreoptics receiver). However I agree that the article as it is now is not optimal (missing references, too many formulas). Fvultier (talk) 10:40, 23 August 2018 (UTC)
 * There are several references in this article related to the technical explanations. While some people would like to see no equations, Laplace equations and bode plots are generally accepted and are excellent short cuts for characterizing electronic circuits.  Op-amps, because they have so much gain, are almost like ideal mathematical building blocks.  They are described with real-value mathematical functions and a roll-off slope for frequency considerations.   Your change that showed the transfer function of a TIA in terms of a theoretical complex function was not required.   There were no references for the transistor TIA ideas.  A TIA could be built with FETs and BJTs, but the circuit would have to have a lot of gain.  So, in effect, you would be building an op-amp from discrete components.  The BJT circuit you showed has the problem of very low reverse bias on the photodiode and an unbalanced circuit using an unmatched NPN / PNP pair.  The low reverse bias would result in a very low frequency response because of the high junction capacitance.  The MOSFET has an even worse problem with frequency response because MOSFET gates have very large capacitance compared to the photodiode junction.   Also, the N-channel and P-channel MOSFET gates are tied together and only one will turn on.   I would be interested in reading a paper or app-note describing these ideas more, but I think they are just theoretical and therefore unencyclopediac.   Single stage differential amplifiers and cascode circuits have been used with photodiodes.  They are not called transimpedance amplifiers.Zen-in (talk) 05:17, 24 August 2018 (UTC)
 * The 2 references you added do not appear to be relevant. One is in German, which most people on this site wouldn't be able to read and could not access, and the other only has a couple of pages on the general subject of TIAs.   A want to see an article or appnote I can read that supports this theoretical idea.  I don't believe either circuit would work as you claim.  That content is POV and does not belong in an encyclopedia.Zen-in (talk) 15:03, 24 August 2018 (UTC)
 * I deleted these sections because, as stated above 1)The example circuits do not work 2) There are no valid references. One reference is in German and the other is just a general electronics book. I have not been able to find any paper that describes where a TIA has been built using MOSFETs or bipolar junction transistors.  I have asked the editor that added this theoretical material to supply better references and he has not done that.   TIAs almost exclusively use op-amps.  Zen-in (talk) 16:53, 9 September 2018 (UTC)
 * I support that. WP is not a textbook, reference book or catalog of circuits. The opamp circuits are fully sufficient to illustrate the article.Constant314 (talk) 00:30, 10 September 2018 (UTC)

Possible error in section Bandwidth and Stability
In the literature known to me as well as on Wikipedia (see Electrical reactance) the reactance of a capacitor is a real quantity and not an imaginary. So the following equation from the article is wrong:

$$\beta = \frac{X_{C_\text{i}}}{R_\text{f} + X_{C_\text{i}}} = \frac{1}{1 + R_\text{f} C_\text{i}s}$$

I propose that we change the word reactance in the text back to impedance and the X in the forumlas back to a Z. Do you agree? — Preceding unsigned comment added by Fvultier (talk • contribs) 15:26, 23 August 2018 (UTC)
 * That equation was copied from Graeme's book.  However it is incorrect because one side is real and the other side is complex.  That mistake is not apparent in his book because that equation is a small part of a larger explanation of the stability analysis of transimpedance amplifiers.  There is no requirement for complex math equations because Laplace transforms and Bode plots are used instead.   I will edit out the complex function since it does not add to the explanation at all.Zen-in (talk) 05:24, 24 August 2018 (UTC)
 * This is a step in the wrong direction.  Beta is complex and becomes almost purely imaginary at high frequency, with phase approaching -90 degrees.  It is this phase shift added to the opamp's phase shift that is responsible for the peaking.Constant314 (talk) 14:20, 24 August 2018 (UTC)
 * Disagree with "There is no requirement for complex math equations because Laplace transforms and Bode plots are used instead." If we consider Laplace transforms we always need to care about math involving complex numbers. Bode plots show magnitude and phase of a complex quantity. There is nothing wrong with a complex beta. All reactances X should be replaced by impedances Z. Fvultier (talk) 14:39, 24 August 2018 (UTC)
 * I agree with that.Constant314 (talk) 14:42, 24 August 2018 (UTC)
 * The frequency-dependent term 1/$$\beta$$ is the slope of the stability intercept line in the Bode plot, figures 4, 5. I explained earlier that Bode plots do not use imaginary numbers.  $$\beta$$ is derived from the incremental model.  It includes reactances which change with frequency. Bode plots have been used for almost 90 years to analyze the stability of feedback systems.Zen-in (talk) 14:56, 24 August 2018 (UTC)
 * The full Bode plot includes a phase plot. The reason that you can sometimes get away with just the magnitude plot is by making assumption about phase.  Without accounting for the phase shift in beta, you won't get that pronounced peaking with is caused by the phase margin getting close to zero.  Constant314 (talk) 15:04, 24 August 2018 (UTC)
 * The stability analysis in this article was lifted almost verbatim from Graeme's book on photodiode amplifiers, although without as much detail. Bode plots do not use complex parameters.  They are log-log plots with frequency (in radians/Sec) on the X axis. If a different method of stability analysis is proposed, it should have equally qualified references.  I suggest people read the article and try to understand this well established method of stability analysis before doing any piece-meal changes.Zen-in (talk) 16:40, 24 August 2018 (UTC)
 * Are you saying that Graeme has $$R_\text{f} + X_{C_\text{i}}$$, a resistance added to a reactance without a j factor in front of the reactance? Constant314 (talk) 16:51, 24 August 2018 (UTC)


 * I found the book on Amazon and using the search feature I found $$\beta = \frac{1}{1 + R_\text{f} C_\text{i}s}$$ in the text on page 39. The figure just below that text has the equation


 * $$\beta = \frac{X_{C_\text{i}}}{R_\text{f} + X_{C_\text{i}}} = \frac{1}{1 + R_\text{f} C_\text{i}s}$$.


 * However, it is mathematical nonsense. The ratio of two real numbers cannot produce a complex result.  (the s in the right hand expression is complex).  The factor -j has been omitted.  The proper mathematical equation would be $$\beta = \frac{-j X_{C_\text{i}}}{R_\text{f} - j X_{C_\text{i}}} = \frac{1}{1 + R_\text{f} C_\text{i}s}$$.


 * This omission might be intentional on the grounds that it did not make much difference to the magnitude plot or it might be an unintentional error.
 * On page 41, Graeme clearly states the both $$ = \frac {1}{\beta } $$ and $$ = A_{OL} $$ have a phase shift of about 90 degrees, hence they must both be complex.
 * I propose that we just use $$\beta = \frac{1}{1 + R_\text{f} C_\text{i}s}$$, which is what Graeme says and we all, I think, agree is correct.Constant314 (talk) 22:05, 24 August 2018 (UTC)

I think what Graeme is doing with the equation no one likes is to show the equivalence of the feedback across Rf and Ci as a frequency dependent resistive divider and as a frequency domain equation. If you just took the real part of both they would be the same. A Bode plot uses frequency domain equations (Laplace Transforms) so I agree the right formula there is $$\beta = \frac{1}{1 + R_\text{f} C_\text{i}s}$$ Zen-in (talk) 03:10, 25 August 2018 (UTC)

2 paragraphs that repeat themselves in DC Operation section
" The transimpedance amplifier presents a low impedance to the photodiode and isolates it from the output voltage of the operational amplifier. In its simplest form a transimpedance amplifier has just a large-valued feedback resistor, Rf. The gain of the amplifier is set by this resistor and has a value of −Rf (because the amplifier is in an inverting configuration). There are several different configurations of transimpedance amplifiers, each suited to a particular application. The one factor they all have in common is the requirement to convert the low-level current to a voltage.

The gain, bandwidth, as well as current and voltage offsets, change with different types of sensors, requiring different configurations of transimpedance amplifiers. "

This text is already present in the introduction of the article. In case that it is voluntary, i'm not deleting it myself. The two sources [3] and [5] are also the same. — Preceding unsigned comment added by The quantum panda (talk • contribs) 10:23, 29 April 2019 (UTC)
 * The duplicated paragraph was added about a year ago when someone added a section about theoretical bjt and MOSFet TIAS.  Most of that was removed but the duplication was overlooked.   Thank-you for bringing it to everyone's attention. Zen-in (talk) 16:43, 19 June 2019 (UTC)