Hi all,
I need some validations on my understanding of feedback system stability. Currently, I am trying to replicate the reference design of a transimpedance amplifier (see link below). With most of the resources I have been reading about this topic, stability of a feedback amplifier depends on gain margin and phase margin. Please correct me if I am incorrect on my understanding: the intuitive way to evaluate stability is to look at when the magnitude bode plot crosses the 0 dB line, the corresponding phase should be less than -180 degree -- thus constitute a positive phase margin; also, when the phase bode plot cross -180 degree, the magnitude should be less than 0 dB -- thus giving a positive gain margin.
My simulation resulted in the bode plot below, and both gain/phase margin are negative it seems, thus rendering the system unstable I think. I am hoping my analysis is incorrect and there is an alternative way to assess transimpedance amplifier (current feedback amplifier?). I have seen others mention looking the at the rate of closure, which look at the intersection of open-loop gain (A_OL) and 1/beta. I am still trying to understanding how these values are obtained for a design. I have found some good resource from the E2E forum (list at the bottom) but still having difficulty understanding it. I have attached the TINA circuit file below if anyone is interested in looking at it. Thank you for all the help in advance!
Reference design: http://e2e.ti.com/support/amplifiers/precision_amplifiers/f/14/t/158884.aspx#584443
TINA circuit TSC file: 6523.TunnelingCurrentPreamp.tsc.zip
Resource:[1] Transimpedance Considerations for High-Speed Operational Amplifiers (PDF 193 KB)http://www.en-genius.net/site/zones/acquisitionZONE/technical_notes/acqt_092407
[2] Operational Amplifier Stability, by Tim Green, Linear Applications Engineering Manager, Burr-Brown Products from Texas Instrumentshttp://www.ti.com/general/docs/litabsmultiplefilelist.tsp?literatureNumber=sboa122
Wayne Chu
Hi Wayne,
Although you are close in your understanding in some areas, you are off in others. The stability of a closed loop system requires looking at the loop gain of the circuit which is not equal to the circuit closed loop Vo/Vin transfer function. The Loop Gain is equal to the difference between the AOL curve and the 1/Beta (inverse feedback factor). You will probably recognize this term as: AOL*B.
The system stability is therefore determined based on the remaining phase margin when the AOL*B product is equal to 0dB. Tim Green's paper's which you've reference go into great detail showing how to obtain the AOL, 1/B, and AOL*B curves which can then be used for stability analysis. In short, you need to break the loop around the amplifier and look at the difference between the source input (injected at the point where the loop is broken), the feedback signal, and the output. I suggest you continue your reseach on Tim Green's papers and I'm sure you will figure it out.
Regarding your circuit, there are some obvious problems with your lack of negative feedback on the composite amplifier second stage that requires some heavy re-work. You may have just flipped the terminals on accident so please double check the circuit.
Hope this helps.
Best Regards,Collin WellsPrecision Linear Applications
Regards,Collin WellsPrecision Linear Applications
Thanks Colin,
I am still going through Tim Green's paper. The first two parts of the paper seems to be most relevant in discussing how to obtain the open loop gain (A_OL) and 1/beta value of a circuit. I followed figure 2.3 given in part 2 (p. 3) to change the circuit in order to conduct a loop gain test: 1) grounding the input, 2) place a high value inductor at the output (block low frequency signal), 3) put a voltage source with a high value capacitor at the start of the feedback loop. The paper tests a first-order circuit, now my question is, how would I use the same technique for a n-order multistage amplifier circuit? Do I test each amplification stage separately; if all stages are stable, then I can assume the entire circuit is stable? Or, do I test the entire circuit together?
I attached the modified TINA circuit, hopefully you can tell me if I am doing this correctly. In it, I test for the second stage of amplification alone, where I defined N2(s)/N1(s) as A_OL and N3(s)/N1(s) as 1/beta. Now, if I were to test both stages together, like N3(s)/N0(s), I will get a floating point divided by zero error, most likely because I am not suppose to measure open gain test like this.
TINA circuit with open loop configuration: 8787.TunnelingCurrentPreamp-OpenLoop.tsc.zip
Just curious, I cannot find the rest of Tim Green's paper (part 11 to 15), was that ever made available online?
Thank you again and really appreciate your help!
Wayne
Sorry for the delay.
I believe the majority of your issues lie within the first stage transimpedance amplifier and not the second stage composite amplifier that you've currently broken the loop on.
A 1GOhm feedback resistor is not very realistic, but the zero that it causes with the input capacitance of the amplifier and photodiode capacitance leads to an unstable rate of closure between the amplifier's AOL and the 1/Beta terms. This effect is described in the following articles:
http://www.ti.com/lit/an/sboa055a/sboa055a.pdf
http://www.eetimes.com/design/analog-design/4009969/Understand-and-apply-the-transimpedance-amplifier-Part-1-of-2-http://eetimes.com/design/analog-design/4009970/Understand-and-apply-the-transimpedance-amplifier-Part-2-of-2-
Please see the attached circuit and results for how to perform a stability analysis of your first stage transimpedance amplifier.
6675.Transimpedance - Stability.TSC
Please let me know if there are other questions.