The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> TIA stability
https://designers-guide.org/forum/YaBB.pl?num=1271387740

Message started by Dipankar on Apr 15^th, 2010, 8:15pm

Title: TIA stability
Post by Dipankar on Apr 15^th, 2010, 8:15pm

Dear All,

I have a very basic differntial TIA - an differential OPAMP with resistor (Rf) feedback (shunt-shunt). To simulate I use "Iac" with ac mag =1 and see the vout. Te transfer function I am interested is H(f)= (VF(/outp)- VF(/outn))/(IF(/inp) -IF(/inn)). Now in low freq H(f) is reasonably starting from 20dB(Rf) but then at a certain freq higher than OPAMP's 1st pole the H(f) is going upwards ( an RHP zero ??? ) only to come down at a very high frequency when phase(H(f)) has crossed -180 - depicting an unstable system. However, the OPAMP is itself stable having a PM > 65. So something going wrong with the simulation set up ? I have never come across a simulation bench for TIA stability. Hope you can help.

Title: Re: TIA stability
Post by aaron_do on Apr 16^th, 2010, 7:45pm

Hi,

I think you should be looking at the loop gain rather than H(f) to see the stability. Also, what do you mean by "the opamp itself is stable"? You need to put the op-amp inside the loop, and then break the loop to check the loop gain for stability.

One more thing, since you are designing a TIA, I assume you are using an ideal current source at the input. I suggest you make sure your source is modeled correctly as it will affect your stability.

cheers,
Aaron

Title: Re: TIA stability
Post by Dipankar on Apr 16^th, 2010, 8:51pm

"OPAMP is itself stable " means : I check the loop gain with stb analysis and it's stable.Also I use the same OPAMP in some other application as voltage amplifier and its T(f)=V(out)/V(in) is stable.

Yes, correctly modeling the TIA input current source is important.

But can you explain why I need not bother about V(out)/I(in) stability ?

Title: Re: TIA stability
Post by aaron_do on Apr 17^th, 2010, 7:47am

Actually my comment was that I don't think you can see the stability directly through H(f). I may be wrong though... ;D

Anyway I think you should be looking at the loop gain, and i'm not sure how your loop gain can be a V/I quantity. Also H(f) is the closed loop response, so I'm a little unsure how you concluded that the system was unstable just by looking at H(f).

BTW you said you check the loop gain with stb analysis and it's stable. When you ran the stb analysis for the op-amp, you included the feedback network in the loop right? Just making sure...

Title: Re: TIA stability
Post by Mayank on Apr 17^th, 2010, 8:58pm

Hello Diapankar,
I think you are simulating a Trans-Impedance Amplifier.
H(f) is nothing but the Impedance offered by the block & as you correctly put it, it's Vout_DIFF / Iin_DIFF .
The reason that it's a Trans-Impedance Block is because you are tapping you output in the form of a Voltage.
Loop Dynamics will be decided by the way the Opamp sees it's input.
When inside the loop, the opamp's only function is to keep Vin+ & Vin- voltages at it's input terminals very close to each other, so as to provide a finite output Voltage.
So, for stability analysis, you better look at the Loop Gain Plot rather than the V/I Plot.

Also, as aaron pointed out, feeding inputs to your ckt is also v imp.
I think the best way is to use an ideal balun and give some ac current as well as DC current & generate some DC bias at input nodes through a Resistor to Ground.

thanks,
Mayank.

btw, returning to Designer's Guide after a long break. Hi all !!

Title: Re: TIA stability
Post by thechopper on Apr 21^st, 2010, 7:27pm

Hi

I agree with Aaron and Mayank....

If you consider Vo/Ii to analyze stability you are still missing the feedback network (Rf)....thus you are NOT looking at the loop gain.
In any case you are looking at the open loop gain (forward path gain), and then to consider for example PM you need to look at the frequency at which the open loop gain crosses the feedback gain level which is Rf (not 0dB)

In addition I do not see how you define in your Bode plot dB since the involved ratio (V/I) is not unit-less....

So if you want to analyze loop gain you either:
1) Convert the input current into voltage and amplify that voltage thru the op-amp...look at Vo/Vi
2) see the output current, treating the opamp as a current amplifier, and analyze Io/Ii

Loop gains have to be unit-less to be meaningful...

Best
Tosei