2
\$\begingroup\$

I am looking at the following figure (Maxim APPLICATION NOTE 5129
Stabilize Your Transimpedance Amplifier):

maxim figure

I just can't understand the following: in the region half-way between fF and fi, the slope difference indicates a 180 degree phase shift between open-loop and the external network, albeit the gain is larger than unity (it is 1 only at the intersection). So, what prevents this compensated op amp from saturating to one rail in this case due to the positive feedback and over-unity gain in some frequency range ? Thanks.

\$\endgroup\$
  • \$\begingroup\$ You're right. I was mis-remembering something. \$\endgroup\$ – The Photon Aug 4 '15 at 4:43
2
\$\begingroup\$

Cipocip - the phase of the loop gain does not assume a value of 180deg between those two frequencies. You must not look at the asymptotic lines only. The real magnitude of the loop gain does not have a roll-off of -40dB/dec. This is true because (a) the open-loop gain of the opamp and (b) the rise of 1/beta both never reach 20 dB/dec (corresponding to a 90deg shift). As you know, each first-order system has a phase shift of 90 deg for infinite frequencies only. In the present case, the situation is even less critical because the 1/beta function has a pole (following the zero). Hence, the maximum phase shift will be (perhaps) only 75...80 deg. - dpending on the distance between both frequencies.

More than that, a clear picture of the stability margin can be gained looking at the NYQUIST plot for the loop gain. The critical point never is encircled.

\$\endgroup\$
  • \$\begingroup\$ Hm, I cannot refute your argument. I wonder if other experts would agree with you. \$\endgroup\$ – cipocip Aug 4 '15 at 8:04
  • \$\begingroup\$ Why not? It is clear that 1/beta NEVER reaches 90 deg. And the same applies for the open-loop gain (if it is app. a 1st order function in the active region). I was teaching feedback techniques and stability problems for 25 years. \$\endgroup\$ – LvW Aug 4 '15 at 8:32
  • \$\begingroup\$ refute=to prove to be false or erroneous; I said cannot refute. \$\endgroup\$ – cipocip Aug 4 '15 at 8:41
  • \$\begingroup\$ OK - you see, I have to improve my english knowledge. However, as I have mentioned - you can prove my statement using the NYQUIST plot and the corresponding criterion (if you trust it). \$\endgroup\$ – LvW Aug 4 '15 at 9:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.