Is this multistage OPAMP stable?

[duet12]

Hi

The opamp is a 3-stage one. As shown in the figure, the phase shift is first negative -90 and then shoot up to become positive phase shift (I think is due to Left Half Plane zero).

As I understand, as long as the phase shift is not passing 180degree either way(either positive or negative), the OPAMP is stable. But I never see any published paper which have such kind of phase plot, so I am wondering maybe I am wrong.

Could somebody explain if this OPAMP is stable?

Thanks for your help.

duet12

 

[matchasm]

In your plot, there is an amplitude peak around 0dB, which indicates bad stability properties. Even if the thing is stable, you will probably see tremendous ringing in the transient behavior.

The phase plot does look a bit weird. Do you have a schematic?

Do you intend to use the amplifier in a feedback loop?

 

[wjxcom]

I do not think this opa is stable, at the same time, you said thisis a 3-stage opa, so I think the gain of this opa -- 80dB is not normal

regard。

 

[hr_rezaee]

Hi
is stable because
in 0dB gain your phase margin is -225+360=135
there for is stable.

 

[caosl]

I think the feedback circuit which use this opamp will not stable. In fact the phase margin in you plot is not enougn seriously.You must first eliminate the peak in you mag response.

 

[culho]

I've had similar phase responses and although the phase margin was good, the transient simulation showed clear oscillations (saturated on VDD-VSS).

After many hours reading literature and analyzing the circuit extensively, i found the root cause.

First, if you have a right halp plane pole, the circuit is unstable as its exponential response is unbounded. You can try to place a right half plane zero on it, but across corners you will end up with a zero pole pair on the wrong side....

Second, when stabilizing your opamp, if it has more than 2 two stages and clever compensation schemes are used (neste miller, hybrid miller) make sure its individual loop is stable and that when put together they still are stable.

In my case, the instability was caused by a parasitic positive feedback loop from the output to the biasing of the differential pair (clever architectures to improve performance).

To sum up, whenever you see a phase response like that, there is a positive feedback somewhere. Be careful.

Regards

 

[LvW]

Yes, I agree with the explanations above.
In general, this proofs that the simulated magnitude response alone cannot reveal possible instabilities. The reason is: AC analyses are small signal analyses based on ideal conditions (no power switch-on, steady-state, 100% constant power supply, no noise).
Therefore, one should carefully study the phase response which can indicate the trend to instablity. And, of course, transient analyses are the best way to analyze stability properties.