Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Is it stable that phase drop to close to 0 but rise to 75 degree when gain is 0dB?

Status
Not open for further replies.

shanmei

Advanced Member level 1
Joined
Jul 26, 2006
Messages
430
Helped
8
Reputation
16
Reaction score
8
Trophy points
1,298
Location
USA
Activity points
4,496
The phase margin is 72 degree when the gain is 0dB. However, the phase margin is close to 0 before the frequency reaches the unity-gain bandwidth.

Is the op-amp stable? Why? Thank you.
stability.jpg
 

It may not be. Two ways to know.
1. generate a Nyquist plot.
2. Run a transient analysis to check the step response for oscillations.

Also, you may try to move the zero towards lower frequencies and thus not let the phse drop so much.
 
Why Phase starts from 180 degree ? Is it an Inverter Amplifier ??
If it's so, the danger is much behind from where the Gain drops down to 0.
While the Gain is higher while the Phase is close to 0, so it will close to oscillate..
 
The bode plot is simple enough (monotonous gain, only one crossing of the 0 dB line) to see that the circuit is unconditional stable. Nyquist criterion can be applied, but it's not necessary to assess stability in this case.

The phase margin is 72 degree when the gain is 0dB. However, the phase margin is close to 0 before the frequency reaches the unity-gain bandwidth.
The correct statement is: "The phase margin is 72 degree". Closed loop phase approaches 0 degree, not phase margin which is only a single value.

Although the amplifier is stable, the "sagging" phase will probably show in the transient response.

Why Phase starts from 180 degree ? Is it an Inverter Amplifier ??
The phase of a negative feedback loop always starts at 180 degree.
 
On the face of it - it is quite stable - adding other delays to bring the total delay to 360 deg is where it will start to oscillate - then the feedback reinforces the input - at 180deg the feedback still subtracts from the input ...
 
The effect of the 2-pole/1-zero phase on closed loop transient response (red) can be seen in comparison with a regular 2-pole amplifier (cyan), both circuits have similar phase margin and gain transient frequency, also shown a 1-pole amplifier (green).

loopgain.PNG closedloop.PNG
 

Attachments

  • closedloop_2p1z.zip
    2.1 KB · Views: 87
  • Like
Reactions: asdf44

    asdf44

    Points: 2
    Helpful Answer Positive Rating
Thank you all of you. Yes, it is an inverting amplifier. Thank FvM to give the simulation comparison.
 

Is this a fixed frequency inverter (50 or 60hz for example)? If so PR (Proportional Resonant) control is another good option.

PR control adds a tuned resonant element to the feedback that works like an AC integrator. Any error signal at the tuned frequency is amplified creating very high gain at a that single frequency.
 
It is an opamp, not a fixed frequency inverter. Thank you for the information.
 

Can I ask what the application is? I’ve experimented with this compensation scheme but have only recently found other examples of it in custom audio opamps.

I speculate that putting a limit on the “extra” pole will limit the large signal impact while maintaining the small signal benefit for locking in output accuracy at mid range frequencies.
 
It is a tranimpednace amplifier (TIA) used for converting the input current into output voltage.

There are two poles, P1 and P2, and a zero, Z1. P1 is the dominant pole, and Z1 samller than P2. Z1 compensates P2.

I am not sure about the influence of the poles and zero to middle range frequency signal.
 

I am not sure about the influence of the poles and zero to middle range frequency signal.
Run an .AC simulation in closed loop to see the effect. You get about 1.6 dB gain peaking in the 1 - 3 kHz range, corresponding to the overshoot in pulse response.
 
I was implying that the added open loop gain in the 1-100hz range would result in lower closed loop error. Looking at these closed loop simulation that's not true however due to the early onset of the peaking. The peaking means the phase delay stays lower at higher frequencies in case that matters.

I expect that if the extra pole is moved left away from crossover the peaking will be reduced.
 
Oh, I see. Yes, the higher gain leads to an accurate settling, and the seperate pole improves the phase margin.

The plots from FvM are used to illustrate that it is still stable even if the phase drops close to zero before the unit gain bandwidth given that the phase margin is large enough at the unit gain bandwidth.

Thank you.
 
  • Like
Reactions: jjx

    jjx

    Points: 2
    Helpful Answer Positive Rating
The phase margin is 72 degree when the gain is 0dB. However, the phase margin is close to 0 before the frequency reaches the unity-gain bandwidth.

Is the op-amp stable? Why? Thank you.
View attachment 157193

When the phase shift is close to 180 degrees, you still have a lot of open loop gain. This means that the denominator of G/1+GH can still be written as ~|GH| which would mean that the transfer function can be simplified as ~ 1/H. Of course there can be a lot of exceptions. But in most cases, it means that the closed loop transfer function does not become infinity at a frequency and therefore it is stable.

Only where the loop gain comes close to 1 (or 0dB) and the phase shift is close to 180 degrees, will there be a chance that the closed loop TF goes to infinity and therefore you will have oscillations.
 
When the phase shift is close to 180 degrees, you still have a lot of open loop gain. This means that the denominator of G/1+GH can still be written as ~|GH| which would mean that the transfer function can be simplified as ~ 1/H. Of course there can be a lot of exceptions. But in most cases, it means that the closed loop transfer function does not become infinity at a frequency and therefore it is stable.

Only where the loop gain comes close to 1 (or 0dB) and the phase shift is close to 180 degrees, will there be a chance that the closed loop TF goes to infinity and therefore you will have oscillations.

Wow, that is a great explanation for this frequency response. Thank you.

You are right, the key is whether the denominator of the loop gain becomes to be zero.

And the gain is not the opamp gain, it is the loop gain.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top