Take a look at http://www.designers-guide.org/Forum/YaBB.pl?num=1182388268, perhaps your circuit has a similar behavior.

icmeiying wrote on Jan 20^th, 2010, 8:31am:


No, I don`t think so. The Bode plot gives you only information on the small signal response of the system.
When your circuit response differs between small and large input signals, I don`t think that this will be caused by pole positions.

Hi,
     You could try this approach -- Obtain Bode Plots for min, max & center input CM levels corresponding to your large signal swing.
  If any of these Bode Plots indicate instability, then you can have problems in large signal swings.

Bode Plots can be trusted only if the phase function's monotonic.

--
Mayank.

hi buddypoor, raja.cedt and Mayank
   Thanks a lot for all your suggestions! Today I replaced a pmos transistor that works in the linear region with a real nulling-zero resistor during the miller compensation, and found that everything was OK including the large signal response. The value of miller capacitor used by a pmos did not change.
   To be honest, I was still a little confused about this phenomenon.

best wishes

icmeiying

raja.cedt wrote on Jan 21^st, 2010, 3:50am:



Hi Rajasekhar,
Integrator "wind-up" is a phenomenon which can be seen in control systems which have an integrating controller. Very often, the actuators driven by the controller (electrical,mechanical,...) come to their limits although the integrating controller further integrates. Now, when the the error signal changes its sign (due to the control action) the integrator must be discharged up to a certain degree before the actuator can respond to the new condition. Thus, the integrator has itself wound-up. This leads to a delay within the loop and causes overshoots.