Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
pixel said:It can be generated by current mirror in differential amplifier... also in the case of pole-zero cancellation, when components are not perfectly matched.
The effect of pole-zero doublet could be not clearly seen in the amplitude and phase response, especiaqlly if the case if they are closed together, but relativlly small mismatch can have high impact on time response.
For step response in the case of zero-pole doublet (especially in area around unity gain frequency) is typical one fast part and one much slower (undershoot), which can significantly degrade settling time.
In the case of pole-zero doublet there is an overshoot with also slow settling time.
meghna said:(1) What is the relation between phase margin and settling time? From rajavi (page 354), I can concldue that better the phase margin, slower is the response or settling time is more. Correct me if I am wrong!
(2) Now if we are talking about pole-zero doublet, we should specify whether it is right half or left half plane zero. In diode connected load, it is a left half plane zero. As I know, left half plane zero increases the phase margin.
Is this the reason why you are saying that doublet causes large settling time?
(3) If it the reason, then pole-zero or zero-pole both will result in good phase margin (bad settling time), but zero-pole doublet will be worse in terms of settling time. Correct me if I am wrong!
meghna said:This is a very good point that settling time is equal to settling time until peaking occurs.
That means, lower the phase margin, lower is the rising time. (is it true for any number of pole system? someone told me that these theories work only for two pole system)
Now settling requirements may depend upon the user. So the question comes who decides the amount of overshoot. Can you suggest some good paper where I can see these calculations.
Thanks a lot!
youyang said:this paper would be helpful
B.Y ESHWANT KAMATH,"Relationship Between Frequency Response and Settling Time of Operational Amplifiers"
pixel said:It is true for standard two pole system until Q factor reach value of 0.7(Butterworth response). For higher peaking you would have faster rise time, but slower settling time(approx Q periodes of osillations).
Here are two unity gain closed-loop response plots for pole-zero (wz=2wp), and zero pole (wz=0.5wp), where N is number of decades from unity gain frequency. You could se that the biggest problem is around unity gain frequency, and that overshoot does not mean fast settling time.
jiangnancai said:So what's the difference between the pole-zero doublet and zero-pole doublet?
If Wz>Wp,is it called pole-zero doublet,and vice versa?
Other question is that in my impression,Q factor is a factor to describe AC frequency response.But for settling time,it is a time domain response, so is it reasonable? Thanks
LvW said:jiangnancai said:So what's the difference between the pole-zero doublet and zero-pole doublet?
If Wz>Wp,is it called pole-zero doublet,and vice versa?
Other question is that in my impression,Q factor is a factor to describe AC frequency response.But for settling time,it is a time domain response, so is it reasonable? Thanks
1.) In short: We have an "doublet" if we introduce a zero Fz1 and a pole Fp2 with the aim to cancel another pole Fp1 - and if Fz1 does not meet Fp1 exactly !
If Fz1>Fp1 the step response will have a "long tail" slightly above the final value and if Fz1<Fp1 the tail will be slightly below the final value.
2.) The Q factor of a pole pair is a measure for the ac response in the transition region around the pole frequency - and at the same time Q detrermines the step response in the time domain. For Q>0.5 we will have a an overshoot which increases and turns into ringing for higher Q values.
Does it help ?