confused about feedback (thinking in time domain)

[suifengpiaoyang]

sometimes I was thinking about feedback circuit, And got some questions about that
Imagine that: A system A with feedback B is shown as below, The system output Vout and the feedback system output VFB.
Assume that the system is stable which means the output Vout is stable before time 0ms.Suddenly the Vout has a puse(the lengh of pulse is 20ms) due to some reasons, And then the feedback system detecs this change and after 20ms delay of system B the output VFB starts a nagtive pulse to stop the chage of Vout.But due to the delay of system A, The VFB signal reach to Vout after 10ms,which of course is late. And in this way,the whole system will not be stable anymore!

So I am confused about feedback system,how dose it make system stable?Or am I thinking in a wrong way?

 

[Chipmunk]

There is indeed a danger that feedback systems may become unstable if not designed carefully. You are right that it's the delay that may cause the instability, but the best way to investigate stability issues is to move to the frequency domain. Google Bode Plots and study them carefully, they are the key for understanding.

 

[LvW]

sometimes I was thinking about feedback circuit, And got some questions about that
...............
So I am confused about feedback system,how dose it make system stable?Or am I thinking in a wrong way?

I think, your description of the sequence is realistic. The output pulse you have mentioned may, for example, occur as the result of an input signal.
In this case - as described by you - it will take some time until the negative feedback network does react and brings the output back to the linear region.
Examples:
* This can be observed for opamps with (negative) feedback. In case of a sufficient large input step the time needed for the feedback signal to arrive at the input results in a drastically increased rise time at the output known as "slew rate effect".
* For lower step amplitudes the amplifier remains within the linear range but may show a kind of "ringing (overshoot)" at the output. In the frequency domain this effect is quantified by a parameter called "phase margin". In the time domain this effect is caused simply by the delay of the feedback signal.
* But note: In both cases we do not speak about "instability". If the system returns back to a stable operating point it is a stable system - perhaps with a stability margin that is too small for certain applications (where no overshoot is allowed). Thus, also the case as described by you does not indicate instability in general.
* However, this feedback delay is a major issue in control systems - in particular the delay caused by (perhaps) unknown "dead times" within the closed loop. Indeed, this may lead to instability of the whole system.

LvW

- - - Updated - - -

Supplement:
Regarding your last sentence ("how does it make system stable") - you must discriminate between static (dc) and dynamic stability.
Negative feedback always improves static stability (operating point) but - at the same time - decreases dynamic stability.

 

[suifengpiaoyang]

There is indeed a danger that feedback systems may become unstable if not designed carefully. You are right that it's the delay that may cause the instability, but the best way to investigate stability issues is to move to the frequency domain. Google Bode Plots and study them carefully, they are the key for understanding.

Thank you！
I do know that moving it to the freq. domain will help us understand something,But ,after all,freq. domain is virtual.Time domain is something that give us a vividly explanation.