Slewing and linear settling

[parkpika]

"The amplifiers don't linearly settle when there are large signals are present as input."

What does settling linearly mean? Signals settle exponentially because of charging/discharging caps? How is this different from slewing?

 

[Analog Ground]

"Settling" usually refers to behavior of an amplifier (or control system) when the output is near a new value after a transition. "Slewing" usually refers to the part of the transition from the start to just before the output gets near the final value and starts to "settle". When an amplifier is not linear, it is no longer behaving according to the nice linear equations for the amplifier. Here is an example. An amplifier's "slew rate" is it's ability to follow a rapidly changing signal. The larger the signal, the larger the slew rate is required to keep up. It is always seen as a time domain characteristic. The important point: If the amplifier is not fast enough to keep up with the predicted rate, it is no longer a linear device. It violates the predicted behavior. The relationship between the output and the input is distorted in a non-linear way. Because of this, "settling" after exceeding the maximum slew rate can be unpredictable. The output may not be continuous and smooth. Certainly not according to the linear equations. It seems your amplifier has problems with large input signals.

Slewing and settling linearly means an amplifier or control system which uses feedback will behave according to it's linear equations and gain characteristics. Settling linearly is often described in the time domain as undershooting or overshooting the final value. If overshooting, a "ringing" or "damped oscillation" can be observed during settling. The may be a lot going on during settling but the output is continuous. In this context, it is good to have a feel for the basics of feedback control systems.

Here is a previous post on slewing and settling: https://www.edaboard.com/threads/80575/

 

[parkpika]

I'm still confused about settling 'linearly'. Intuitively, when the output is near the final value, it will behave exponentially, getting infinitesimally closer to the final value. How can this behavior 'linear'?

 

[Analog Ground]

Your question is a good one. I think "linear" in the amplifier sense is a term which has evolved over the years and varies in use with the type of circuit. For power amplifiers in audio and RF, people seem to use the term to mean the output is proportional to the input and ideally will be at a constant gain and in phase for all frequencies in the signal. When displaying the input and output on an oscilloscope as an X-Y display, the trace is a straight line ("linear" as in line). If this input-to-output relationship holds for all the frequencies, a square wave in will be a square wave out. A common term used is "distortion" to describe non-linear behavior. Non-linear is a bad thing.

The world of signal processing and op amp based functions is different. Most functions are implemented which, by design, do not maintain a constant gain and phase for all frequencies. There are probably many designers in this field which never use the X-Y mode of an o'scope and don't know how to interpret one. The display will almost never be a straight line. "Linear" has come to mean "continuous", "smooth" or "regular".

"Settling", as defined in the first reply, has to do with the circuit behavior near a new output value. The test used in the signal processing world which is closest to the X-Y display in the power amplifier world is to put a square wave in and see how the output settles to the constant value at the top and bottom of the square wave. The output can approach the value with an exponential behavior (as you mention) or overshoot and settle with an oscillation ("ringing"). As long as the output is continuous and smooth, the behavior is described as "linear". The use of "linear" is not intuitive, as you point out.

I would be interested to hear how others use "linear" and how they think the term has evolved.