Hi

Trying to figure this regulator out (see link). But I have difficulty with "compensation network" for "loop stability" (pin 18 "COMP" and page 39 in the datasheet)

What exactly is the "compensation network"? what does it compensate for? how does it work?

Anyone that can help me out with understanding this concept?


https://www.ti.com/lit/ds/symlink/t...ed-dc-dc-switching-regulators%2Foverview.html

 

https://www.analog.com/media/en/technical-documentation/application-notes/AN149fa.pdf

It's like correction (stability control) for over-steer while driving a car.

 

The dynamics of any control loop are governed the components involved. The purpose of "compensation" is to avoid building an oscillator from an amplifier. This can happen when the loop gain exceeds unity AND the phase shift is any multiple of 360°. This is usually accomplished by adding phase shift in the vicinity of the unity gain crossover.
N.B. This explanation won't mean much if you have not studied Laplace Transforms and Differential Equations. If that is the case, just tuck it your back pocket and pull it out at the appropriate time.

 

ah.. so it is for dampening the speed of the voltage correction (when a change in the load occurs) to counteract an overcompensation and "over steering" of the regulator output (that could result in oscillation)?
am I understanding the idea correct?

 

ah.. so it is for dampening the speed of the voltage correction (when a change in the load occurs) to counteract an overcompensation and "over steering" of the regulator output (that could result in oscillation)?
am I understanding the idea correct?

That is essentially a good heuristic explanation. It kind of depends on the operational mode of the converter and the control methodology. Some converters do control the the voltage; they do not exhibit the same set of problems that you see in converters that control the maximum current in the inductor. Also I'm not sure on this converter if its is CCM or DCM. A CCM converter with max current in the inductor being controlled is where the greatest need for compensation occurs. This compensation takes the form of slope control for charging and discharging the inductor to prevent sub-harmonic oscillations. This is something that you don't see in linear analog systems.

 

ok, I see.
A follow-up question: what is the advantage/disadvantage of controlling the voltage or controlling the current? should they not all need to control the voltage? or how to otherwise know if Vout is correct? how to do this if they only control the current? or is that they control the voltage AND current? or?

 

ok, I see.
A follow-up question: what is the advantage/disadvantage of controlling the voltage or controlling the current? should they not all need to control the voltage? or how to otherwise know if Vout is correct? how to do this if they only control the current? or is that they control the voltage AND current? or?

If you read the datasheet carefully you will see that there are multiple nested control loops. They are all working in concert in parallel. The main reason for adding current control loops is that the response to a load transient is faster and it reduces the order of the system. Reducing the order of the system is part of the strategy to ensure there are no cases where oscillations can occur. I've been around for half a century and I have not yet seen a first order system that will oscillate.

 

OK, I got to understand the issue a bit better.
Thank you (both) very much for your help. Really appreciated.