Calculation PID for Boost Converter

[amg66]

Hi,
What is the good of for boost converter PID or PI for closed loop?

I want the value of PID or PI im using Matlab Simulink

How can Calculate PID value like Ki, Kp, Kd?
if the input Vmin= 30 Dc
Vout= 116 Dc
if i want constant for 120 DC
what is the Ki=
Kp=
Kd=
Revrnce block=

Thank you.

 

[Warpspeed]

Its not that simple, because a boost converter can run either in continuous or discontinuous mode of operation, and can also have a right half plane zero.
You will need to either build it or model it, then do some dynamic testing under the expected extremes of operation.

It will very likely require a very slow integral control loop for unconditional stability.

 

[KlausST]

Hi,

This has been discussed many times here.

Look for "PID ziegler nichols"

Klaus

 

[asdf44]

Also be aware that though there are some commonalities there are no standards in how those coefficients are defined and there are multiple ways to arrange the P, I and D blocks. Though different sources or programs may use the same names don't assume they're defining them the same way or that you can plug exact numbers from one to the other.

Ideally you should model your system and find its transfer function. Then decide how you want to compensate it. Then choose a control structure and coefficients to achieve that. Are you even sure you need PID rather than PI for example, or Type II compensation ("modified PI")?

I recommend PSIM as a great tool for doing power topology and control design. PSIM has a range of PID blocks and good ac analysis.

You can also do effective PID control design in LTSpice if you become comfortable working with S domain transfer functions. Converting PID into an S domain function and running S domain transfer functions in LTSpice are both things that should be easy to google.



Another quick comment that's not easy to figure out on your own is that compensation is more about performance than it is about stability. If you don't have high bandwidth requirements you might not need a P or a D term at all. A sufficiently low frequency integrator (pole) will be stable as Warpspeed also implied.

 

[Warpspeed]

There is more to this than just plugging some simple numbers into a simulator or formula and getting a result.
That usually works for most process control, but there can sometimes be some complicating problems.

How do you deal with a closed loop system that has more than one mode of operation with vastly different transfer characteristics, that suddenly switches back and forth between modes ?

How do you deal with a non linear system, where one or more of your "constants" is actually a highly non linear variable ?

How do you keep a system stable that first corrects in one direction before it finally corrects in the desired direction as a normal part of its operation, where the initial response to change is positive feedback ? Think very seriously about that, because that is exactly what continuous mode boost converters do. And it does not show up on a Bode plot, because its a transient dynamic condition.

There is a lot more going on here than simple classical closed loop theory might suggest.

 

[schmitt trigger]

The dreaded RHP-zero.

 

[Warpspeed]

The dreaded RHP-zero.

Yes indeed, and I doubt if any circuit simulator is likely to be able to duplicate it.
It does not exist with steady state phase and amplitude circuit analysis.

To unleash the whole terror, you need an actual functioning circuit.

 

[asdf44]

A RHP Zero is another thing that's only relevant if someone is pushing their control bandwidth up near their switching frequency. If loop bandwidth is sufficiently slow (~<1/100th of switching), the RHP Zero can safety be ignored.


The RHP Zero is a simple result of the low side position of the main switch - when it turns on output voltage actually goes down briefly though ultimately a longer on-time results in higher output voltage.

Any simulator that's modeling the operation of the basic topology including the inductor, switch and diode will fully model that behavior in a transient sim like a step response. AC analysis may be a different story but you can validate your control design in time domain transient sims, including the RHP Zero, with only the usual simulator caveats.

 

[CataM]

First find out the Transfer Function of your model. Once you have it, you do not insert a PID because sounds good and everybody talks about it, instead, depending on your transfer function, you start with P regulator, then PD, if the 1st one does not suit your need, then PI regulator if the 2nd one does no suit your need and finally with PID if the 3rd one does not suit your need.

Simulink has a great toolbox which adjust P,PI,PD,PID regulator very easily (one can adjust the regulator with that toolbox at the same time is drinking a coffee and reading the newspaper) once you have a transfer function.

 

[Warpspeed]

If you keep the loop bandwidth much lower than the switching frequency, and use the smallest tolerable amount of inductance in your boost inductor, it should work fine.

I would start off with an Integral only controller with a single dominant pole. Provided it slews slowly enough it will be stable.
Adding any proportional gain to that can be counter productive, because it amplifies the initial complete phase reversal, but try it and see.
Derivative action simply has no place here, and would guarantee unmanageable instability.

If you really do need much wider loop bandwidth than this simple integral solution provides, keep your boost converter in discontinuous mode right up to full power.
The peak currents and conduction losses will be horrific, but if the total power output requirement is low, it can be very simple solution to the whole control loop problem.