Buck Converter with PID Controller in LTSpice

Thread Starter

Thurg38

Joined Feb 18, 2021
4
Hi all,

I'm designing a Buck DC-DC Converter that is meant to be able to convert from a 50V input, run it out at 0-50V with 5A max current. I currently have the design as below and am having difficulty with the PID controller not giving me anywhere close to my worst case scenario of 25V @ 50% Duty Cycle and my course head has also been slightly stumped by this design. Any suggestions or comments on the design would be very helpful.BuckWithPID.png
 

Papabravo

Joined Feb 24, 2006
16,107
AFAIK there is no evidence that a PID controller using voltage feedback will be able to perform nearly as well as either a CCM or DCM with either peak or average current feedback. The major advantage of these methods is that they reduce the order of the control loop from 2nd order to first order. You should read Basso for an education on these circuits.

https://www.amazon.com/Switch-Mode-Power-Supplies-Second-Simulations/dp/0071823468/ref=sr_1_2?dchild=1&keywords=Basso,+Christophe+Switch-mode+power+supplies&qid=1613678569&sr=8-2

You should also be aware that a duty cycle in excess of 50% presents some other really nasty problems.
 
Last edited:

Papabravo

Joined Feb 24, 2006
16,107
Why so many op amps?
That circuit looks way more complicated then it needs to be.
They are only needed if all three feedback proportionality constants are non-zero. If any constant is zero, that path can be eliminated. I would really like to know why the TS thinks this is a good way to go.
 

Thread Starter

Thurg38

Joined Feb 18, 2021
4
They are only needed if all three feedback proportionality constants are non-zero. If any constant is zero, that path can be eliminated. I would really like to know why the TS thinks this is a good way to go.
Honestly, it was a marking criteria on my assignment however the tutor who set is also unsure of whether it is possible and I wanted to ask people with more expertise than I.
 

Thread Starter

Thurg38

Joined Feb 18, 2021
4
AFAIK there is no evidence that a PID controller using voltage feedback will be able to perform nearly as well as either a CCM or DCM with either peak or average current feedback. The major advantage of these methods is that they reduce the order of the control loop from 2nd order to first order. You should read Basso for an education on these circuits.

https://www.amazon.com/Switch-Mode-Power-Supplies-Second-Simulations/dp/0071823468/ref=sr_1_2?dchild=1&keywords=Basso,+Christophe+Switch-mode+power+supplies&qid=1613678569&sr=8-2

You should also be aware that a duty cycle in excess of 50% presents some other really nasty problems.
Could you expand on the 50% issues? I realise that the error will become smaller the closer it is to the original input voltage and therefore the error becomes smaller and harder to integrate and then differentiate
 

Papabravo

Joined Feb 24, 2006
16,107
Could you expand on the 50% issues? I realise that the error will become smaller the closer it is to the original input voltage and therefore the error becomes smaller and harder to integrate and then differentiate
The phenomena is called sub harmonic oscillation. It occurs when the duty cycle exceeds 50% and causes an instability in the current waveform in the inductor. It is usually dealt with using slope compensation. You don't allow the current in the inductor to reach a peak if the initial value of the current starts at a higher level than you intended. Instability in the current waveform makes regulation difficult.

The problem with controlling only the voltage is that the response time is too slow. Any system with a time delay as an element in the forward path is going to be difficult to control via linear methods. Using current control simplifies your life because it reduces a 2nd order loop into a first order loop. Much easier to deal with.

As a pedagogical exercise go ahead with your research and see what you come up with. I'm guessing you will find that the differential term needs to be small if not zero.

PS -- if you can get away with it substitute an actual switch for the MOSFET(s) and behavioral sources for gain/attenuation blocks. Not sure they'll let you get away with it but the sims will run blindingly fast if you make that substitution.
 
Last edited:

Papabravo

Joined Feb 24, 2006
16,107
I ran across a couple of simulations I did for a buck converter using PI control for voltage and PI for max current with slope compensation. I can post the sim files for you if it will be useful. You're going to have to round up a copy of the LTspice Control Library which I think is available from @Bordodynov 's website, and on GitHub

http://bordodynov.ltwiki.org/
the file you want is l.b.zip and is about 16 MB and it has other stuff in it

Turns out the part you need is reasonable size included below
 

Attachments

Thread Starter

Thurg38

Joined Feb 18, 2021
4
I ran across a couple of simulations I did for a buck converter using PI control for voltage and PI for max current with slope compensation. I can post the sim files for you if it will be useful. You're going to have to round up a copy of the LTspice Control Library which I think is available from @Bordodynov 's website, and on GitHub

http://bordodynov.ltwiki.org/
the file you want is l.b.zip and is about 16 MB and it has other stuff in it

Turns out the part you need is reasonable size included below
Amazing, thankyou I'll take a look through all of this, can't thank you enough for your help
 
Top