Op Amp closed loop control

Thread Starter

300-3056

Joined Sep 9, 2022
26
Hi there.
In another forum I have been talking about closed loop control and found this circuit that seems to do what the group is trying to build.
That is an automatic gasious fuel air mixer with feedback.

Some differences, they don't need the push pull circuit at the end because they have servo motors

The set point needs to be done from O2 sensor ( I wonder how many newbies like me have come here asking about something like this using analog parts or an arduino ? )
A narrow band sensor would be nice a wide band may be required ( 0-1 volt narrow 0-5 volt for the more precise wide )

1665068089253.png

Where is s good place to look for something like this to adapt?
If I had to guess it was designed to use something like a dual supply OP-amp and these guys are interested in single supply.

It's been a lot of years since I played with any of this stuff.
I can look at the block diagrams and tell you what they are doing, but I can't recall much else about how to design something like this or where I would even look for help.
 
Last edited:

dcbingaman

Joined Jun 30, 2021
1,065
Hi there.
In another forum I have been talking about closed loop control and found this circuit that seems to do what the group is trying to build.
That is an automatic gasious fuel air mixer with feedback.

Some differences, they don't need the push pull circuit at the end because they have servo motors

The set point needs to be done from O2 sensor ( I wonder how many newbies like me have come here asking about something like this using analog parts or an arduino ? )
A narrow band sensor would be nice a wide band may be required ( 0-1 volt narrow 0-5 volt for the more precise wide )

View attachment 277806

Where is s good place to look for something like this to adapt?
If I had to guess it was designed to use something like a dual supply OP-amp and these guys are interested in single supply.

It's been a lot of years since I played with any of this stuff.
I can look at the block diagrams and tell you what they are doing, but I can't recall much else about how to design something like this or where I would even look for help.
What you are dealing with here is nothing more than a analog implemented PID control loop. U2 is proportional gain, U3 integral gain and U4 derivative gain and U1 creates the error term that is: setpoint minus feedback reading. You would have a lot more flexibility if you use a microcontroller to implement the PID. Because every application is unique there is no 'one size fits all' solution for PID, except that you can more easily tune a PID control implemented with microcontroller along with setting control limits and integral windup limits etc. I don't think most modern PID controls are using Analog implementations anymore due to the extremely hard time it takes to tune them properly for the Plant being controlled. Recommend a simple microcontroller PID implementation. The primary task then just boils down to digitizing the O2 sensor output, that is some analog conditioning followed by and ADC, the ADC typically already lives in the microcontroller along with say a PWM output of the microcontroller for the Servos. The PWM output would have some analog filtering and conditioning based on your needs.
 
Last edited:

crutschow

Joined Mar 14, 2008
34,281
You might also consider using Fuzzy Logic with a microprocessor.
PID assumes a linear system, and can have problems controlling a system with non-linearities or hysteresis, whereas Fuzzy Logic does not need the system to be linear.
I like Fuzzy Logic because it basically uses a series of If-Then-Else statements to perform a function, which are generally easier to understand than the PID equations, where the digital processor is trying to digitally emulate the analog integral and differential functions.
If interested, here's a short article I wrote about using it, along with a reference to a newbie who incorporated it in a dynamic pressure/flow control system, where it worked better than the PID system he had previously tried.
 
Last edited:

dcbingaman

Joined Jun 30, 2021
1,065
You might also consider using Fuzzy Logic with a microprocessor.
PID assumes a linear system, and can have problems controlling a system with non-linearities or hysteresis, whereas Fuzzy Logic does not need the system to be linear.
I like Fuzzy Logic because it basically uses a series of If-Then-Else statements to perform a function, which are generally easier to understand than the PID equations, where the digital processor is trying to emulate the analog integral and differential functions.
If interested, here's a little article I wrote about using it, along with a reference to a newbie who incorporated it in a dynamic pressure/flow control system, where it worked better than the PID system he had previously tried.
I may use that technique on the second version of a project I am working on. Currently we are using a microcontroller PID implementation. That appears to be working but I did have to deal with integral windup error, though I did find a solution to deal with integral windup error in our application, the fuzzy logic idea may give better results. Nice article you put together. I plan to read over the other two articles you referenced.
 

Thread Starter

300-3056

Joined Sep 9, 2022
26
Fuzzy logic.
I have some reading to do.
I used to have fuzzy understanding of how to code a Z80!
Sounds like this might be entertaining to try.

Here is a link to Drive on Wood.
A forum about building biomass gasification systems.
And this thread is Jesse's electronic automixer.

It was of on a false start but I am optimistic with some help from some modern electronics people we can come up with a simple one size fits all automixer ( just joking ) .

https://forum.driveonwood.com/t/jesses-electronic-automixer/6937/2
 

Thread Starter

300-3056

Joined Sep 9, 2022
26
What you are dealing with here is nothing more than a analog implemented PID control loop. U2 is proportional gain, U3 integral gain and U4 derivative gain and U1 creates the error term that is: setpoint minus feedback reading. You would have a lot more flexibility if you use a microcontroller to implement the PID.
We already have a fellow on team woodgas that built an ardiuno based automixer.
And it works pretty well....

But I recall my college days and we did this with op-amps before we got to the Digital stuff.
And on the surface this is not complicated problem in need of a lot of processing power so my gut tell me a hand full of op amps might be as complicated as needed.

Also if you are familiar with KB electronics drives they still make caveman drives that are all analog.
I work on ABB and GE drives that were all analog and worked in closed loop applications .
They did a fine job until they went out of fashion...

So how about a push for a guy many years out of school how to do I about building this and if possible I would like to pull as many of the tuning parameters out to trim pots as possible so this can be tuned by hand for guys without the ability to program in C.


Where do I start to research this?
Any suggestions?

My first thought is use LM324, that I do recall will work single sided supply

Where to start.
I found this article that goes with the previous diagram.
it seems well written enough so I will try and follow it.

https://www.nutsvolts.com/magazine/article/the_pid_controller_part_1
 
Last edited:

dcbingaman

Joined Jun 30, 2021
1,065
Do you have the original page in order to down load the tutorial? it just shows a PDF page
Building a Fuzzy Logic Controller
Can you say a little about what the project does?
I would be interested in hearing later how using Fuzzy Logic for it works out.
I can say a little, but it is proprietary work for a customer for controlling a car engine while idling. That is keeping engine RPM 'constant and stable' during engine idle regardless of engine loads that come and go (like Air Conditioning). The project is currently in beta testing. If I end up having issues with the implemented PID, I may consider Fuzzy Logic as it sounds tempting. If I do go that route, I will try to let you know how that works out. At the moment that is a few months down the road.
 

dcbingaman

Joined Jun 30, 2021
1,065
We already have a fellow on team woodgas that built an ardiuno based automixer.
And it works pretty well....

But I recall my college days and we did this with op-amps before we got to the Digital stuff.
And on the surface this is not complicated problem in need of a lot of processing power so my gut tell me a hand full of op amps might be as complicated as needed.

Also if you are familiar with KB electronics drives they still make caveman drives that are all analog.
I work on ABB and GE drives that were all analog and worked in closed loop applications .
They did a fine job until they went out of fashion...

So how about a push for a guy many years out of school how to do I about building this and if possible I would like to pull as many of the tuning parameters out to trim pots as possible so this can be tuned by hand for guys without the ability to program in C.


Where do I start to research this?
Any suggestions?

My first thought is use LM324, that I do recall will work single sided supply

Where to start.
I found this article that goes with the previous diagram.
it seems well written enough so I will try and follow it.

https://www.nutsvolts.com/magazine/article/the_pid_controller_part_1
If you want to go analog, I would recommend using modern op amps that have rail to rail inputs and rail to rail outputs. You can also use a single supply if you want. The 741 and LM324 are inferior to modern op amps. Just do a search on Digikey, Mouser, Newark for rail to rail input and output op-amps. We now have really low quiescent power, insanely low input leakage currents. The rail to rail ones will make your design task a lot easier and will also allow for a single supply voltage. Being Band Width is not an issue for your design, just about any modern op amp will work. Also recommend using SMD devices and electronic parts. You can create your own board using free online tools like KiCad or similar. You may find though that the task would actually be less expensive and more flexible with a low end 8 bit microcontroller. The cost of these is now almost down to the pennies. Did you want help with this project beyond advice on this forum? It sounds like an interesting project regardless how you implement in being analog or digital.
 

Thread Starter

300-3056

Joined Sep 9, 2022
26
Things have changed a lot.
I learned 8 bit processors using the Z80 and 6800 .
I blinked and there was the PI and Arduino.

Op amps have changed lol.
You know everything changes and before you know it know one knows what your talking about.
I was taught basic TV on B&W heathkit trainers with tubes lol.....
I actually know what a Mag amp is lol...

35 years later non of these things even work with modern electronics.
 

Papabravo

Joined Feb 24, 2006
21,159
Things have changed a lot.
I learned 8 bit processors using the Z80 and 6800 .
I blinked and there was the PI and Arduino.

Op amps have changed lol.
You know everything changes and before you know it know one knows what your talking about.
I was taught basic TV on B&W heathkit trainers with tubes lol.....
I actually know what a Mag amp is lol...

35 years later non of these things even work with modern electronics.
Ve get too soon oldt, und too late schmart
 
Top