Implementing PID on Temperature Controller

lloydi12345 · Joined: 23 Aug 2011 Posts: 5

Hi CCS people. It is my first time to post here. I have a project which is a temperature controller and I would like to apply the principle of PID. It's a fryer and it is tasked to follow temperature settings for its appropriate time settings. I have read a lot of PID threads here and able to get along what PID really is. One of the threads I've read is this one: http://www.ccsinfo.com/forum/viewtopic.php?p=58275 I would like to apply PI only without D(I don't know if it's okay). I'm having problems about this code:

SherpaDoug · Joined: 07 Sep 2003 Posts: 1640 Location: Cape Cod Mass USA

1) Errors can be positive or negative. There is nothing odd about a negative error. Make very sure 'sum' can not roll over from very big positive to very big negative or the opposite. That could lead to a very dangerous error!

2) The time scale depends on your fryer and how close you want to regulate the temperature. Consider the fryer full on for 20 seconds starting slightly too cold. Will it be too hot by the time 20 seconds is over? If so 20 seconds is too long.
_________________
The search for better is endless. Instead simply find very good and get the job done.

lloydi12345 · Joined: 23 Aug 2011 Posts: 5

Thanks SherpaDoug for the reply. I see. I get confuse with the output. Is the output not incremented to its previous value? like:

SherpaDoug · Joined: 07 Sep 2003 Posts: 1640 Location: Cape Cod Mass USA

If Ki and Kp are scaled right then output will be the duty cycle or the power required. Higher output = more power needed. If the error is negative then the output will start to drop, as it should if things get too hot.
_________________
The search for better is endless. Instead simply find very good and get the job done.

lloydi12345 · Joined: 23 Aug 2011 Posts: 5

I see. How about the sum? What should be the value of the max positive and the max negative? What will I do if it reaches those values? Should I reset them to 0?

SherpaDoug · Joined: 07 Sep 2003 Posts: 1640 Location: Cape Cod Mass USA

Over time the sum should average to zero as positive and negative errors are added. In PID loops it is common to limit integrator "windup" for step changes in input by limiting the range of the integrator term.

http://en.wikipedia.org/wiki/Integral_windup
_________________
The search for better is endless. Instead simply find very good and get the job done.

lloydi12345 · Joined: 23 Aug 2011 Posts: 5

gpsmikey · Joined: 16 Nov 2010 Posts: 588 Location: Kirkland, WA

Last year I built a PI controller for the heat plate on the bottom of the hummingbird feeder (quite a few stay around the Seattle area even when it is well below freezing). The heater kicks on when the ambient air temp gets below 34 deg F and off when the ambient air temp goes above 40 deg. F. The proportional controller has a couple of extra features (I use a 2.5 second cycle time on the heater and change the "on" time for the heat to maintain the heat plate at 70 degrees. One of the things I did since I was using remote sensing of the temperatures is use a measurement "window" where I turn the current to the heater off at the end of the cycle, measure the temps then turn the heat back on for the amount needed (even with separate grounds for the sensors, I was getting some error due to current). Below is a portion of the PI code I created that seems to work fine (the settings.variable are things I had stored in eeprom for getting the response the way I wanted it) - note this is NOT the complete program but does show the PI portion of the code:

(here is a link I found quite helpful in understanding it all:
http://igor.chudov.com/manuals/Servo-Tuning/PID-without-a-PhD.pdf )
also, the June 2006 Circuit Cellar had an article called "Home-brewed HERMS" that was about using PID for controlling home brewing stuff you would probably find helpful.

Here is a picture of it in "action" ...

lloydi12345 · Joined: 23 Aug 2011 Posts: 5

These codes are really helpful gpsmikey thanks. I already found the article you've recommended me last time but didn't really understood them all. So I'll give this one again another try and see if this time I'll be able to understand them. Thanks again.

asmboy · Joined: 20 Nov 2007 Posts: 2128 Location: albany ny

I've done PID controls for motion apps several times in recent years, and
I did a project for a fast food commercial 2 gallon tank coffee brewer , demand heater last year.

Actually with a safety cutout on the high side of the demand heater control TRIAC drive, and a pair of LM35's doing failsafe sensing for each other , a well tuned PID algorithm tested out no better at all than a rudimentary "click bang", very ordinary set point method.
The simple set point method with a built in 2 degrees of hysteresis
was able to reach the desired temperature faster actually.

Is s small amount of overshoot really an issue with this app?

I think the key to good temp regulation is well coupled temp sensors actually.

In the case of something like a fat fryer - so long as there is a PHYSICAL over temp sensor ( say NC bimetal ) I feel you will be hard pressed to show any performance improvement no matter how hard you push a PID approach.

If simply learning to do a PID control?
Thats another kettle of fried fish;-))

The ONLY situation where think you see PID being an advantage is to paper over and compensate for unusually poor coupling of your sensors to the bucket of fat ;-))