Heating Element PID Control

[dvdmoylan]

Hi All,

I have a heating element (230VAC 250mA) which is used to heat pressurized air (18 SLM) as it passes through the element via an SSR. No matter what PID settings i use i cannot prevent oscillation, ie overshot. I graphed the output of the PID controller and the temperature, when the temp goes slightly below the setpoint the PID kicks in for about 300msec, this it enough time to cause overshoot. Other than adding a step down transformer is there any other method i should try to dampen the current drawn by the element when the PID kicks in as it seams that the element is too responsive.

Thanks,

David

[nathand]

What method did you use to tune the PID? Sounds like you have too much integral control. Try using a larger value for the integral time - note that in the LabVIEW PID, the integral value is time not gain, so a larger value produces less integral response (longer time).

[dvdmoylan]

Thanks for the response,

I had used various integral times of between 0.001-1.5 the lower i went the more unstable to system became, i also tried setting it to 100 but the response was similar to 1.5. Will try more setting between 1.5& 100 to see if it works, also should i go greater than 100?

Thanks,

David

[johnsold]

How long does it take for your sensor to respond to a change in the input?  If you turn on the heater for 1 second, when does the temperature begin to change and how big is the change? What is your target error tolerance on temperature?

Lynn

[nathand]

Did you try a real tuning method, for example Cohen-Coon or the open-loop Ziegler-Nichols method? If not, you should start there. It's not difficult. Turn on the system at some fixed output (that you know is safe) and wait until the temperature is completely stable. This could take a few hours. Log data throughout that time. You then use that data to calculate initial gains using simple formulas. If you're just using trial and error, it's very difficult to get it right, especially if your system responds slowly.