Temperature control of a specimen inside a temperature chamber

[yavs]

Hello,

I have a thermal chamber that I'm controlling with serial Modbus. The only thing I can control is air temperature inside the chamber and gradient. On the other side I have some thermocouples measuring the temperature of the specimen inside the chamber. I would like to control the temperature of the specimen with labview.

I know I need to do a cascada control. The first PID would be the difference between the desired specimen setpoint - specimen temperature. The thing is I don't know how to "convert"/"relate" temperature of the specimen with air temperature inside the chamber.

Convection formula is : E=h*A(Tair-Tspecimen), this formula could relate Tair and Tspecimen--> Tair = (E/h*A)+Tspecimen... but I don't know "E".

Someone has worked with something similar in labview and knows how to solve this?

[Norbert_B]

I have not designed control circuits yet, but the following feed for thought:

1. PID is a control circuit design amongst many. Depending on the control behavior, PID might be a bad selection. Examples:
   • How about overshots
   • How about sticking at low temperatures too long if the control is very slow reacting
2. Consider how stable your parameters are and if there are additional dependencies. Example:
   • Your specimen start to generate temperature by themselves at one temperature point. Read: you can easily run into a vicious cycle if the control doesn't account for that
   • Air usually is quite inert regarding temperature. You could result into blowing real hot air into the chamber to speed up heating. However, this can lead to local hot spots having too high temperatures for specimens
3. Cycle time of control circuit. As stated above, air is usually inert to control regarding temperature. A cycle time in range of a second could easily be enough. However, the specific number has to be figured out. A 'real' control circuit requires determinism (read: cycle time must be kept at all cost). Running the software on non-real time systems can result in cycle time being inconsistent, maybe for longer periods. Is this something your system (read: specimen in chamber) can handle?