[time-nuts] Achievable temperature stability for Thunderboltenvironment
holrum at hotmail.com
Mon Jan 17 04:00:35 UTC 2011
There is some info on implementing the temperature controller in the comments at the start of the file heather.cpp
You can use something the simple single ended transistor driver that Warren described or use a DC solid state relay. The program can manipulate two of the RS-232 port control signals (one for enable and the other for heat/cool). The simplest implementation uses just the heat/cool signal for controlling the fan. Heating uses self-heating of the Tbolt. Cooling uses the fan to add room air. I use a DC solid state relay with one end driven by the heat/cool signal and the other by the enable signal. I have a current limiting resistor and a zener diode across the relay to prevent overdriving the device.
The TT command sets the control temperature and enables the PID with default parameters. The KA command does an auto-tune procedure to fine tune the PID parameters. You can also individually change the PID parameters using various K commands. Once you have a set if parameters that works well, you can specify them on the command line (or in a configuration or keyboard script file)
When using a fan, you should set the control temperature somewhere between the maximum room temp seen and the no-fan-on self-heating temperature of the unit in the box.
And there are rumors of a list member volunteering to write a Heather manual...
More information about the time-nuts