Thermal control system design

[Robert Wolcott]

I am designing a closed loop thermal control unit that would like to be very
accurate (+/- .2 degrees C). I plan to use a PID controller with an RTD
sensor and a small tape style heater.

Most PID controllers only provide a signal out and do not directly power the
heating element. If I had a PID controller providing a low voltage signal
(mV), what sort of intermediate circuit would I need to power a 10-50 watt
heating element? My main goal is not necessarily accuracy, but stability.

Thanks,
Bob

 

[John Popelish]

I am designing a closed loop thermal control unit that would like to be very
accurate (+/- .2 degrees C). I plan to use a PID controller with an RTD
sensor and a small tape style heater.

Most PID controllers only provide a signal out and do not directly power the
heating element. If I had a PID controller providing a low voltage signal
(mV), what sort of intermediate circuit would I need to power a 10-50 watt
heating element? My main goal is not necessarily accuracy, but stability.

Many PID controllers provide a pulsing relay power output, that if the
pulse time can be made significantly shorter than the dominant thermal
time constant of the process, approximates a continuously variable
power output. There also solid start relay outputs that control
integer half cycles of the line waveform of on time for a low noise
version of this sort of burst control.

Other controllers provide an analog output, usually 4 to 20 milliamps
that represents the 0 to 100% power output. The simplest way to
connect that to a small heater is with a solid state relay that
converts the variable current into a phase controlled output (the sine
wave has various early parts of each half cycle of the line waveform
turned off, with the rest of each half cycle turned on. This is still
a form of pulse width modulation, but is much higher frequency than
the relay output.
http://rocky.digikey.com/WebLib/Crydom/Web Data/PCV Series.pdf

The ultimate in low noise, continuously variable heat control would be
a linear voltage regulator controlled by the 4-20 ma output, but that
would be used only for very low power heaters with extreme precision
requirements.

I have never seen a PID controller with a millivolt output.

A big problem for your controller will be the tiny voltage represented
by a less than .2 degree error (and the controller has to be able to
measure the temperature to a finer precision than you can expect it to
control). If the temperature is moderate, you may do better with a
thermistor even though the absolute calibration will not be as good,
because of the higher sensitivity to very small temperature changes.

 

[Robert Wolcott]

John,

The controller I had in mind was an Omron unit and I believe it had an
current output instead of the voltage output I had stated. The temperature
I am looking for is 30-40 degrees C. Do you think a thermistor would be
better for that range?

Thanks,
Bob

 

[John Popelish]

John,

The controller I had in mind was an Omron unit and I believe it had an
current output instead of the voltage output I had stated. The temperature
I am looking for is 30-40 degrees C. Do you think a thermistor would be
better for that range?

That is an ideal temperature range for many thermistor sensors. You
will probably need a transmitter (signal conditioner) to convert the
variable resistance of the thermistor to the standard 4-20 ma
controller input. If you pick a transmitter that allows a narrow
range of resistance to be expanded to the full input signal range, you
will increase the resolution, though the stability and calibration
issues remain.
some examples that you can mine for key words for searches:
http://www.ametherm.com/ntc_thermistor_probes/welcome.htm
http://www.betatherm.com/
http://www.rtie.com/ntc/probe.htm

 

[[email protected]]

Maybe I'm just not familiar with it, but why would you want to use a
PID controller for a thermal process? A thermal system is much more
complicated than a simple spring/mass/damper system.

I would assume a PD controller with an experimentally obtained offset
would be adequate.

Dave

 

[John Popelish]

Robert, have a look at the stuff from wavelength electronics, it may give
you some ideas - there is not a lot in some of the controllers. I used one
of their small PID units for controlling a Peltier, had to add extra
heatsinking. It provided direct drive of the load and had analog in & out
signals as well.

www.wavelengthelectronics.com

I have used these units to control both laser current and
temperature. And the peltier control units accept thermistor sensors,
directly and control clean DC to run the peltier. I suspect one of
these units could be used as is, if with slight modification, for the
complete heater control.

 

[Tim Wescott]

Maybe I'm just not familiar with it, but why would you want to use a
PID controller for a thermal process? A thermal system is much more
complicated than a simple spring/mass/damper system.

I would assume a PD controller with an experimentally obtained offset
would be adequate.

Dave

(A) because you can buy off-the-shelf PID controllers.
(B) because everyone knows how to tune a PID (well, sorta).
(C) because PID is often good enough.
(D) because if PD is good, PID is better and no harder (see A).

 

[Tim Wescott]

John,

The controller I had in mind was an Omron unit and I believe it had an
current output instead of the voltage output I had stated. The temperature
I am looking for is 30-40 degrees C. Do you think a thermistor would be
better for that range?

Thanks,
Bob

Surely Omron has one with relay output? That's a very very common
output for temperature control.

 

[Rob]

I am designing a closed loop thermal control unit that would like to be very
accurate (+/- .2 degrees C). I plan to use a PID controller with an RTD
sensor and a small tape style heater.

Most PID controllers only provide a signal out and do not directly power the
heating element. If I had a PID controller providing a low voltage signal
(mV), what sort of intermediate circuit would I need to power a 10-50 watt

Robert, have a look at the stuff from wavelength electronics, it may give
you some ideas - there is not a lot in some of the controllers. I used one
of their small PID units for controlling a Peltier, had to add extra
heatsinking. It provided direct drive of the load and had analog in & out
signals as well.

www.wavelengthelectronics.com

rob