Department of Computer Science and Electrical Engineering
Control Of A Thermoelectric Cold Plate Utilizing Experimental Design Methods
Darren Coleman
Abstract:The
controller for a Cold Plate utilizing a peltier thermoelectric device as
its actuator was to be designed using analytical design methods.It
was estimated that the design would require the development of an accurate
mathematical model from which a design using classical compensator design
techniques could be developed.
Development
of the mathematical model was to be performed using theoretical analysis
and then confirmed by comparison with the physical system.However,
it was found that development of a theoretical model would be difficult
due to the many system specific influences such as thermal interfaces,
the characteristics of the heat sink used and the influence of the difference
in temperature between the cold side and the hot side of the thermoelectric
device.Therefore, the theoretical
method for developing a model for the Cold Plate was abandoned, with experimental
methods used in its place.
Experimental
methods for finding a model of the Cold Plate involved analysis of the
system step response characteristics.It
was found that the Cold Plate exhibited a response close to that of a lagging
first order system.However, closer
examination of the step response curve revealed a higher order component.Further
analysis of this higher order component revealed that, although appearing
as a second order component by nature, its actual characteristics were
dependent on the surrounding conditions.It
was estimated that further analysis of the Cold Plate to determine a more
accurate model would require much more time and effort, and so was again
abandoned.
With
the uncertainties created by the inaccuracy of the developed model for
the Cold Plate, the design of a compensator using analytical design methods
that produced the specified performance appeared a formidable task.For
this reason, the analytical approach was also abandoned for an experimental
design method called the Process Reaction method.This
experimental method of control compensator design did not require an intimate
understanding of compensator design or a model of the system to be developed
but utilized only the step response of the system.It
was found that this method produced a design with adequate performance,
but without the development costs that are associated with the analytical
design methods.
Thus, it was concluded that for small projects in industry that require variable control, the experimental methods for the design of PID type control compensators are ideal for producing accurate controllers at a fraction of the time required in design using one of the corresponding analytical methods.
Complete thesis:
Thesis.pdf
Conference Paper:
Conference
Paper.pdf
Additional material:
Operational
Code for the Cold Plate Controller