Analysis software aids selection of thermo-electric cooler
EILEEN MacKENZIE and TIM FLEURY
Thermo-electric coolers (TECs) are becoming more widely used, particularly in the optoelectronics industry. Devices such as laser diodes and charge-coupled-device (CCD) detectors must be controlled at a specific temperature over a wide ambient temperature range. Failure to do so can result in the instability of the output wavelength of the laser diode, or increased noise levels and decreased sensitivity in the CCD. The control temperature, however, is often much lower than that of the ambient temperature, which results in the need for a temperature-control system involving a TEC. (see "TE coolers use the Peltier effect," below).
Selection criteria
The purpose of a thermal analysis is to expose potential thermal problems as early in the design phase as possible and to properly size the thermal management components to save cost, size, and weight. So a steady state thermal analysis can provide the information required for selection of the thermal-control components, and a TEC can be selected based on the range of ambient temperatures.
The optimum performance of the TEC should occur at the ambient temperature with the highest frequency of occurrence. Once the TEC operating characteristics have been determined, the ultimate cooling system (fans, heat sinks, and so on) required to remove the heat generated by the TEC and device can be selected.
Sudden changes in environmental conditions that occur during events such as transportation can cause changes in the operating temperatures. Transient conditions also occur during startup. So the required time to stabilize, warm up to the minimum allowed operating temperature, or cool down to the maximum operating temperature can be critical to the design. Determining the response of the thermal-management control system to transient boundary conditions is often also required.
The available size, available power, and required performance must be determined to select the proper TEC. Maximum current, voltage, heat load, and temperature differential from hot side to cold side are specified for different TECs. The maximum current is the most efficient current. It is the point at which the I2R losses begin to dominate the Peltier cooling effects, the heat pumping capacity is maximized, and further increase in current will reduce the capacity. The maximum voltage is simply the potential that produces the maximum current. The maximum heat load is the point at which there is no net cooling and the temperature differential is zero. The maximum temperature differential for a single-stage TEC can be as high as 67°C.
Accurate modeling
With the above in mind, software can incorporate a TEC and control its operation with a thermostat. Previously TECs were either not modeled in a thermal analysis or modeling required an iterative process with the TEC properties being updated after each solve. The typical solution involved estimating the hot- and cold-side temperatures and the heat load on the TEC. The estimation approach produces an inaccurate result because it does not account for the nonlinearity in TEC performance. TEC power requirements and temperature differentials change with operating conditions.
For an accurate solution, inputs to the thermal model must include the TEC geometry, top and bottom material thickness and properties, type of TE material, and properties of the temperature-dependent material. The temperature-dependent inputs should include the Seebeck Coefficient, resistivity, and thermal conductivity of the couple material. For a transient analysis, the capacitance and temperature-dependent properties of the TEC must be taken into account in the model.
A full library of TEC parts should be available so that the desired TEC can be easily selected. In the software, TECs can be represented like any other part of the model by extruding brick or plate faces (see Fig. 1). The software then calculates the number of couples based on the area ratio of the TEC modeled to the actual TEC size, as this does not have to be exactly the same as the real TEC size. The voltage or current can be regulated, and the models can be solved for a transient as well as steady-state solution. Heat loads and ambient temperatures can be varied with time. The rest of the thermal model is created in the usual way with convection and radiation added where necessary.
In electro-optic devices the heat source is often very small, resulting in very small elements in the thermal model. Extruding these through to the TEC would increase the solve time. The particular software used contains a feature called SynchroMesh, which allows different meshes to be attached and is particularly helpful for annular TECs.
Once the model is complete and all the boundary conditions have been added, the model is almost ready to solve. The TEC voltage or current, however, must be defined and can be related to a themostat in the model. In the thermostat control box, "+1" signifies TEC cooling, and "-1" signifies heating—these values are then multiplied by the specified voltage or current to find the required operating power. The closer the estimated voltage or current value is to the final value, the quicker the solving time.
Once the model has solved, post processing will provide all the relevant TEC data, which includes the TEC voltage, current and power, and the cold-side heat load, Qc. Having Qc as an output is very helpful and no longer means that TEC selection will be based on estimates. The data found from the thermal analysis can then be used in software for a quick look at the benefits (or not) of using other TECs or to vary the TEC to ambient thermal resistance to quickly find the TEC power needed.
Different scenarios can be quickly and easily investigated. Tradeoff studies can be performed by selecting a different TEC or changing the input voltage or current to the TEC, changing the boundary temperature or material property (such as the TEC attachment), and resolving the model. The model can be exercised for transient conditions by using the steady-state results as initial conditions, or initializing to a cold or hot soak temperature.
Post processing
A model was created of an electro-optic laser module using software with an embedded TEC library (see Fig. 2). The goal was to find a suitable TEC to maintain the internal parts at 25°C in a 70°C ambient. The laser-power dissipation was known, and so for the first analysis a TEC was modeled that was thought to be suitable. All leads, wirebonds, and convection and radiation were added to the model to get an accurate figure for the heat load on the cold side of the TEC.
Once the model had solved, TEC data was provided that showed that cold-side total power was only 1 W. The power dissipated by the laser was 0.9 W so the additional passive heat loads only accounted for a further 0.1 W. With this information it was found that a smaller TEC than initially selected could be used. Without the software a larger, more expensive TEC would most likely have been selected resulting in an overdesign of the module.
Today's thermal-analysis packages, which include the TEC operating parameters, now allow for accurate predictions of the TEC performance by taking into account the system-level thermal management.
EILEEN MacKENZIE is a thermal analysis consultant at EMK-Analysis, 19 School St., Port Charlotte, Isle of Islay, Scotland PA48 7TW; e-mail: [email protected]. TIM FLEURY directs engineering services at Harvard Thermal, 249 Ayer Rd., Harvard, MA 01451; e-mail: [email protected]
TE coolers use the Peltier effect
Thermo-electric coolers operate on the principle of the Peltier effect, which works in the reverse of a thermocouple or Seebeck effect. A thermocouple consists of two dissimilar metals joined at one end (junction). A potential is generated at the opposite end based on the junction temperature. When the junction is heated or cooled, the potential generated at the opposite end changes. The Peltier effect is produced by applying the potential at one end to drive the temperature at the junction. The junction will increase or decrease in temperature, depending on the polarity and magnitude of the applied potential. The TEC is therefore capable of heating or cooling the device, but at the cost of added power requirement to the system.
A thermostat is used to control the TEC to maintain the required operating temperature range of the device. The thermostat system consists of a temperature sensor located on or near the device, and a voltage or current controller for the TEC. The temperature sensor must have a small thermal capacitance compared to the device to reduce the time lag between the sensed temperature and device temperature.
Quite often a negative temperature coefficient (NTC) thermistor is used because it has a rapid response time in temperature sensing, a high resolution over a narrow temperature range, is inexpensive, and its small packaging makes its application versatile. Thermistors are temperature-dependent resistors. Their resistance decreases, nonlinearly, with increasing temperature and vice versa. A control system is used to monitor the resistance changes and adjust the input (voltage or current) to the TEC to compensate.