Design of a universal thermoelectric module test

Design of a universal thermoelectric module test system for testing rat brain thermoelectric hypothermia H. Demirel, B. Ciylan, B. Erkal and S. Yılmaz Abstract: A universal microcontroller test system, which aims to determine the parameters of thermoelectric modules under various thermal loads, has been designed and realised using a novel test method. The test system has been designed according to a more simplified form of the present formula set, which has been made to accept minimum variables as the input in order to obtain more precise results. It measures all the parameters of a working thermoelectric module by measuring only the hot-side temperature, module operation voltage, current and thermal voltage values of the module. The new test system has been used to measure a standard thermoelectric module (Melcor CP 1.4-127-10L) in order to verify its performance. It has also been used to test the operation of an experimental medical apparatus, which is used to induce hypothermia (low body temperature) in the brains of rats using a thermoelectric module.

1

Introduction

Since the 1960s, thermoelectric modules have been in widespread use in the military, aerospace, science and medical technology fields as coolers, heaters and thermoelectric generators, because of their compactness, static operation, reliability, suitability for effective temperature control and DC operation [1 –3]. The studies made on thermoelectric coolers have concentrated on the development of new thermoelectric materials and production techniques. However the methods, used to analyse the performance of the modules, are as important as the new materials [4 – 6]. The studies on this subject have focused on obtaining those microparameters of a module (a, Seebeck coefficient of the semi-conductor material; R, resistance of the material; K, thermal conductance of the material; and Z, figure of merit) which cannot be extracted easily. These parameters are important when finding the thermal characteristics of the module. Since these parameters are highly dependent on the temperature and the physical dimensions of the elements, the classical methods occasionally fail to accurately determine the actual thermal parameters of modules. It may be impossible to determine them because the module is installed in an appliance. So, it is necessary to develop new methods to investigate the thermoelectric properties of semi-conductor materials and the thermal parameters of modules which are in operation in appliances such as a refrigerator or a rat brain thermoelectric hypothermia apparatus. # The Institution of Engineering and Technology 2007 doi:10.1049/iet-smt:20060083 Paper first received 28th February and in revised form 16th October 2006 H. Demirel and B. Erkal are with the Zonguldak Karaelmas University, Karabuk Vocational College, 100 Yıl Street, 78100 Karabuk, Turkey B. Ciylan is with the Gazi University, Technical Education Faculty, Electronic and Computer Education Department, Besevler, 06500 Ankara, Turkey S. Yılmaz is with the Zonguldak Karaelmas University, Technical Education Faculty, Machine Department, 100 Yıl Street, 78100 Karabuk, Turkey E-mail: [email protected]

160

There are various methods and devices for the measurement of the values of a, R, l and Z of semi-conductor materials [7 – 11]. Similarly, there are methods and devices to measure the thermal parameters of modules [12, 13]. In addition, special computer simulation programs, which use the same theoretical foundations as above, have been developed [14]. In this study, a new mathematical model which employs the easily and accurately measurable parameters of an operational thermoelectric module such as: thermal voltage, temperature, terminal voltage and current is developed and a microcontroller system is designed and realised to quietly extract and translate these parameters into the output parameters of a module. 2

Method

According to the classical method, the thermal balance equation of a thermoelectric module which shows the heat absorbed from the cold side is presented in (1) as [15] QC ¼ aITC 0:5I 2 R KðTH TC Þ QL

ð1Þ

where QC is the heat pumped from the cold side in watts, I is the current applied to the module in amperes, R is the resistance of the thermoelement in ohms, K is the thermal conductance of the thermoelement in W/8C, and a, is the total Seebeck coefficient of the thermoelement. Additionally, DT ¼ (TH 2 TC) and TH , TC are the thermal difference, hot-side and cold-side temperatures, respectively. Here QL represents the actual total external thermal load in watts, including the thermal losses [15]. The power consumed by the module is expressed as P ¼ I 2 R þ aðTH TC ÞI

ð2Þ

One of the thermal characteristics of a thermoelectric module is the coefficient of performance-(COP). The COP of a thermoelectric module is expressed as the ratio IET Sci. Meas. Technol., 2007, 1, (3), pp. 160 – 165

So, R can be rewritten according to these equations as

of the cold-side heat, QC to the power consumption, P COP ¼

QC P

The figure of merit, Z of a module is represented by (4) Z¼

a2 RK

ð4Þ

These equations establish the basis for the analysis of the thermal performance of a thermoelectric or a cooling system constructed with these modules [15]. Present methods use this equation set to calculate the thermal parameters. These formulae do not provide accurate results, as they are used without a proper measurement method, and also they do not represent the thermal parameters of an actual working module which experience various thermal loads. The reason behind this is that the variations in a R, l and Z of a semi-conductor pellet are dependent not only on the temperature, but also on the physical dimensions of the semi-conductor pellet [6, 15, 16]. Additionally, the technological and structural factors are not considered in these equations. The classical method can only produce accurate results, if every thermoelement is equipped with a pair of thermocouples, one to measure TH and the other to measure TC . If the high number of thermoelements in a typical thermoelectric module is considered, it can be seen that the practicality and the cost of the classical method mean that it is not feasible. Also, it is very hard to calculate the thermal load over the module, QL , using (1)– (4), because, it is impossible to measure the cold-side temperature, TC and the thermal conductance, K of an operational module in many cases, since the system is developed primarily to test or gather the performance of a module which is installed in a refrigerator or similar appliance. So, it is imperative to develop a new, low cost, practical method to investigate the real thermal properties of a working module from the point of view of both practical applications and theoretical studies. Despite the fact that the basic principles of the new method depend on (1) – (4), these equations are modified to reach the objectives of this study. The result of these equivalent modifications is that, only the values of the applied current and generated thermal voltage, E, are needed to calculate all the thermal parameters of a working module. The current and thermal voltages of a module can be measured very accurately and easily. As it is known, when QL is zero in an ideal thermoelectric module under no-load conditions, QC is zero under steady-state conditions. In this case the temperature difference T will be DTmax and the cold-side temperature, TC will be TCmin . Under these conditions, the current and voltage of the module will be Imax and Vmax , respectively, and the thermal generated voltage (Seebeck voltage) by the module will be Emax . The thermal properties of a module can be obtained by employing these macroparameters in (1) –(4). It is well known that Vmax ¼ aTH

ð5Þ

However, the maximum voltage can be expressed as follows Vmax ¼ Imax R þ aDTmax ¼ Imax R þ Emax

R¼

ð3Þ

ð6Þ

IET Sci. Meas. Technol., Vol. 1, No. 3, May 2007

ð7Þ

ð8Þ

If we put all these in the thermal equilibrim equation of the module, QC. can be expressed as 2 R KDTmax ¼ 0 QC ¼ aImax TCmin 0:5Imax

ð9Þ

From (7) we can obtain TCmin ¼ TH

Emax a

ð10Þ

If we put (8) and (10) into (9) we find an alternative expression for QC KEmax ð11Þ ðVmax Emax ÞImax 0:5ðVmax Emax ÞImax ¼ a If we now, withdraw the term K from this expression K¼ ¼

0:5aðVmax Emax ÞImax Emax 0:5Vmax ðVmax Emax ÞImax TH Emax

ð12Þ

If we assume that QL ¼ 0 and insert (8) and (12) in the first classical equation of QC , (1) QC ¼ aTC I

0:5I 2 ðVmax Emax Þ Imax

aDT ðVmax Emax ÞImax Emax

ð13Þ

We can generalise (7) and use it to obtain the cold-side temperature for any value of thermoemf of the module, E TC ¼ TH

E a

ð14Þ

Using (13) and (14), we can eliminate the term TC from (13) E 0:5I 2 ðVmax Emax Þ QC ¼ a T H I a Imax

EðVmax Emax ÞImax Emax

ð15Þ

In many cases the hot-side temperature of the module is held at a constant value and is seldom changed. So, the changes in the current and heat transfer rate of the module affects TH only vary slightly and thus it is suitable to employ the expression for Imax given in (5), in (15). In this case the final modified expression of QC is QC ¼ Vmax I

0:5I 2 ðVmax Emax Þ Imax

0:5ðVmax Emax ÞImax E Iþ Emax

ð16Þ

Similarly, the power consumption of the module can be expressed as

Emaxcan also be expressed as

aDTmax ¼ aðTH TCmin Þ ¼ Emax

Vmax Emax Imax

P¼

I 2 ðVmax Emax Þ þ EI Imax

ð17Þ 161

The COP is has been given in (3) and the figure of merit can be expressed as Vmax Emax Z¼ 0:5ðVmax Emax Þ2 TH

By measuring the TH and E parameters at any time, one can calculate the cold-side temperature at that time. In this study, a microcontroller-based multifunction test system has been realised using the new method, and it has been tested using a standard thermoelectric module.

The thermoelectric module test system consists of three main blocks: the power supply, cooling system, and the electronic control and data acquisition unit. The power supply is a switched mode type and a water circulation type heat exchanger is used as the cooling system. Temperature measurements are taken via a K-type thermocouple and the current is sampled by a Hall-effect sensor. The front end for temperature measurements is also suitable for T-type thermocouples. All of the test system is controlled via a microcontroller board. The system is capable of making all the necessary calculations. A schematic diagram of the mechanical structure of the test system is shown in Fig. 1. The hot side of the thermoelectric module is cooled by a water cooling system. There is a thin layer of copper sheet between the brass cooling plate and the hot side of the thermoelectric module, whose purpose is to provide a thermally conductive means in which the thermocouples can be freely attached to measure the hot-and cold-side temperatures. There is a brass heater plate between the cold sides of two thermoelectric modules. One of the thermoelectric modules is the device under test, while the other one, which is identical with the module under investigation, is used to distribute the heat load posed by the heater equally between its two sides. All the test apparatus is covered by a 5 cm polyurethane coating to minimise the external heat load and the internal heat losses. The electronic measurement and control system is the most important part of the test system and its block diagram is shown in Fig. 2. The data is collected from sensors and stored by the microcontroller board, which makes the necessary calculations and represents the results. A Hall-effect current sensor is used and thus its insertion

DC TC thermocouple DC TH thermocouple

thermoelectric module

polyurethane foam


thermal load copper sheet thermal grease copper sheet thermal grease brass plate

circulation pump

heat exchanger

Fig. 1 162

water tank fan

Cooling unit of microcontroller test system

240*128 LCM display

Yükselteç op-amp (OP-07)

ADC (AD 7731)

4*4 keypad

thermocouple amplifier (AD595)

amplifier op-amp (OP-07)

amplifier op-amp (OP-07)


voltage sensor

current sensor

serial interface (RS 232)

ON/OFF switch

power supply

Fig. 2 Schematic diagram of the electronic measurement and control system

Thermoelectric module test system

DC

2*24C64

microcontroller (ATMEGA128)

ð18Þ

Finally, the cold-side temperature, TC can be written as E ð19Þ TC ¼ TH 1 Vmax

3

memory

error is very low. The LEM LA55-P current sensor used here provides an output sensitivity of 100 mV/A. There is nothing special about the voltage sensor. The temperature measurement is realised through a thermocouple amplifier which uses a K-type chromel – alumel thermocouple to measure the hot-side temperature of the thermoelectric module. The Thermocouple amplifier utilises a circuit which has cold junction compensation built-in. At an ambient temperature of þ25 8C, it provides an output sensitivity of 10 mV/8C with a K-type thermocouple. In the 0 –50 8C ambient temperature range, the measurement error of the temperature measurement system is +0.6 8C. The total uncertainty in the measurement, including the thermocouple, is +1.2 8C. The relay is used to control the current. The power supply is separate and isolated from the circuit supply. The voltage and current limit of the power supply can be adjusted from 0 – 40 V and 0 – 10 A, respectively, according to the circuit conditions and module requirements. The ripple factor of the power supply is below 1%. The values of QC , TH , COP, power consumption (P), and the figure of merit of the module (Z) are shown on an LCM display when the tests have been concluded. The temperature, current and voltage sampling intervals and digital calibration data and all the commands are entered via a 4 4 matrix keypad. All the temperature, current and voltage measurements are logged by the system and can be transferred to an external computer terminal via a serial link for further analysis. A module is tested in two stages. The method is similar to the Harman method, but differs in the sense that macroparameters are used to extract the actual thermal parameters of a working module [17]. In the first stage, the module is operated with a suitable voltage level, which is set according to the module’s catalogue data. Under unloaded thermal conditions the Vmax , Imax , Emax parameters of the modules are obtained firstly. At this stage the system closes the relay contacts and thus applies current to the module. At the same time, the current, voltage and hot-side temperature of the module are sampled at predetermined intervals. The microcontroller compares every temperature sample with previous ones using a predetermined tolerance factor, and looks for a zero difference. When this condition is satisfied, it cuts the current of the module. The microcontroller waits for a short while. This is necessary for the measurement system to obtain a sample which is free from switching noise and retention by the contacts. This time delay is kept sufficiently small to ensure that it does not effect the measurements (t ¼ 0.5 s). After this time delay the IET Sci. Meas. Technol., Vol. 1, No. 3, May 2007

microcontroller takes a final voltage reading. This value is recorded as Emax . The voltage and current readings just before the current is cut off are recorded as Vmax and Imax , respectively. The first stage of the testing procedure necessitates connecting the module to the measurement apparatus as shown in Fig. 1. However, the values obtained at the end of this stage can also be entered manually by taking these data from the module’s datasheet. In this case it is not necessary to remove the module from its place. If the operational performance of the module which is attached to a system is the one which is in question, the second stage of the test procedure can easily be realised with the module operating in its place. In the second stage, the microcontroller applies current again. But in this stage the module is thermally loaded by a suitable thermal load which is determined from the datasheet of the module. This thermal load may also be the natural thermal load which the module sees from its position in the system. The user commands the controller to cut off the current via the keyboard at a suitable time. In this case the voltage, current and hot-side temperature of the module are sampled and then the thermal voltage, E of the module is also sampled after current has been cut off. However, just after the sampling of the thermal voltage of the module, the current is applied again. If the user wants to take another sample the procedure is repeated. Thus, after every sampling, the data obtained from the both first and second stages are processed by the equation set and the calculation results are displayed immediately. The measurement of the thermal parameters of a working module takes place in the second stage, whereas the steady-state limit parameters of the module are measured in the first stage. The first-stage measurements are taken once and are used in the second-stage calculations as the input parameters to determine the actual condition of the module. So, the system is also useful as a controller and as a health monitor for a module installed in an appliance. In this case, the system can alert the user when a critical shift from the specifications of the module occurs. Measured values of Vmax , Imax , Emax , TH , I, V and E are used in (16) to calculate the cooling energy QC , in (17) to find the power consumption, P, in (3) to find the module’s COP, in (18) to find the figure of merit, Z, and finally, in (19) to find the cold-side temperature, TC (found automatically by the microcontroller system) and the calculated values are displayed. 4 Determination of performance of thermoelectric test system A series of experiments have been carried out to determine the performance of this system. Two thermoelectric modules (Melcor CP 1.4-127-10L), whose specifications are well known from the manufacturer’s datasheet, have been used in the experiments [14]. The test apparatus to which the modules have been attached is shown in Fig. 1. One of the modules has been used as a cooler in the second stage. It cools the heater in order to distribute its heat equally between the cold sides of the two modules. The other one has been subjected to hosting. Its current, voltage and temperature have been monitored by the system during the experiment. Both modules have been supplied with the same voltage and current. Power to the heater has been applied from an adjustable DC power supply. The power is determined by the voltage and current measurements taken by two Philips IET Sci. Meas. Technol., Vol. 1, No. 3, May 2007

Fluke 45 digital multimeters. The voltages and currents of both modules have been simultaneously measured by separate Philips Fluke 45 multimeters. The hot-side temperature of the tested module has been measured simultaneously by the test system and a data acquisition system (HewlettPackard Agilent 34970A), the cold-side temperature is also measured by the data acquisition system for control purposes. All the measured values have been compared with those taken by the system itself. A total of 21 measurement points has been taken, including the loadless case. Every measurement has been repeated at least five times and their average has been taken as the actual value. Synchronisation between all the measurement systems is accomplished using time stamps along the sampled data. A chronometer, which is adjusted to the local time clock of the data acquisition system, is used to indicate the sample times for the Philips multimeters, where the sample interval is relatively long and the measured data is only changes occasionally. Multimeter readings are used to calibrate the test system. In the second place, the system is used to collect the actual thermal parameters of a working module which is installed and used to cool a rat’s brain in an experimental medical apparatus. The details of this special purpose medical apparatus have been given elsewhere [18, 19]. The hot side of the module is cooled by a water circulation system and the water temperature of the return pipe is measured as the hot-side temperature. Again, the module in the system is thermally isolated from the environment and the first stage of the testing procedure is executed using the test system. The Vmax , Emax , Imax parameters obtained at the end of this stage are listed in Table 1 along with the end results obtained after the second-stage measurements. The second-stage measurements are collected under the loaded regime of the module. The data in Table 1 is sampled under steady-state thermal conditions. 5

Results and discussion

The performance of the realised system has been tested using a pair of thermoelectric modules (Melcor CP 1.4-127-10L) and then the acquired data has been compared with manufacturer’s datasheets for these modules. All the output parameters have been calculated using the manufacturer’s datasheet values as the input parameters and they have been compared with those were reported by the system. These last calculated values have been found using the classical equations (1)– (4) [6 – 16]. It is found that the indirect measurement results of the test system and the calculated values are within close range, as the graphs revealed. Fig. 3 shows the graphical representation of change in measured QC against the thermal load posed by the external heater, QL , with the calculated QCM value added. QL is determined according to the electrical power measurement of the heater resistance. As can be seen from the figure, the average of the differences between these three parameters does not exceed 0.7 W. It can also be clearly seen from the figure that, despite the thermal load being zero, the calculated and measured QC values are non-zero. The reason is the thermal losses which arise from the non-ideal thermal isolation. Its value is 3.46 W for the realised system, thus it is understood that thermal losses in the measurement apparatus are 3.46 W. The isolation can approach the ideal by using a vacuum chamber [6]. However, the system can calculate the exact value of this thermal loss, which cannot be avoided in many circumstances. 163

Table 1: Measurements from the operational module installed in the rat brain thermoelectric hypothermia apparatus Vmax , V

Imax , A

Emax , V

TC , 8C

TH , 8C

E, V

QC , W

P, W

COP, %

Z, 1023 K21

14.00

2.6

3.20

27.05

34.70

1.90

9.43

34.29

0.28

2.50

COP (%), P (W), DT (°C)

30 25 20 15 10 5 0 0

5

11

16

21

27

32

2.6 2.6 2.6

10–3 Z

QC, QCM, W

35

2.6

100 90 80 70 60 50 40 30 20 10 0

40

2.6 2.5 2.5 3.9

10.6

16.3

22.1

28.7

34.4

QC, W

QL, W

Fig. 3 Thermal load over module obtained using two different methods

Fig. 5 Measurement of parameters of a module using microcontroller test system

—V— QC —B— QCM

—B— —V— —O— —X—

P DT COP Z

TC, TCS, TH, THS, °C)

50 40 30

is possible to compare the data with the datasheet data of the module under test. Thus it can be seen that the microcontroller test system has the capability to indirectly measure all the parameters of a module for different thermal loads.

20 10 0 –10 –20 –30

6

–40 0 2

4 5

7 9 11 13 14 16 18 20 21 23 25 27 28 30 32 33 35 QL, W

Fig. 4 Temperature measurement calibration results for various thermal loads —V— TC —B— TCS —O— TH —*— THS

The hot- and cold-side temperatures being measured by the system have also been recorded using the Hewlett-Packard Agilent 34970A data acquisition system and these values are shown in Fig. 4, where TC and TH are the cold- and hot-side temperatures measured by the data acquisition system, respectively, and TCS and THS are the cold- and hot-side temperatures measured by the test system itself, respectively. As can be seen from the figure, both systems have values of these parameters which are very close to each other. It can be seen also, that the hot-side temperature remains constant at 30 8C. The hot-side temperature does not change and stay constant in many systems especially if they are water cooled. So, the proposed measurement system has been tested under constant hot-side temperature conditions. But, this does not mean that it cannot measure correctly with different hot-side temperatures, since it is one of the input parameters of the system. Fig. 5 shows all the parametric measurement results of an operational module using the microcontroller test system. The data in this figure are the sole measurement data of the system taken with the sample module from Melcor. It 164

Conclusions

The microcontroller test system, realised in this study, depends on the measurement of the Seebeck voltage of the thermoelectric module. The calculated output parameters of a module according to these rules are its real values under various thermal loads. As a result, the problem of testing a thermoelectric module while it is solved in a practical and convenient way by using this new method. Investigation of parameters of a standard thermoelectric module manufactured by Melcor using this system yielded very similar results to those obtained using classical methods, as can be seen in Figs. 3 – 5. For instance, the difference between the measured thermal load value of the microcontroller system and the calculated value from datasheets provided by Melcor has not exceeded 0.7 W. The new system has also measured temperatures with a maximum of deviation from the calibration instrument 0.5 8C. 7

References

1 Andersen, J.R.: ‘Thermoelectric air conditioner for submarines’, Adv. Energy Conv., 1962, 2, pp. 241– 244 2 Mei, V.C., Chen, F.C., and Mathiprakasam, B.: ‘Comparison of thermoelectric and vapor cycle technologies for groundwater heat pump application’, ASME J. Solar Energy Eng., 1989, 111, pp. 353– 360 3 McNaughton, A.G.: ‘Commercially available generators’ in Rowe, D.M. (Ed.) CRC handbook of thermoelectrics vol. 1 (CRC Press, FL, USA, 1995), pp. 459–469 4 Min, G., and Rowe, D.M.: ‘Improved model for calculating the coefficient of performance of a Peltier module’, Energy Convers. Manage., 2000, 41, pp. 163 –171 5 Xuan, X.C., Ng, K.C., Yap, C., and Chua, H.T.: ‘A general model for studying effects of interface layers on thermoelectric IET Sci. Meas. Technol., Vol. 1, No. 3, May 2007

6 7 8 9 10 11

devices performance’, Int. J. Heat Mass Transf., 2002, 45, pp. 5159–5170 Huang, B.J., Chin, C.J., and Duang, C.L.: ‘A design method of thermoelectric cooler’, Int. J. Refrig., 2000, 23, pp. 208 –218 Crucq, A., and Degols, L.: ‘A new method for the measurement of the thermoelectric power of sintered semiconductors’, J. Phys. E: Sci. Instrum., 1972, 5, pp 81– 83 Waclawek, W., and Zabkowska, M.: ‘Apparatus for the measurement of thermoelectrical properties’, J. Phys. E: Sci. Instrum., 1981, 14, pp. 618–620 Goldsmid, H.J.: ‘A simple technique for determining the Seebeck coefficient of thermoelectric materials’, J. Phys. E: Sci. Instrum., 1986, 19, pp. 921 –922 Sumanasekera, G.U., Grigorian, L., and Eklund, P.C.: ‘Low-temperature thermoelectrical power measurements using analogue subtraction’, Measure. Sci. Technol., 2000, 11, pp. 273 –277 Min, G., and Rowe, D.M.: ‘A novel principle allowing rapid and accurate measurement of a dimensionless thermoelectric figure of merit’, Measure. Sci. Technol., 2001, 12, pp. 1261–1262

IET Sci. Meas. Technol., Vol. 1, No. 3, May 2007

12 Buist, R.J.: ‘Methodology for testing thermoelectric materials and devices.’ in Rowe, D.M. (Ed.) CRC handbook of thermoelectrics vol. 1 (CRC Press, FL, USA, 1995), pp. 189–209 13 Xuan, X.C.: ‘Investigation of thermal contact effect on thermoelectric coolers’, Energy convers. Manage., 2003, 44, pp. 399– 410 14 http://www.melcor.com, accessed June 2004 15 Ioffe, A.F.: ‘Poluprovodnikonie Termoementi’ in Gultaiev, P.V. (Ed.) Polucenie Holoda (pres. Ahademia Nauk SSSR, Moscow, 1960), pp. 102-107 16 Uemura, K.: ‘Commercial Peltier modules’ in Rowe, D.M. (Ed.) CRC handbook of thermoelectrics vol. 1 (CRC Press, FL, USA, 1995), pp. 621– 631 17 Harman, T.C.: ‘Special techniques for measurement of thermoelectric properties’, J. Appl. Phys., 1958, 29, pp. 1373–1379 18 Demirel, H., and Ahıska, R.: ‘Microcontroller based rat thermohypotherm system’. Proc. III. Int. Advanced Technologies Conf., Ankara, Greece, 2003, pp. 165 –173 19 Ahıska, R., Demirel, H., and Erkal, B.: ‘Post traumatic protection of brain in rats using rat termohypotherm device’, J. Sci., 2004, 17, (4), pp. 29–38

165