Cryogenics 42 (2002) 5966 www.elsevier.

com/locate/cryogenics

Construction and performance of a thermoacoustic refrigerator

Department of Applied Physics, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, Netherlands

Abstract This paper deals with the construction and performance of a thermoacoustic refrigerator. The manufacturing of the dierent components of the apparatus will be explained along with the reasons for using specic materials. The setup consists of three major parts: The refrigerator which is contained in a vacuum vessel, the electronic apparatus necessary for the measurements and acquisition of the experimental data, and the gas-control panel which is used to ll and purge the system and to prepare gas mixtures. The system is assembled and the rst measurements show a good behavior. A low temperature of 65 C is achieved which is one of the lowest reported temperatures up to date. 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Thermoacoustic; Refrigeration; Performance

1. Introduction During the last two decades thermoacoustic refrigeration is explored as a new cooling technology. Thermoacoustic refrigerators are systems which use sound waves to produce cooling power. They consist of a loudspeaker, attached to one end of an acoustic resonator (tube), which is closed at the other end. The loudspeaker sustains an acoustic standing wave in the resonator. A stack of parallel plates and two heat exchangers are appropriately installed in the resonator. The thermal interaction between sound and the surface of the stack generates a heat-pumping process. The way these systems work is extensively explained in the literature [1]. This paper describes the construction of a thermoacoustic refrigerator. The design procedure of the thermoacoustic refrigerator is described elsewhere [2]. The manufacturing process will be explained and the rst performance measurements will be presented. This paper starts with a description of the design and construction of the dierent parts, followed by an explanation of the preparation of the gas mixtures. The rst performance measurements results of our refrigerator will be given at the end of this paper.

2. Thermoacoustic refrigerator A schematic illustration of the thermoacoustic refrigerator is shown in Fig. 1. It consists of an acoustic driver placed in a housing, a gas-lled resonator in which a stack and two heat exchangers are placed, and a vacuum vessel in which the resonator is contained. The system is made out of separate components so that specic parts can be exchanged. In the following the design and construction of the dierent parts will be described in detail. 2.1. Acoustic driver The driver consists of a commercial moving-coil loudspeaker [3]. The choice of the driver is based on requirements like compactness, lightweight, low losses, and high Bl-factor [4]. The loudspeaker has been modied, as illustrated in Fig. 2, to meet some specic requirements. The dierent components are shown and are numbered for reference. A plastic housing, which originally covered the back of the loudspeaker, was removed so that our resonance back volume control system could be mounted. The original dome is not rigid enough to generate the high dynamic pressures required for our experiments. The fabric dome was cut o near the voice coil (2) and replaced by an aluminum cone which is glued onto the voice coil. Furthermore, the loudspeaker coil has a diameter of 54 mm, but a resonator with a diameter of

Corresponding author. Tel.: +31-40-247-4215; fax: +31-40-2438272. E-mail address: a.t.a.m.d.Waele@tue.nl (A.T.A.M. de Waele).

0011-2275/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 1 - 2 2 7 5 ( 0 1 ) 0 0 1 8 0 - 1

60

M.E.H. Tijani et al. / Cryogenics 42 (2002) 5966

Fig. 1. Cross-sectional illustration of the thermoacoustic refrigerator, showing the dierent parts.

Fig. 2. Illustration of the driver and driver housing. It is shown how the system is mounted. The dierent components are numbered: (1) Magnet. (2) Coil. (3) Reducer cone. (4) Accelerometer. (5) Dynamic pressure transducer. (6) Rubber O-ring. (7) Capillary. (8) Rolling diaphragm. (9) Electrical feed-through plug. (10) Electrical feed-through plug for the resonator sensors. (11) Bolts to attach the driver to the housing. (12) Bolts to attach the resonator. (13) Bolts to attach the vacuum vessel. (14) Water cooling tubes. (15) and (16) Accelerometer signal cables. (17) Dynamic pressure transducer cable. (18) Connectors. (19) Bolts to attach the lid to the housing . (20) Back volume control system. (21) Gas ll port. (22) Support system for the accelerometer cable. (23) Feed through channel for the wires coming from the resonator.

38 mm is used to meet the required cooling power. Hence, a reducing cone (3) is used with a taper from 54 to 30 mm. The light-weight aluminum cone has a thickness of 0.1 mm and is made of a conical part to which a base plate can be xed using six bolts, as shown in Fig. 3. The base plate has a thickness of 0.6 mm. The plate provides the needed driving area and works as a support for the accelerometer (4) which is glued to the back of the plate. With this construction for the cone one has access to the accelerometer, by removing the plate, for eventual reparations. A plate thickness of 0.3 mm was used rst, but after some troubles with the oscillation modes of the plate which caused the accelerometer to be detached from the back, we chose to use a thickness of 0.6 mm. The total mass of the reducer cone is 3.5 g. A rubber rolling diaphragm (8) is used to seal the driver housing from the resonator [5]. This diaphragm consists of an annular rolled membrane, which is xed between the basis plate and the cone (Fig. 3). The membrane is clamped between the plate and the cone. The outer side of the membrane is clamped between two

annular aluminum plates who support the driver and contain the capillary (7). The stainless steel capillary (7) serves to equalize the static pressure (10 bar) between the housing and the resonator and to provide acoustic sealing. A capillary is also mounted in the piston of the gasspring system for pressure balancing between the cylinder and the housing. The driver is mounted on a support brass plate which forms the interface with the bottom of the housing. The driver has a good thermal contact with the bottom of the housing, so that the heat dissipated in the driver is removed by the cooling water circulating in the copper tubing. An accelerometer (4) [6] is glued on the base plate of the pusher cone. It measures the acceleration from which the velocity u at that location can be deduced. A dynamic pressure transducer (5) [7] is also mounted near the moving cone to measure the amplitude of the dynamic pressure p at that location where a pressure

M.E.H. Tijani et al. / Cryogenics 42 (2002) 5966

61

Fig. 3. The rolling diaphragm mounting procedure. The annular membrane is rst clamped between two aluminum clamping plates which support the loudspeaker, protect the cone during motion, and contain the capillary which connects the housing and the resonator. The clamping plates are put together by six M3 bolts (only one bolt is shown here). Then, the membrane is xed between the cone and the basis plate via six M1 bolts.

volume can be varied from the outside of the housing by using a cylinder and a piston (Fig. 1). In this way an extra spring can be added to the loudspeaker. By varying the spring constant, the mechanical resonance frequency of the loudspeaker can be shifted to match the acoustical resonance frequency of the resonator. The back volume can be varied by varying the height of the piston. The gasspring system consists of a brass cylinder which is mounted on the back of the driver. By turning a crank, xed to the lid of the housing, a screw system varies the height of the piston. The cylinder has an inner diameter of 8 cm and a height of 15 cm and accepts a piston system mounted on the lid of the driver housing. A second cylinder of diameter 4.5 cm and height 15 cm can be used instead of the large diameter cylinder whenever a ne tuning is necessary. 2.3. Resonator The requirements for the dierent parts of the resonator have been discussed elsewhere [2,4]. The resonator consists of a large-diameter tube which contains the stack, a small-diameter tube, and a buer volume. A contraction is used to connect the large and small diameter tubes. An illustration of the resonator is depicted in Fig. 4. Actually, the large diameter tube consists of a part which is situated in the driver housing of length 3.2 cm, a part contained in the hot heat exchanger ange of length 0.5 cm and the stack holder of length 8.5 cm. In the following the design and construction of the dierent parts will be discussed. The requirements for the stack holder are rigidity and low thermal conductivity. We decided to make the holder out of the material POM-Ertacetal [4]. It has a an inner diameter of 38 mm, a wall thickness of 2 mm and a length of 85 mm. A rubber O-ring, mounted into the ange of the hot heat exchanger, is used for sealing. At the taper side, which is the cold side, the stack holder is attached to the tapered part. As this side becomes colder than 40 C, we used an indium O-ring to seal this junction. The tapered section constitutes the contraction part which connects the large diameter stack holder to the small diameter tube. The taper is machined out of an aluminum block. A taper angle of 30 is used in order

antinode exists. After many fatigue problems with the wires of the accelerometer, which broke down during operation of the refrigerator, we ended up with a good solution which consists of using coated stainless steel wires with a diameter of 0.125 mm which are soldered to the output connectors of the accelerometer. The wires are then glued to the top of the accelerometer. The wires are formed into a conical shaped spring, and enter a guiding stainless steel tube which is supported by a structure xed to the top of the driver by means of three legs (22), glued to the driver housing. Via standard cables and feed throughs (18), the signal is led to the outside of the set up. The dynamic pressure transducer is a dierential pressure sensor. This means that this transducer measures the acoustic pressure by reference to the static pressure at its back. A circuit is used to protect the loudspeaker coil from burn out, whenever high current is used. The circuit is based on the fact that the resistance of the coil is a function of temperature. The method consists of superimposing a small dc current of 10 mA on the alternating current. The corresponding dc voltage is a measure of the temperature. A maximal tolerable temperature of 100 C is considered. 2.2. Gasspring system The electroacoustic eciency of the loudspeaker, dened as the output acoustic power divided by the input electric power, can be maximized by matching the mechanical resonance frequency of the driver to the acoustical resonance frequency of the resonator [4,8]. It was shown that this shift can be realized by using the volume of gas at the back of the driver housing. This

Fig. 4. A schematic illustration of the resonator. The small diameter resonator is welded to the tapered section and the buer volume.

62

M.E.H. Tijani et al. / Cryogenics 42 (2002) 5966

that the connection can be kept short. The bottom side of the taper is welded to the small diameter tube. The stack side of the tapered section is connected to the cold heat exchanger with an indium O-ring. The small diameter tube is also made of aluminum, it has an inner diameter of 2 cm, a wall thickness of 3 mm, and a length of 20 cm. It is welded to the tapered section and to the buer volume. The buer volume is also made of aluminum. It has a volume of one liter, a wall thickness of 1 mm and a length of 27 cm. It is welded to the small diameter tube. A taper angle of 9 is used to reduce irreversibilities. The total mass of the resonator including stack holder, tapered section, small diameter tube, and buer volume is 350 g. 2.4. Stack The design and material requirements for the stack, which forms the heart of the refrigerator, have been discussed elsewhere [2,4]. To guarantee low thermal conductivity, Mylar [4] material is chosen. Because of the diculty of construction and the fragility of the parallel-plate stack, an approximate geometry is usually used [911], which consists of winding a long sheet around a rod to get a spiral stack. The spacing between the layers is realized by shing line spacers glued onto the surface of the sheet. The gaps between the layers are considered to approximate the parallel-plate channels. This assertion will be checked experimentally. We have constructed two spiral stacks, and seven parallel-plate stacks with a spacing varying between 0.15 and 0.7 mm. In the following the construction of the stacks will be described. 2.4.1. Spiral stack The spiral stack consists of a long polyester Mylar sheet, wound around a support PVC-rod with a diameter of 6 mm into a spiral, as shown in Fig. 5. A channel structure is realized between the layers of the spiral by

using shing line spacers of 0.38 mm diameter. The sheet has a thickness of 0.06 mm. A length of 3.8 m was necessary to obtain a stack diameter of 38 mm. In the rst 50 cm of the sheet a spacer distance of 3.5 mm is used, above this length a distance of 7 mm is used. Two stacks have been manufactured. The two stacks have the same spacing and length. The only dierence is the tension which is used during the rolling process. For the rst stack an excess tension was used which resulted in a stack with deformed channels. The second stack was carefully rolled up with just enough tension which results in an uniform channel structure. These two stacks are used to show the inuence of the rolling process on the performance of the cooler [4]. 2.4.2. Parallel-plate stack The parallel-plate stacks consist of parallel plates which are spaced by shing line spacers glued between the plates, as shown in Fig. 6(b). This type of stack is more dicult to manufacture than the spiral stack. Spiral stacks, once rolled up, results in a rigid structure. Parallel-plate stacks are more fragile. One of the problems was to stick nylon shing line onto polyester plates.
(a)

(b)

Fig. 5. An illustration of the spiral stack.

Fig. 6. Illustration of the manufacturing technique of the parallelplate stack. A plate is placed between the brass rods, then the glue coated shing line coming from the syringe is wound around the rods to get parallel strings. The strings are then pressed onto the plate and an other plate may be placed. The procedure is repeated until the needed height of the stack is reached: (a) setup; (b) a parallel plate stack as released from the work plate.

M.E.H. Tijani et al. / Cryogenics 42 (2002) 5966

63

A shing line, with a rough surface, in combination with 3M Scotch-Grip 7312 glue gave a good result. For ease and speed of construction a special mounting setup is built, as shown in Fig. 6(a). During the winding process just enough tension is used, to avoid deformation of the stack. After the hardening of the glue, the rectangular stack was machined into a cylindrical shape using a willing technique. The building of the stack takes one day, and the shaping process two full days. In total seven stacks have been manufactured with a spacing varying between 0.15 and 0.7 mm. A detailed description of the manufacturing procedure of the stacks can be found elsewhere [4,12]. 2.5. Heat exchangers The requirements for the heat exchangers have been discussed elsewhere [2]. In the following, we will explain the construction procedure of the copper heat exchangers. As the hot and cold heat exchangers dier only in length, the following discussion is valid for both. The heat exchangers have a sine channel structure which is chosen because of ease of construction. Two copper sheets are spirally wound together to provide this geometry. The sheets are 0.1 mm thick and 2.5 mm wide for the cold heat exchanger and 5 mm wide for the hot heat exchanger. One sheet is at and the other sheet has a sine shape, as shown in Fig. 7. The sine shape is achieved by passing a at sheet between a toothed wheel system. The top-to-top height of the sine structure is controlled by the spacing between the wheels and the tooth size. This height is chosen to be 0.5 mm so that the porosity of 70% is obtained. The heat exchangers are manufactured in the following way: The two sheets are rst soldered together at one end. Then the two sheets are wound together to obtain a spiral structure, as shown in Fig. 7. The sine copper sheet provides both spacing (channels) and contact for the radial heat transfer path. Once the spiral is rolled up, the other end of the two sheets is soldered to the external surface of the spiral. Finally, to improve the radial heat

transfer and to hold the spiral together so that rigidity can be ensured, a copper screen 0.5 mm thick is soft soldered to one face of the spiral. During the soldering, care was taken to avoid the ow of solder into the channels. The total sheet length of 216 cm is rolled up to obtain a spiral with a diameter of 38 mm. The next step is to mount the heat exchangers in the refrigerator. The hot heat exchanger is soft soldered in a copper ange which also provides connection to the stack holder and the driver housing. The copper ange has a good thermal contact with the bottom of the driver housing through which cooling water is circulating. The cold heat exchanger is mounted and soldered in a copper ring which can be screwed in the aluminum tapered section.

3. Gas mixture control In order to ll and evacuate the resonator, a gas panel is used. As the driver housing and the resonator are connected via the capillary with a high ow resistance (Fig. 2), a quick lling of the system will result in a pressure dierence over the thin aluminum driver cone which may be damaged. The maximal pressure dierence which the cone can withstand is 0.5 bar. The maximal lling speed is 12 mbar/s. In order to prepare the gas mixtures, three containers of one liter are placed in the gas panel. During the preparation of the mixture a waiting time of about one and half an hour is allowed for the mixture to equilibrate. The refrigerator is lled up to 10 bar. We determined the composition of the mixture by measuring the acoustic resonance frequency of the refrigerator. This frequency is related to the resonance frequency of the system for pure helium by the expression s fHe xM2 1 xMHe ; 1 fmix MHe where fHe , and fmix are the resonance frequencies for pure helium and mixture, respectively; M2 and x are the molar mass and mole fraction of the other noble gas component, MHe is the molar mass of helium.

4. Thermal insulation The heat leak as from the hot side of the stack to the cold one is minimized by using low thermal conductivity material for stack and stack holder, as discussed previously. Radiation is reduced by using Multilayer Insulation (MLI, usually named superinsulation). The aluminum resonator is polished to provide low emissivity. Furthermore, gravity driven convection in the refrigerator is avoided by putting the cold side down.

Fig. 7. Illustration of the heat exchanger. Two copper sheets, one at and the other with a sine shape, are rolled up together to realize a spiral structure with sine channels (left side). Expanded view of the channel (right side).

64

M.E.H. Tijani et al. / Cryogenics 42 (2002) 5966

The residual heat leak can be determined by means of a warm up measurement in the following way: The refrigerator is allowed to cool down and reach the steady state. Then the cooler is switched o. Subsequently the system warms up. The heat leak is given by _ leak C dTc ; Q dt 2

5.2. Electroacoustic eciency The determination of the electroacoustic eciency of the driver requires knowledge of the input electrical power and the acoustic output power [8]. The determination of the electrical power requires the knowledge of the voltage over the speaker, the current through the coil, and the phase dierence between them. Analogous a knowledge of the acoustic pressure amplitude, acoustic velocity, and the phase dierence between them is required to determine the acoustic power. We use four lock-in ampliers [13] to measure the four signals and their phases. In this way the powers can be calculated and hence the electroacoustic eciency. For electroacoustic eciency measurements a small electrical input power is used (<200 mW). In summary, the equations used for the electroacoustic eciency calculations are the electrical power input to the driver [4] _ e 1 VI cos h; P 2 4

where Tc is the temperature at the cold heat exchanger and C is the heat capacity of the cold side. If we assume that the heat leak can be written as _ leak RTh Tc ; Q where R C =s is a thermal resistance, s is a time constant, then the time dependence of Tc has the form Tc t Th DT0 et=s ; 3

where Th is the constant temperature of the hot heat exchanger and DT0 is the temperature dierence in the beginning of the warm up measurement. The time constant s is obtained by tting the expression (3) to the warm up data. The leak determined by this method is then reduced by the leak due to the thermal conductance in the stack region (gas + material), because this leak is an internal system leak. Once the stack is mounted in the resonator, the resonator is coupled to the driver housing via the hot heat exchanger copper ange (Fig. 1). Between the stack and the cold heat exchanger a spacing of 0.6 mm is used to prevent the channels of the stack and those of the cold heat exchanger from blocking each other. This spacing is small in comparison with the gas excursion at that location. At the interface between the stack and the hot heat exchanger, a spacing of 0.2 mm is used.

where V is the amplitude of the voltage over the driver, I is the amplitude of the current through the coil, and h is the phase dierence between the voltage and the current. The acoustic input power is given by [4] _ 1 pU cos u; W 2 5

where p is the amplitude of the dynamic pressure, U is the volume velocity amplitude and u is the phase difference between the pressure and the volume velocity. The electroacoustic eciency is then given by [4,8] gea _ W : _ Pe 6

5.3. Refrigerator performance The performance of the refrigerator is dened as the ratio of the transferred heat at the cold heat exchanger and the acoustic power used to achieve the transfer. The heat load is supplied by an electrical heater which is placed at the cold heat exchanger. Thermometers are used to monitor the temperatures at the hot heat exchanger, the cold heat exchanger, and at the buer volume [14]. A typical performance measurement is achieved as follows: rst the refrigerator is pressurized with a given gas mixture. Then the resonance frequency is determined. After that the amplitude of the dynamic pressure is increased. When the desired amplitude is reached the refrigerator is allowed to cool down and stabilize. Then step wise a heat load is applied to the cold heat exchanger. Each time when the steady temperature is reached a set of parameters is recorded, consisting of Tc ,

5. Measurement procedure 5.1. Resonance frequency Usually, the rst thing to determine is the fundamental acoustic resonance frequency of the refrigerator which is the operation frequency of the system. Two lock-in ampliers [13] are used to measure the velocity and the acoustic pressure signals and their phases. The phase dierence is used to determine and control resonance. The resonance frequency is dened as the resonance for which the phase dierence is zero. The resonance frequency can also be determined by measuring the pressure amplitude as function of the frequency. Whenever resonance takes place a peak shows up.

M.E.H. Tijani et al. / Cryogenics 42 (2002) 5966

65

Th , f, p1 ; u1 ; pm , etc. [4]. Once the measurement for a given type of gas and for given operation conditions is nished, operation conditions may be changed or another gas mixture may be lled and the measurement procedure is repeated. The coecient of performance COP of a refrigerator is dened as COP _C Q ; _ W 7

_ C is the cooling power and W _ is the acoustic where Q _ power used to pump QC : The quantity COPC Tc Th Tc 8
Fig. 8. A typical cool-down measurement showing the temperatures of the hot heat exchanger Th , the buer volume Tbuffer , and the cold heat exchanger Tc as functions of time. The used average pressure and the drive ratio are indicated in the graph. The working gas is helium at 10 bar.

is called the Carnot coecient of performance which is the maximal performance for all refrigerators. The coecient of performance relative to Carnots coecient of performance is dened as COPR COP : COPC 9

6. Measurements The performance measurements are described in much more detail elsewhere [4,12,15]. Here we will limit ourself to a cooldown measurement and some performance measurements with pure helium. The performance measurements for the refrigerator are presented in plots of COP, COPR, and DT given by DT Th Tc 10

_ which is the sum of as functions of the total heat load Q the heat load applied by the heater and the heat leak. A parallel-plate stack with a plate spacing of 0.3 mm was mounted in the resonator. Helium at 10 bar was used as working gas. Three dierent drive ratios have been used. The drive ratio D is dened as the ratio of the amplitude of the dynamic pressure p1 and the average pressure pm . A typical cooldown measurement was performed using a drive ratio D 2:1%. The behavior of the temperatures of the hot heat exchanger Th , buer volume Tbuffer , and cold heat exchanger Tc as functions of time are plotted in Fig. 8. Initially, the refrigerator is at room temperature. Once the driver is switched on, the temperature of the cold heat exchanger Tc drops quickly, followed by the temperature of the buer volume Tbuffer until steady state is reached. The temperature of the hot heat exchanger is held constant by the cooling water. The whole part of the cooler below the cold heat exchanger cools down. A low temperature of 65 C was reached in this measurement.

Fig. 9. Performance measurements with helium showing the eect of the drive ratio. An average pressure of 10 bar, and a stack with a plate spacing of 0.3 mm is used. (a)(c) The drive ratios are indicated at the corresponding curves: (a) COP; (b) COPR; (c) DT .

66

M.E.H. Tijani et al. / Cryogenics 42 (2002) 5966

Fig. 9 shows the performance measurements using three dierent drive ratios. As can be seen from Fig. 9(a), the COP increases as the heat load increases for all drive ratios. The slope of the COP curves decreases as the drive ratio increases. This behavior of COP as functions of the drive ratio is to be expected, because the cooling power is proportional to the square of the drive ratio [1]. The COPR shows a parabolic behavior with a maximum for all drive ratios, but the magnitude and the position of the maximum changes as the drive ratio changes (Fig. 9(b)). The value of the maximum increases as a functions of the drive ratio up to 1.4% and then remains nearly constant. As can be seen from Fig. 9(c), the temperature dierence is practically a linear function of the heat load. The slope of the DT line decreases as the drive ratio increases. The optimization of the refrigerator is described elsewhere [8,12,15].

References
[1] Swift GW. Thermoacoustic engines. J Acoust Soc Am 1988;84:114680. [2] Tijani MEH, Zeegers JCH, de Waele ATAM. Design of thermoacoustic refrigerators. Cryogenics [submitted]. [3] The performance measurements are all measured with a Dynaudio loudspeaker type D54 AF. [4] Tijani MEH. Loudspeaker-driven thermo-acoustic refrigeration. PhD Thesis, Unpublished, Eindhoven University of Technology, 2001. [5] DIACOM Corp. 5 Howe Drive, Amherst, NH 03031, USA. [6] Kistler Instrument Corp. Type 8614A1000M1, Range 1000 g, sensitivity 2.45 mV/g, mass 0.7 g. [7] Endevco Corp. Piezoresistive pressure transducer, Model 8510B-5, Range 34.5 kPa, Sensitivity 9.36 mV/kPa. [8] Tijani MEH, Zeegers JCH, de Waele ATAM. A gasspring system for optimizing loudspeakers in thermoacoustic refrigerators. J Appl Phys [submitted]. [9] Hoer TJ. Thermoacoustic refrigerator design and performance. PhD Dissertation, Physics Department, University of California at San Diego, 1986. [10] Garrett SL, Ade JA, Hoer TJ. Thermoacoustic refrigerator for space applications. J Thermophys Heat Transfer 1993;7:595. [11] Garrett SL. Thermoacoustic life science refrigerator. NASA Report, No. LS-10114. 1991. [12] Tijani MEH, Zeegers JCH, de Waele ATAM. The optimal stack spacing for thermoacoustic refrigeration. J Acoust Soc Am [submitted]. [13] Lock-in amplier, EG&G Princeton Applied Research, Model 5210. [14] Two-terminal IC temperature transducers (Analog Devices AD590) have been used. [15] Tijani MEH, Zeegers JCH, de Waele ATAM. The Prandtl number and thermoacoustic refrigerators. J Acoust Soc Am [submitted].

7. Conclusion In this paper the manufacturing procedure of a thermoacoustic refrigerator is discussed. The construction of the dierent parts of the refrigerator is described in detail. The system has been assembled and the rst performance measurements have been done. The measurements show that the system behaves very well as expected. A low temperature of 65 C is achieved. The refrigerator is used to study the eect of some important thermoacoustic parameters, such as the Prandtl number using binary gas mixtures, and the stack plate spacing. The results of this measurements can be found elsewhere [12,15].