Measuring Methods of Thermoelectric Coolers eng

MEASURING METHODS OF THERMOELECTRIC COOLERS NONSTATIONARY DYNAMICS IN Z-METERING
1
2
1
L.Yershova , I.Drabkin , V.Volodin , D.Kondratiev
τ max =
1
Institute of Physics and High Technologies, Moscow, Russia
Institute of Chemical Microelectronics Problems, Moscow, Russia
Fax: +7-095-132-5870
1. Introduction
Temporal characteristics of a thermoelectric cooler (TEC) are important
performance parameters for any device involving TEC. In paper [1]a single
thermoelectric pellet non-stationary processes in the so-called regular mode [2]
were studied and the obtained formulae were generalized for one-stage TECs
loaded by various heat to be pumped. In paper [3] there are derived expressions
for the time constant of single-, two- and multi-stage TECs. This paper studies
transient processes concerned in Figure-of-Merit measurements with the Z-meter.
It compares experimental and theoretical [3] results and yields the evaluating
approach for obtaining relaxation time values in real thermoelectric devices.
2. TEC Time Constant Theory
Let us define the time constant as the period enough for the temperature
difference between the initial and stationary states decreases e-proportionally. In
paper [3] corresponding differential equations are solved and maximum time
constants for the slowest exponential processes are found. The thermoelectric
cooler (TEC) electrical current limit is as follows:
j <<
αL
,
κ0
(2)
where С1 is the heat capacity of cold side junctions, substrate and objects to be
cooled; N is the module pellets number, s0 – a pellet cross-section.
2) One-stage TEC, its substrates in free heat exchange with the medium
(free TEC):
1
2
LC1
,

Lαj 
 1 +
 s0κ 0 N
κ 0 

(1а)
where j – the current density, α – the Seebeck coefficient, κ0 – the average n/p
pellets thermal conductivity, L – pellets height. Eq. (1) means the TEC current
must be much lower than the maximum module current Imax, that is the current
at which the TEC yields the highest temperature difference at zero heat load:
I << I max ,
(1b)
Designate the slowest expo-process as τmax. There are certain cases when
this value is defined.
1) One-stage TEC, its hot side temperature constant and cold side
adiabatically isolated:
τ max =
C1C 2 L
,
(C1 + C 2 )κ 0 Ns0
(3)
where C1, C2, are the heat capacities of all the elements on the cold and hot
substrates.
Eq. (3) is the maximum time constant at j=0. When j≠0 the time constant
is to be found via numerical solution of the corresponding characteristic equation.
For the free module case this time is approximately twice lower than for
the one with the thermally stabilized hot side, see (2).
3) Two-stage TEC, its hot side temperature constant and cold side
adiabatically isolated. For this variant two exponential processes – slow and fast –
manifest themselves. Their time constants are:
τ max ≈ τ 1 max + τ 2 max ,
(4а)
τ=
Here
τ 1 max
and
τ 2 max
τ max 1τ max 2
(τ max 1 + τ max 2 )
(4b)
are time constants of the one-stage TECs formed by each
cascade.
Therefore for an n-stage TEC time constant can be expressed as the sum of
the each cascade times and all possible combinations like that in (4b):
τ max ≈ τ 1 + τ 2 + ... + τ n ,
(5а)
τ ij =
τ max iτ max j
(τ max i + τ max j ) ∀ i , j
(5b)
3. Time Constant Measurement
The Device DX3065 allows measuring the parameters of a TEC both in
the air and while effective heat rejection from the TEC hot side is carried out, i.e.
for the TEC mounted onto some heat sink.
The function of the device is double. On the one hand, it enables testing
the TEC Figure-of-Merit by the Harman method; on the other hand, it
telemetrically tracks the kinetics of the TEC transfer to the stationary state, for it
is this state the Harman method can be applied to. So, one device is not only Zmeter but also τ-meter. The accompanying software allows observing the kinetics
data by measuring the temporal behaviour of Seebeck voltage Uα(t) and
interpolating the data by the exponent:
(6)
U α (t ) = Ustα (1 − e −t / τ ) ,
where Ustα - the stationary Seebeck voltage, аnd τ - TEC time constant.
Fig. 1. Z-meter exterior
For checking if approximating the kinetics by a single exponent is correct
additional study in the semi-logarithmic scale is possible:
 Ustα − U ( t ) 
 ,
f ( t ) = ln
Ustα


(7)
4. One-stage TEC: Experiment and Theory
Here there are experimental and theoretical [3] data for one-stage TECs
both in the free substrates option and with the hot sides thermally stabilized.
Denote I for the electric current, τexp, τtheory for the measured and calculated time
constants, D for dimensionless root-mean-square declination, normalized to the
stationary Seebeck voltage. In the calculations hereafter the following parameters
are used: thermoelectric material heat capacity 0.13 J/g, density 7.5 g/сm3, the
corresponding ceramics values are 0,8 J/g and 3,5 g/сm3, for the solders – 0,17 J/g
and 9,3 g/сm3, the ceramics is 0.5mm thick.
Table 1.
TECs measured parameters
TEC Type
Substrate sizes, mm2
s 0,
mm2
L, mm N
Imax, A
Cold
Hot
1MT03-004-13 2,0х1,0
2,3х2,3
0,09
1,3
8
0,3
1MC04-032-15 6,4х6,4
6,4х8,0
0,16
1,5
64
0,5
1MC06-030-05 8,2х8,2
8,2х8,2
0,36
0,5
60
3,2
1MC06-060-05 10,0х12,0 12,0х12,0
0,36
0,5 120
3,2
1MC06-018-15 6,0х6,0
6,0х8,0
0,36
1,55 36
1,1
1MC06-024-15 8,0х8,0
8,0х8,0
0,36
1,55 48
1,1
1MT04-059-16 8,0х7,0
8,0х7,0
0,16
1,6 118
0,45
Experiment and theory were taken at two current values 5mА and 25 mА. Table
1 shows that these currents satisfy the requirement (1b).
Table 2.
Measured and calculated one-stage free-substrated TECs time constants
TEC Type
I, mА
D
τtheory, s
τexp, s
1MT03-004-13
5
1.92
2.43
0.0018
25
2.87
2.55
0.0007
1MC04-032-15*
5
2.56
3.20
0.0003
25
3.53
3.14
0.0001
1MC06-030-05*
5
0.82
0.75
0.0012
25
0.82
0.71
0.0003
1MC06-060-05
5
0.80
0.68
0.0009
25
0.80
0.65
0.0002
1MC06-018-15*
5
2.56
2.23
0.0007
25
2.57
2.26
0.0002
1MC06-024-15
5
2.56
2.44
0.0005
25
2.95
2.47
0.0002
1MT04-059-16
5
2.91
2.47
0.0002
25
2.91
2.50
0.0001
The sign «*» in Table 2 marks TECs, tested also on the heat sink – see
Table 3.
Table 3
Measured and calculated time constants of the one-stage TECs mounted on
the heat sink
TEC Type
I, mА
D
τtheory, s
τexp, s
1MC04-032-15*
5
6.59
6.10
0.0057
25
6.43
5.77
0.0098
1MC06-030-05*
5
1.65
1.28
0.0036
25
1.64
1.45
0.0062
1MC06-018-15*
5
4.77
3.96
0.0049
25
4.71
3.96
0.0071
Table 3 proves that calculated time constants are mainly 10 – 20 % higher
than the tested ones. This may be explained by the heat exchange with the
environment being allowed for not accurately enough. These processes are apt to
diminish time constant. It is also seen that time constants for the free module are
nearly twice lower than those for the modules on the heat sinks.
Typical telemetric data are given in Fig. 2.
0
120
80
Ln(U), Thot=const
-1
U in air
-2
60
lnU
U, mV
100
40
-3
Ln(U) in air
-4
U, Thot=const
20
-5
0
0
5
10
t, s
15
20
-6
0
2
4
6 8 10 12 14
t, s
Fig. 2. Measured Seebeck voltage values for the TEC 1МС04-032-15 at 25mА
The data in the logarithmic scale depend on time linearly. Corresponding
time constants found from the linear approximation yield 3s and 5.5s, which is
agreeable with the results of Tables 2, 3. It is clear that one-stage TEC kinetics is
covered by single characteristic time.
5. Two- and Three-stage TECs: Experiment and Theory
The calculations for the multi-stage TECs were only taken at 5 mА and
only for the thermally stabilized hot sides. Measurements were carried out at 5
mА both for this very case and for the free modules. The parameters of the tested
two-stage TECs are in Table 4.
Table 4.
The two-stage TECs parameters
Substrate sizes, mm2
s0,
TEC Type
N1
2 l, mm
Сold Medium Hot mm
2MC06-10-10 3,2х3,2 4,0х4,0 4,0х4,0 0,36 1,05
6
2MC04-039-15 4,9х4,9 6,5х6,5 6,5х6,5 0,16 1,55 36
N2
14
42
Imax,
A
1,3
0,3
Here are the calculations results.
Table 5.
Calculated time constants of the two-stage TECs mounted on the heat sink
TEC Type
τ1, s
τ2, s
τ1+τ2, s
τ1τ2/(τ1+τ2), s
2MC06-10-10
5.04
3.53
8.57
2.07
2MC04-039-15
6.83
9.80
16.63
4.02
We see that the temporal relaxation for two-cascade modules reveal a fast
and slow components. So a single exponent approximation is only admitted as a
sort of estimation. To what extent is it acceptable? In Table 7 the experimental
results are given.
Table 7.
Experimental time constants for the two-stage TEC
Mounted Module
Free Module
TEC Type
D
D
τexp, с
τexp, с
2MC06-10-10
3.1
0.0012
5.02
0.0011
2MC04-039-15
9.3
0.0007
12.5
0.0001
Consider time behaviour of 2МС04-039-15. Fig. 3 depicts the Seebeck
voltage dynamics for two heat exchange variants.
Table 9.
Calculated time constants of the three-stage TECs
mounted on the heat sink
TEC Type
τ1, s
τ2, s
τ3, s
τ1+τ2+τ3, s
3MC06-024-13
6.11
5.32
4.73
16.16
3MC07-098-115
9.54
5.32
3.08
17.94
Ln(U), Thot=const
-t/12.5
Ln(U) in air
0
5
10
15
20
25
30
35
40
t, s
In the three-cascade case except the slow evolution described by the
summed time constant there are three faster transitory processes referring to (5b).
The averaged experimental data on a single exponent regression are offered in
Table 10.
Fig. 3. 2МС04-039-15 dynamics for two heat exchange variants
It is seen that for the mounted option the logarithm has a linear character.
Fig. 4 discovers the same TEC dynamics in the free option.
TEC type
3MC06-024-13
3MC07-098-115
lnU air
0,0
-1,0
-t/15-0.34
-2,0
-t/6.8
-3,0
-4,0
0
5
10
15
20
t, s
25
30
35
40
Both TECs are very sensitive to the environmental change on the hot side.
The reason may be a high cascading coefficient for both these modules (see the
similar case for 2MC06-10-10, Table7).
Below there are time behaviour pictures for the TEC 3MC06-024-13. In
Fig. 5 one sees the dynamics for two heat exchange variants.
Fig. 4. Logarithmic time behaviour for 2МС04-039-15 in the free heat exchange
option at 5mA
Table 8.
Substrate sizes, mm2
Imax,
s0,
TEC type
2 l, mm N1 N2 N3
A
Сold Medium Hot mm
4,0х4,0,
3MC06-024-13 2.5х2.5
6.1х6.1 0.36 1.3 6 14 32 1.05
6.1х6.1
3MC07-09810х10,
8х8
12х12 0.49 1.15 36 42 132 1.45
115
12х12
0,0
7
U in air
6
4
U, Thot=const
3
Ln(U), Thot=const
-0,5
5
U, mV
Time constants defined in the semi-logarithmic scale are τ =15s и τ=6.8 s,
which is quite close to the calculated results of Table 5.
The study for the three-stage TECs is simular. The parameters of the threestage TECs and the theory results can be found in Tables 8 and 9.
Table 10.
Measured results for the three-stage TECs
Free Module
Mounted Module
D
D
τexp, s
τexp, s
5.9
0.0009
11.2
0.0005
4.1
0.0006
8.7
0.0003
lnU
lnU
0,0
-0,5
-1,0
-1,5
-2,0
-2,5
-3,0
-3,5
-4,0
-1,0
-1,5
Ln(U) in air
2
-2,0
1
0
-2,5
0
10
20
30
t, s
40
50
60
0
4
8
t, s
12
16
Fig. 5. Measured Seebeck voltage dynamics for the TEC 3МС06-024-13 at 5 mА
The logic found out on the two-stage option remains fair. It is illustrated
by Fig. 6.
0,0
-0,5
-0,2
-0,4
lnU air
-1,0
-t/10.7-0.5
-1,5
-2,0
-t/3.8
-2,5
-t/7.7-0.5
-3,0
0
2
4
6
8 10 12 14 16
t, s
lnU Thot=const
0,0
-0,6
-0,8
-t/11.2-0.02
-1,0
-1,2
-1,4
-1,6
0
2
4
6
8 10 12 14 16
t, s
Fig. 6. Logarithmic time behaviour for 3МС06-024-13 for the free heat exchange
and mounted options at 5mA.
Fig. 5 convinces one of the fast component presence in the free module
testing. This can be captured by the kink of the curve in the semi-logarithmic
scale.
The three-cascade case reveals at least three linear curve parts
corresponding to different time constants. Comparison of the theory and
experiment shows that the tested values are some 30 % lower than the calculated
ones, which is no wonder if taking into account the one-stage preceding
disagreement.
Conclusion
With the help of the Z-meter DX3065 it is possible to study the process of
TECs transfer to the stationary state both in the free heat exchange option
(standard Z-meter configuration) and in real performance conditions for modules
mounted on the headers, heat sinks, in devices, etc.
Comparison of the theory and experiment shows that the tested values
are 10 – 30 % lower than the theoretically predicted figures. However the theory
correctly catches the basic relaxation characteristics of TECs as dependent on
their real operational-environmental conditions.
Time constants evaluated by Z-meter allow more accurately describing
non-stationary kinetics in real thermoelectric devices.
1. E.I.Astakhova, V.P.Babin, U.I.Ravich. Calculation and Measurement of the
Cooling Thermoelement in the Regular Mode. Eng.Phys.J., 62, 1992, 284.
2. G.M. Kondratiev. Regular Thermal Mode. Moscow, GITTL, 1954, 408.
3. I.A. Drabkin. Transient Processes in Thermoelectric Cooling Modules and
Devices. (Ibid)