Anda di halaman 1dari 3

Comment on Adaptive tuning of an electrodynamically driven

thermoacoustic cooler [J. Acoust. Soc. Am. 111(3),


12511258 (2002)] (L)
Robert W. M. Smitha)
Penn State University Applied Research Laboratory, State College, Pennsylvania 16804-0030
Received 19 August 2003; revised 15 December 2003; accepted 16 December 2003
Corrections for equations contained in a recent paper Li et al., J. Acoust. Soc. Am. 1113, 1251
1258 2002 are provided and unclear items are discussed. Comments are provided on some of the
observations and conclusions of this paper. 2004 Acoustical Society of America.
DOI: 10.1121/1.1647511
PACS numbers: 43.35.Ud, 43.58.Wc, 43.38.Dv RR Pages: 973975

In a recent paper, Li et al. J. Acoust. Soc. Am. 1113, be two definitions of the variable U c (s). U c (s) is first de-
12511258 2002 developed a model for electro-acoustic fined in the text, and explicitly in Fig. 5 to be ...the voltage
conversion efficiency for an electrically driven thermoacous- on the output side of the amplifier... Later, in the text im-
tic refrigerator. In the model they employed a transfer func- mediately leading Eq. 11, which is the first mathematical
tion formulation, shown in their Fig. 5. From this formula- expression containing U c (s), the authors appear to establish
tion, the authors obtain several relations, primarily aimed at a second definition, such that this quantity represents coil
obtaining the electro-acoustic conversion efficiency. voltage.
First, some corrections are appropriate. In the paper, Assuming the first definition of U c (s), Eq. 11 in the Li
Eqs. 8 and 10 are missing a factor of K in the numerator. paper is observed to be in error, since this equation must
Equation 9 is missing a factor of K(Bl) in the numerator. simply reduce to the factor K for the case when one chooses
In the development of subsequent equations, interpretation of H(s)0. For this case the correct form of the equation is
the paper becomes more difficult because of what appear to given by

U c s K LCs 2 RCs1 1G m s AG PV s Bl 2 G m s Cs
. 1
U s LCs 2 RCs1 1G m s AG PV s Bl BlKH s G m s Cs

Assuming the second possible definition of U c (s), and here this second definition shall be renamed as U coil(s), then Eq. 11
in the Li paper is also in error. This is most readily observed again in the limiting case where H(s) is zero. For such a case,
the equivalent electrical impedance of the postamplifier system is

Z eq
1
G e s
Bl 2
G m s
1G m s AG PV s
. 2

Then, the ratio of the postamplifier voltage to the coil voltage can be equivalently expressed as the ratio of the impedance with
and without the series capacitance. Since U c (s)KU(s), when H(s)0 this results in a form closest to Eq. 11 in the Li
paper in

U coil s
U c s
H(s)0

U coil s

Cs LsR 1G m s AG PV s Bl 2 G m s
KU s LCs 2 RCs1 1G m s AG PV s Bl 2 G m s Cs
. 3

One may see in Eq. 11 of the Li paper that K appears only this second possible definition.
in products of H(s). Using only the transfer function formu- The equations referenced are combined to produce their
lation presented in the Li paper, it is does not appear straight- result for efficiency shown in Eq. 13, which may be recog-
forward to formulate a full corrected version of Eq. 11 for nized to be in error since the second term in the denominator
is dimensionally incompatible. If we define here a new quan-
tity, Upa,I , to be the phase angle between the postampli-
fier voltage and the current, a correct replacement for the Eq.
a
Electronic mail: rws100@psu.edu 13 by Li et al. is given here in Eq. 4, after some further
discussion.
J. Acoust. Soc. Am. 115 (3), March 2004 0001-4966/2004/115(3)/973/3/$20.00 2004 Acoustical Society of America 973

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 142.150.190.39 On: Sat, 20 Dec 2014 06:39:34
With UcI defined as the phase between coil voltage and 13. This shows that the addition of a capacitor in series
current, as described in the Li paper, Eq. 13 in the Li paper with the coil does not affect the efficiency. This observation
can be corrected with the addition of a factor of G m (s) in the is in concert with established theory.1,2 However, the empha-
second term of the denominator if the series capacitor is sis on the capacitance, and lack of mention of the role of the
omitted, since this case collapses to our Eq. 4. With this machine inductance, could be misleading, since it too has no
correction it is believed that this equation also produces the
impact on the efficiency. In fact, it is not necessary, as ap-
correct solution when a capacitor is present, because one
pears to have been the goal in formulating the equations, to
does not expect a series electrical nondissipative element to
have any effect on the efficiency, and one expects the impact try isolate the capacitance C. Using U c (s) and the other
on efficiency of H(s) to be tantamount to varying the input variables as defined in the paper, and Upa,I , which as a
voltage on an assumed linear system. reminder is the phase between the postamplifier voltage
About their result in Eq. 13, the authors of the paper U c (s) and the current, the following equation for efficiency
observe that ...the capacitance C does not appear in Eq. results:

A G PV s Bl 2 G m s 2 cos PV
s . 4
1AG PV s G m s LsR1/Cs 1AG PV s G m s Bl 2 G m s cos Upa,I

In this case, while L and C appear explicitly in the equation, frequency of the source. Their model predicts a maximum
they also appear implicitly in cos(Upa,I), as is true for L in electro-acoustic conversion efficiency of 27.4%. Theory pre-
cos(UcI) in the Li paper, in such a manner as to cancel any dicts that 70 percent is possible, based on the source param-
impact on efficiency. An alternate form for the efficiency eters given in their Table I, for a properly coupled source.
shows this more explicitly. Again, using the block variables They state that a mere 0.8-Hz change in frequency from
from the Li paper, and defining for convenience the quantity the optimum efficiency point to the acoustic resonance re-
which appears in a block in Fig. 5b sults in a factor-of-4 reduction in efficiency, to 7.4%.
The conclusion that the authors reach, based in part on
G m s
G B s , 5 their model predictions, is that ...the acoustic resonance fre-
1AG PV s G m s quency does not match the frequency for optimal effi-
ciency... This is undoubtedly true. In practical systems,
the efficiency is
however, the magnitude of the impact is substantially smaller
Re AG PV s than this model might suggest and is often negligible. Our


s , 6 experience3 indicates that for practical systems i.e., systems
1 1 1
Re Re with optimum achievable efficiencies very near the theoreti-
G B s Bl 2 G B s 2 G E s
cally achievable electro-acoustic conversion efficiency, based
where G E represents the blocked electrical admittance of the on the driver parameters, that the indirect efficiency con-
transducer and a series capacitor, as defined in their paper. trol employed by Hofler4 and others5,6 can work very well
Since only the real part of (1/G E (s)) enters this equation, the i.e., tracking and operation at the acoustic resonance and
series electrical impedance associated with the coil induc- can produce efficiencies often reaching within measurement
tance and the capacitor are explicitly seen to have no impact error a fraction of a percent of the predicted optimum
on the efficiency. achievable efficiency. In general, this optimum will occur at
In Fig. 6 of the Li paper, the authors compare predicted some frequency slightly different than the acoustic resonance
and measured efficiency using their model. Assuming the for the case when the source mechanical and acoustic reso-
authors have correctly calculated the efficiency, the result nance frequencies are close but not exactly the same. Given
shows a notable difference between the frequency approxi- the relative simplicity of tracking the acoustic resonance, and
mately 1.3 Hz or 1%) and predicted magnitudes 16% the fact that the penalty is negligible for not directly detect-
relative error for the location and magnitude of the effi- ing the maximum efficiency, tracking the acoustic resonance
ciency, although the two curves are not identified in the plot. is an effective and practical engineering strategy.
It is also not clear how G PV was obtained for the model, to In observing that the acoustic resonance does not corre-
compare the model and the measurement. The differences spond to the optimum efficiency, and the substantial degra-
apparent in Fig. 6 might be attributed to nonlinearity or fre- dation in efficiency for their modeled system corresponding
quency dependence in the motor parameters, but the authors to operating at acoustic resonance, Li et al. cite Hunt2 in
make no comment on the origin of the discrepancy. support to say While it is customary to assume that the
This same model is also exercised to predict the opti- maximum in efficiency always occurs at resonance, a closer
mum efficiency for an electrodynamic source that drives a examination of Eqs. (4.19) and (4.20) reveals that a still
thermoacoustic resonator having a resonance frequency that higher efficiency can be obtained for some other value of
is roughly twice the expected in vacuo mechanical resonance frequency or reactance than the one for which p0X M

974 J. Acoust. Soc. Am., Vol. 115, No. 3, March 2004 Robert W. M. Smith: Letters to the Editor

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 142.150.190.39 On: Sat, 20 Dec 2014 06:39:34
XL . In fact, Hunt points out that this higher efficiency is Research Laboratory, the Office of Naval Research, and Ben
only available for transducers that have complex as opposed and Jerrys Homemade for their continued support of ther-
to purely real transduction coefficients. An imaginary com- moacoustic research.
ponent in the transduction coefficient does not appear to be
1
modeled by Li et al., per the parameters in their Table I, or R. S. Wakeland, Use of electrodynamic drivers in thermoacoustic refrig-
erators, J. Acoust. Soc. Am. 1072, 827 832 2000.
elsewhere in the paper. Our measurements of transducers of 2
F. V. Hunt, Electroacoustics: The Analysis of Transduction, and its His-
similar design from the same supplier exhibit a negligible torical Background Acoustical Society of America, 1984.
complex component in the transduction coefficient.7 For 3
T. L. Shearer, Thermoacoustic refrigeration using a linear actuator with-
transducers with real transduction coefficients, any nonzero out pressure and displacement transducers, Masters thesis, Penn State
Dept. of Electrical Engineering May 2002.
combined acousto-mechanical reactance experienced by the 4
T. J. Hofler, Accurate acoustic power measurements with a high-intensity
source reduces the efficiency from what theoretically could driver, J. Acoust. Soc. Am. 832, 777786 1988.
5
have been achieved. S. L Garrett, J. A. Adeff, and T. J. Hofler, Thermoacoustic refrigerator for
space applications, J. Thermophys. Heat Transfer 74, 595599 1993.
6
R. B. Byrnes, Electronics for autonomous measurement and control of a
ACKNOWLEGMENTS thermoacoustic refrigerator in a space environment, DTIC Report No.
AD B141 388 1989.
I would like to thank Steven Garrett for guidance and 7
R. W. Smith, High efficiency two kilowatt acoustic source for a thermoa-
Thomas Gabrielson for helpful discussions and corrections. I coustic refrigerator, Masters thesis, Penn State Dept. of Engineering
would also like to thank the Penn State University Applied Science and Mechanics Dec. 2000.

J. Acoust. Soc. Am., Vol. 115, No. 3, March 2004 Robert W. M. Smith: Letters to the Editor 975

Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 142.150.190.39 On: Sat, 20 Dec 2014 06:39:34

Anda mungkin juga menyukai