Sie sind auf Seite 1von 8

Baltimore, Maryland

NOISE-CON 2004
2004 July 12-14

Measurement of Sound Transmission Loss Using a Modified
Four Microphone Impedance Tube

Andrew R. Barnard
Mohan D. Rao
ME-EM Department
Michigan Tech University
Houghton, MI 49931 USA

1. INTRODUCTION
With the increasing awareness of consumers to product noise combined with stringent noise
regulations, acoustic comfort has become a primary interest in the design and manufacture of mechanical
systems. A popular way to reduce the undesirable or harmful effects of sound is to block the sound
transmission paths with passive acoustic material treatments. These treatments are generally multi-
layered, consisting of some type of dense barrier material in combination with porous, open celled
absorptive materials. The barrier layers can be made of materials from rubber to hard plastic. Typically
the absorptive materials are made of either polyurethane foam or some type of compacted fibrous
material. It is important for manufacturers and engineers to understand how effectively an acoustical
treatment can block incident sound when designing a mechanical system.
Transmission loss is a key quantification of the effectiveness of acoustical treatments for engineering
applications. The sound transmission coefficient, , of an acoustical treatment is a function of frequency
defined as the ratio of sound energy transmitted through the treatment to the amount of sound energy
incident on the source side of the material
1
. Sound transmission loss (STL) is then simply defined as the
transmission coefficient expressed in decibels (dB) according to Equation (1) below:
dB STL |

\
|
=

1
log 10
10
(1)
Most engineers and scientists who deal with these types of noise control treatments perform
experiments to compare the STL effectiveness of different treatment options. The standard test methods
for STL involve using two adjacent rooms with an adjoining transmission path. The treatment under test
is placed between the two rooms in the adjoining transmission path, sound is generated in one room, and
measurements are taken in both the source and receiver room to characterize STL. These methods are
well defined, time tested, and reliable. Unfortunately, implementing these testing methods reliably
requires large and expensive test chambers. In many situations where STL tests are necessary but
infrequent, this cost and space burden is unacceptable. A STL testing procedure that is less costly and
requires less space would be of great interest in this situation.
A method exists that can fit these low cost and space specifications. It utilizes a modified impedance
tube with four microphones. This method measures what is known as the transmission loss matrix of the
acoustical treatment material, from which STL can be extracted. This study, conducted at Michigan Tech
University (MTU), attempts to implement and validate this transfer matrix method by testing four materials
with traditional two-room methods and the modified four microphone impedance tube. Brief theory of the
two-room techniques and extended theory of the modified impedance tube technique is presented. The
strengths and weaknesses of the three two-room methods and the transfer matrix method are described.
The four multi-layered acoustical materials tested in this study were donated to MTU by Blachford,
Inc. Below are descriptions and model numbers of the materials.
Blachford BYY-14E: 5 mm polyurethane foam - 1 mm rubber barrier - 25 mm polyurethane foam
Blachford BPRPR-14I: 19 mm cotton fiber - 1 mm rubber barrier - 25 mm cotton fiber
Blachford BPRPR-14E: 19 mm cotton fiber - 1 mm rubber barrier - 19 mm cotton fiber
Blachford 2-DT/USC3: 5 mm textured rubber barrier - 12 mm dense polyurethane foam
2. TWO-ROOM MEASUREMENT TECHNIQUES

A. Theory
The theory of interest here involves a two-room setup consisting of adjacent reverberation and
anechoic chambers. Methods also exist for two reverberation room configurations but they are not
utilized here.
The first technique utilizes sound intensity to experimentally determine STL. This method has been
standardized by the American Society of Testing and Materials (ASTM) in the standard ASTM E2249
2
. A
broadband sound source is placed in the reverberation, or source, room. The material under test is
secured using an open window between the rooms. The sound intensity incident on the material, I
i
, is
calculated in Equation (2) from the space averaged sound pressure in the source room, p, under the
assumption that the sound field is diffuse. Here, is the density of air and c is the speed of sound in air.
The sound intensity transmitted through the material, I
t
, is then measured in the anechoic chamber, or
receiver room, using a standard two microphone intensity probe. The intensity probe is positioned
perpendicular to the test sample and can be scanned or moved point by point over the material surface to
obtain the averaged transmitted sound intensity. The transmission coefficient is then the ratio of
transmitted sound intensity to incident sound intensity as shown in Equation (3) and the STL can be
computed using Equation (1).

c
p
I
i

=
4
2
(2)
i
t
I
I
= (3)
A second method exists and is defined by the Society of Automotive Engineers (SAE) in the standard
SAE J1400
3
. This method uses the same setup presented above, except sound pressure is used to
calculate STL rather than sound intensity. The noise reduction (NR) is measured using pressure
microphones. NR is simply the difference between the volume averaged sound pressure level (SPL) in
the source room and the material surface averaged SPL in the receiver room. A frequency dependant
correction factor is then subtracted from the NR to directly yield STL. This correction factor is computed
using the mass law for a homogeneous limp mass panel. The theoretical STL of this reference panel is
computed using Equation (4) and the NR is measured with the reference panel positioned between the
chambers. Here, w is the surface weight of reference panel in kg/m
2
and f is the frequency vector in Hz.
The correction factor (CF) is then computed by subtracting the theoretical STL of the reference panel
from the NR measured with the reference panel in place.
dB f w STL 47 log 10 log 20
10 10 theory
+ = (4)
The final method used in this study is defined by ASTM E336 in appendix A2.3
4
. This method
measures what is known as the field transmission loss (FTL). It is intended measure FTL of partitions
that are installed in a working environment. Here the same setup as above was used to measure the
FTL. The volume averaged sound pressure level (SPL) in the source room and the material surface
averaged SPL in the receiver room were measured to yield NR. The FTL is then computed using
Equation (5). It is evident that this is the same equation used to compute the STL in the SAE J1400
technique except that the correction factor has been set to 6 dB for all measured frequencies.
dB NR FTL 6 = (5)

B. Experimental Methods
All three two-room methods presented here were performed in the reverberation room - anechoic
chamber suite at MTU. The suite consists of a reverberation room of volume 48.388 square meters
joined to an anechoic chamber by a 61 cm square opening. The four materials were cut to this dimension
and secured in the opening individually for testing. A Brel & Kjr (B&K) type HP1001 sound source and
B&K type 4205 noise generator were used to excite the source room with broadband white noise. Two
pressure microphones were used in the reverberation room to collect time averaged SPLs at 6 locations
to yield the volume averaged SPL in the source room. A two microphone sound intensity probe, G.R.A.S.
50 AI-B, was used in the receiver room to record both surface averaged transmitted sound intensity and
SPL. The surface averaged measurements were taken at a distance of 34 cm from the material sample
in the receiving room, based on the standard
4
, and were averaged using a probe scanning technique. A
01dB Symphonie data acquisition was used to collect data for all the two-room tests.
Based on the reverberation room volume and room absorption, the valid frequency range lower limit
of the tests were determined to be about 300 Hz. All measurements were therefore made in third octave
bands from 315 Hz to 10 kHz. An aluminum plate of thickness 3.175 mm, and density 2.8 Mg/m
3
was
used to compute the correction factor for the SAE J1400 method.

3. MODIFIED FOUR MICROPHONE IMPEDANCE TUBE MEASUREMENT TECHNIQUE

A. Theory
Many new STL measurement variations involving a modified standing wave tube have been
developed in recent years. These techniques were originally developed for the measurement of STL of
muffler systems
5
,

but have been adapted to measure STL of acoustic materials. The technique utilized in
this study involves a four microphone technique and an impedance tube described by Philippe Callec
Solutions
6
. Similar techniques have been used by Song and Bolton
7
and Tao and Seybert
5
. A standard
two microphone impedance tube is modified by placing two additional microphone holders on the receiver
side of an acoustic treatment test sample, as seen in Figure (1). Sound pressure is recorded at all four
microphone positions. Through the following algorithm, STL can be computed by solution of the
transmission loss matrix.
The basis for the algorithm lies in the definition of the transmission loss matrix which is unique for all
acoustic materials. As the sample rests in the tube, there are four acoustic waves relating to the material
sample, as shown in Figure (1). The forward traveling wave is defined by its incident, A
1
, and transmitted,
B
2
, pieces. The backward traveling wave is also defined by its incident, B
1
, and transmitted, A
2
, pieces.
The transmission loss matrix is then defined as the matrix relating the forward and backward traveling
acoustic waves as shown in Equation (6).

)
`

=
)
`

) (
) (
) ( ) (
) ( ) (
) (
) (
2
2
1
1
f B
f A
f f
f f
f B
f A


(6)
The coefficient of interest here is the first term of the transmission loss matrix, (f). This term is the
same sound transmission loss coefficient defied above as . So by characterizing (f) = , the STL of the
material is simply defined as Equation (1). The four values A
1
, A
2
, B
1
, and B
2
can be used to solve for the
sound transmission loss coefficient. Equations (7) through (10) define A
1
, A
2
, B
1
, and B
2
in terms of the
sound pressure readings, P
i
, and distances of the microphones, X
i
, as shown in Figure (1). In these
equations j is the complex representation of 1 and k is the wave number.

2
1
) sin(
) ( ) (
2
) (
1
2 1
1
X jk
X jk
e
X k
e f P f P j
f A


= (7)
3
4
) sin(
) ( ) (
2
) (
4
3 4
2
X jk
X jk
e
X k
e f P f P j
f A

= (8)

2
1
) sin(
) ( ) (
2
) (
1
2 1
1
X jk
X jk
e
X k
e f P f P j
f B

= (9)
3
4
) sin(
) ( ) (
2
) (
4
3 4
2
X jk
X jk
e
X k
e f P f P j
f B

= (10)
From an experimental viewpoint, the sound pressure at specific points in space and time are related
to pressure measurements at other points in space and time through phase relationships. These
relationships are typically gathered through the acquisition of complex ensemble averaged cross spectra,
xy
G . A measure of strict sound pressure magnitude at a point is acquired as an ensemble averaged
autopower spectra,
xx
G . These two measurements can be utilized in this case to modify Equations (7)
through (10) to yield Equations (11) though (14).
2
1
) sin(
) ( ) (
2
) ( ) (
1
12 11 *
1 1
X jk
X jk
e
X k
e f G f G j
f G f A


= (11)
3
4
) sin(
) ( ) (
2
) ( ) (
4
13 14 *
1 2
X jk
X jk
e
X k
e f G f G j
f G f A

= (12)
2
1
) sin(
) ( ) (
2
) ( ) (
1
12 11 *
1 1
X jk
X jk
e
X k
e f G f G j
f G f B

= (13)
3
4
) sin(
) ( ) (
2
) ( ) (
4
13 14 *
1 2
X jk
X jk
e
X k
e f G f G j
f G f B

= (14)
By acquiring complex cross spectra from microphones 2, 3, and 4 using microphone 1 as a reference,
and the autopower spectra of microphone 1 with the impedance tube in two different end conditions, the
sound transmission coefficient can be solved from Equation (6) using the determinant method as in
Equation (15). The two end conditions must be very different and are generally open ended, or anechoic,
represented below with a subscripted O, and closed ended, or reverberant, represented below with a
subscripted C.
( )( ) ( )( )
( )( ) ( )( ) ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (
) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (
) (
*
1 2
*
1 2
*
1 2
*
1 2
*
1 2
*
1 1
*
1 2
*
1 1
f G B f G A f G B f G A
f G B f G A f G B f G A
O O C C C C O O
O O C C C C O O


= (15)
It should be noted that since the determinant method of solving a system of equations, or Cramer's
Rule, is being used in this algorithm, that the algorithm will not yield a solution if the transmission loss
matrix is symmetric. This occurs when the material sample under test is symmetric through its thickness
6
.

B. Experimental Methods
A B&K type 4206 impedance tube was modified for this experiment. Two separate diameter tubes
were used to measure STL, 100 mm and 29 mm. The frequency ranges on the two tubes were assumed
to be 100 Hz to 1.6 kHz for the large tube and 500 Hz to 6.4 kHz for the small tube per the manufactures
impedance testing specifications. The microphone positions for the 100 mm tube were as follows: X
1
=
50 mm, X
2
= 100 mm, X
3
= 304 mm minus the thickness of the material sample, and X
4
= 50 mm.
The microphone positions for the 29 mm tube were as follows: X
1
= 20 mm, X
2
= 22 mm, X
3
= 230 mm
minus the thickness of the material sample, and X
4
= 20 mm. An open ended tube condition and a hard
closed ended tube condition were used for both sized tubes. Two " condenser microphones were used
to measure the one autopower spectrum and the three cross spectra in each end condition. A Spectral
Dynamics Siglab data acquisition unit was used to collect the ensemble averaged auto and cross spectra.
Custom designed post processing software was developed to process the auto and cross spectra,
computing STL for each sample. The software was implemented in the Mathwork's MATLAB
environment. The features of the software include:
Imports spectra from Siglab acquisition system.
Computes STL curves from imported spectra.
Combines STL curves from large and small tubes for each material if appropriate.
Averages many STL tests of the same material in either tube.
The four Blachford materials were cut to the inner dimensions of both tubes using a water jet cutting
machine to ensure accurate dimensional tolerances. It was speculated that very small differences in the
diameters of the samples and the nominal tubes would cause large variations in STL. To test this theory,
samples that were oversized by 2 mm on the diameter were tested along with the nominally sized
samples. A minimum of 4 samples for the nominal boundary condition and two samples for the oversized
boundary condition were averaged to obtain results.

4. COMPENSATION FOR BOUNDARY CONDITION INDUCED STL SHIFTING
The modified four microphone technique was first tested on all four materials, using both sized
impedance tubes, under both nominal and oversized boundary conditions. The trends for all the
materials were similar and Figure (2) shows the results for the BYY-14E material. It is evident that the
STL shifts upwards across all frequencies as the boundary conditions are pressure loaded. This is not
unexpected since the pressure loaded boundary condition causes the boundary pores of the absorptive
layers to close and also causes the barrier layers to stiffen. As expected, this phenomenon is much more
pronounced in the small tube due to greater pressure on the samples. These results solidify the
assumption that boundary condition pressure loading causes large errors in STL computation.
The other evident feature in Figure (2) is that small tube STL and the large tube STL do on align for
the nominal boundary condition data. This was unexpected. The diameters of the small tube samples
were measured and found to be between 0.5 mm and 1.5 mm undersized for all the materials. This
undersizing created air gaps between the tube wall and the specimen boundary, accounting for the
downward shift in STL as compared to the large tube data. The effects of undersizing can be quantified
using Equation (16) where A
gap
is the surface area of the gap, A is the total surface area of the tube cross
section,
gap
is the sound transmission coefficient of the gap fluid (air in this case), and
actual
is the true
sound transmission coefficient of the material
1
.
(
(

|
|

\
|
+ = 1 1 log 10
10
actual
gap gap
actual measured
A
A
STL STL

(16)
Under the assumption that the large tube samples were cut to exactly the correct dimension, the
actual STL of the material in the third octave bands between 500 Hz and 1.6 kHz was set to the STL
measured over that frequency range in the large tube. The average difference between the actual STL
and measured STL over these third octave bands was then used to shift the small tube STL upward.
Using this shift, the gap area, A
gap
, and consequently the area of the material sample itself could be
calculated in each third octave band. This calculated area was then compared to the actual measured
area of the small tube samples. Figure (3) displays the percent error in the diameter computation of all
the materials based on the undersized boundary condition STL shift in the small tube. It was assumed
that if the computed diameter was within five percent error of the actual diameter, that the shift could be
considered valid and the large and small tube STL curves could be combined in the valid frequency
regions.
To test this hypothesis, another small tube sample was cut using the water jet cutter for the BYY-14E
material. This material was deemed to be the material most likely to be able to be cut accurately due to
its stiffness and porosity characteristics. Several samples were cut until a nominal diameter of 29 mm
was attained. This truly nominal sample was then tested in the small impedance tube and a nominal
large tube sample was again tested. This time, the frequency ranges used in both tests were 100 Hz to
10 kHz. After the STL had been computed for both tubes, the results were averaged in third octave
bands from 500 Hz to 6.3 kHz.
Figure (4) displays the original results from the large tube samples up to 1.6 kHz, the original shifted
results from the undersized small tube samples, and the averaged small and large tube results for the
truly nominal samples. It is evident that these curves align very closely across all third octave bands
tested. This validated that the shifting and combining of the undersized small tube samples with the
nominally sized large tube samples produced accurate results. The final reported results for the four
microphone method for all the materials is the large tube STL combined with the shifted undersized small
tube STL over the third octave bands where the percent error in calculated versus measured diameter
was small.

5. COMPARISON OF TECHNIQUES
Figures (5) through (8) display the STL results for each material tested using the four techniques
presented above. The two-room STL data for all of the acoustic treatments follow the same trends. The
sound intensity technique and FTL technique yield curves of the same shapes, with the sound intensity
technique showing increased STL. The sound intensity technique will only take into account the sound
energy traveling perpendicular to the test sample in the receiver room. The FTL technique, on the other
hand, measures SPL in the receiver room which could conceivably come from paths other than the
material sample. These secondary sound paths are known as flanking transmission paths and can
adversely affect results
2-4
. Also the underlying assumptions of these two techniques are slightly different
so the results are not expected to be equal
4
. The SAE J1400 technique appears to approximate the FTL
technique at low frequency and the sound intensity technique at higher frequency for all of the materials.
This is due to the difference in the correction factor between the SAE J1400 and the FTL methods. The
frequency dependant correction factor computed with the SAE J1400 method should account for flanking
transmission paths between the chambers as well as other frequency dependant measurement errors. It
appears that the flanking transmission paths between the chambers have a greater effect at high
frequencies, above 3 kHz. Again, this method is based on slightly different assumptions than the other
two-room methods, so the STL measurement is not expected to be exactly the same.
The four microphone modified impedance tube method most closely approximates the FTL two-room
method. The four microphone technique gave the best results in the tests of the Blachford BYY-14E and
2-DT/USC3 materials. The Blachford BPRPR-14E and BPRPR-14E materials yielded less accurate STL
results. This was expected since the BPRPR-14E material was symmetric about its thickness and the
BPRPR-14I material was nearly symmetric about its thickness. As discussed in section 3A, this
symmetry will cause the transmission loss matrix to be symmetric, resulting in numerical linear algebra
errors during the application of the Cramer's Rule solution technique.

6. CONCLUSIONS AND RECOMMENDATIONS
A four microphone modified impedance tube method for the measurement of STL was implemented
at MTU using custom designed post processing software. Three two-room STL measurement methods
were compared with the modified impedance tube STL measurement for verification purposes. The four
microphone technique most closely approximates the FTL technique presented in ASTM E336
4
. The
measurement of STL using the modified impedance tube appears to yield accurate results for comparison
of materials that are non-symmetric through their thickness. Materials that are symmetric or close to
symmetric through their thickness do not produce reliable results using this method.
Boundary conditions within the modified impedance tube can cause wild fluctuation in the expected
STL results. Undersized samples, or air gap boundary conditions, will cause reduced STL. This
reduction can be predicted and accounted for using Equation (16) as shown in section 4. Oversized
samples, or pressure loaded boundary conditions, will produce increased STL measurements using this
method. Both of these boundary condition effects are magnified as the diameter of the impedance tube is
decreased.
Several recommendations for future work on the four microphone modified impedance tube method
are made. The undersized boundary condition shift algorithm should be tested using several different
non-symmetric materials and multiple diameter impedance tubes. Formulas for the prediction the STL
shift induced by a pressure loaded boundary condition should be developed. This most likely would
involve the inclusion of the material properties such as porosity, tortuosity, resistivity, and bulk modulus
into some type of finite or boundary element solution.
A more stable linear system solution algorithm could be developed so that this technique could be
used with symmetric samples. An iterative linear solver would be recommended based on either the
Jacobi iteration or the Gauss-Seidel iteration. This might eliminate the symmetry error induced using the
current Cramer's Rule solution by avoiding division by the determinant of a symmetric matrix.
A phase calibration of the microphones may provide more accurate results. A phase calibration
similar to that used in a two-microphone acoustic impedance test, ASTM E 1050-98
8
, would most likely be
adequate. Frequency dependant phase calibration functions could be measured and used to calibrate
the recorded auto and cross spectra. Any phase calibration error is thought to be small here since the
microphones were manufacturer phase calibrated.
Finally, the elimination of flanking transmission paths between the two chambers used in the two-
room tests could produce better verification of the four microphone technique.

Acknowledgement

These authors would like to extend special thanks to the John Deere Technical Center, in particular
Mr. Loren DeVries, for their support of this work and guidance throughout the initial discovery process.
We would also like to extend thanks to Blachford, Inc., for the donation of the four materials used in this
study.

References
1. H.A. Evensen and M.D. Rao, "Sound Fields & Room Acoustics", Section 3 in Supplemental Notes to
MEEM 4704 Acoustics and Noise Control, edited by H.A. Evensen and M.D. Rao, (Michigan Tech
University, Houghton MI, 2002).
2. Standard Test Method for Laboratory Measurement of Airborne Transmission Loss of Building
Partitions and Elements Using Sound Intensity, American Standard ASTM E 2249 - 02: 2003
(American Society for Testing and Materials 2003).
3. Laboratory Measurement of the Airborne Sound Barrier Performance of Automotive Materials and
Assemblies, International Standard SAE J1400 - 199005: 1990 (Society of Automotive Engineers
1990)
4. Standard Recommended Practice for Measurement of Airborne Sound Insulation in Buildings,
American Standard ASTM E 336 - 71: 1971 (American Society for Testing and Materials 1971).
5. Z. Tao and A.F. Seybert, "A Review of Current Techniques for Measuring Muffler Transmission Loss",
Proc Soc Automotive Eng Noise and Vibration Conf, (2001).
6. Phillipe Callec Solutions, "Customized Sound & Vibration Software: Acoustical Loss", A pamphlet,
France, (2003).
7. B.H. Song and J.S. Bolton, "Enhancement of the Barrier Performance of Porous Linings by Using
Internal Constraints", Noise Control Eng. J., 51(1), pp.16-35 (2003).
8. Standard Test Method for Impedance and Absorption of Acoustical Materials Using a Tube, Two
Microphones and a Digital Analysis System, American Standard ASTM E 1050 - 98: 2003 (American
Society for Testing and Materials 2003).

Figure (1) - Schematic of four microphone modified impedance tube
10
2
10
3
10
4
0
5
10
15
20
25
30
35
40
45
50
Third Octave Center Frequency (Hz)
S
T
L

(
d
B
)
Large Tube STL (nominal)
Small Tube STL (nominal)
Large Tube STL (oversized)
Small Tube STL (oversized)

Figure (2) - STL measured in the modified impedance tube (nominal and oversized boundary conditions)
10
1
10
2
10
3
10
4
0
10
20
30
40
50
60
Third Octave Center Frequency (Hz)
E
r
r
o
r

(
%
)
BYY-14E
BPRPR-14I
BPRPR-14E
2-DT/USC3

Figure (3) - Percent error between calculated and actual diameters of 29 mm undersized samples
10
2
10
3
10
4
0
5
10
15
20
25
30
35
Third Octave Center Frequency (Hz)
S
T
L

(
d
B
)
STL, Large Tube
STL, Small Tube (Original Undersized)
STL, Small Tube (Shifted)
STL, Average Large Tube (Ext. Freq) and Small Tube (recut)

Figure (4) - Verification of 29 mm modified impedance tube STL shifting algorithm for BYY-14E
10
2
10
3
0
10
20
30
40
50
60
Frequency (Hz)
T
L

(
d
B
)
STL, Averaged Large --Small (shifted) Tubes
STL From intensity (ASTM E2249)
FTL from pressure (ASTM E336)
STL (SEA J1400)

Figure (5) - STL measurements from all four methods for Blachford BYY-14E material
10
2
10
3
0
10
20
30
40
50
60
Frequency (Hz)
T
L

(
d
B
)
STL, Averaged Large --Small (shifted) Tubes
STL From intensity (ASTM E2249)
FTL from pressure (ASTM E336)
STL (SEA J1400)

Figure (6) - STL measurements from all four methods for Blachford BPRPR-14I material
10
2
10
3
0
10
20
30
40
50
60
Frequency (Hz)
T
L

(
d
B
)
STL, Averaged Large --Small (shifted) Tubes
STL From intensity (ASTM E2249)
FTL from pressure (ASTM E336)
STL (SEA J1400)

Figure (7) - STL measurements from all four methods for Blachford BPRPR-14E material
10
2
10
3
0
10
20
30
40
50
60
Frequency (Hz)
T
L

(
d
B
)
STL, Averaged Large --Small (shifted) Tubes
STL From intensity (ASTM E2249)
FTL from pressure (ASTM E336)
STL (SEA J1400)

Figure (8) - STL measurements from all four methods for Blachford 2-DT/USC3 material

Das könnte Ihnen auch gefallen