Beruflich Dokumente
Kultur Dokumente
ACOUSTICAL
Volume
JOURNAL
OF THE
SOCIETY
30
OF
Number 11
AMERICA
NOVEMBER.
1958
Armour Research Foundation, Riverbank Acoustical Laboratories, Geneva, Illinois (ReceivedJuly 28, 1958)
A new soundtransmission lossmeasuringfacility has been in use at Riverbank AcousticalLaboratories for more than a year now. This paper describes two reverberation the rooms,installationof the test panel, instrumentation,and measurement technique.The several checksmade for measurementaccuracyand repeatability are discussed. Test resultsare given comparingthe old method of test with the new, and resultsare alsogiven for a folding doormeasured four testinglaboratoriesin the United Statesand Canada. at
I.
INTRODUCTION
HE is definedas the ratio of incident or by 14 ft wideby "11.5 high,and hasa volumeof sound loss of (TL) ft partition transmission awall to
sound sourceand the larger room receivesthe transmitted sound.The new room is rectangular, ft long 20
are transmitted sound energy, with a reverberantsound about 3200 cu ft. The insidesurfaces hard plaster and there are severalprotuberances into field in both sourceand receiving rooms. Becauseof and concrete, the necessity reverberant for sound fields,the measure- the room which help to increasesound diffusion.The ment of sound transmission requires loss special building room absorptionbelow 1000 cps is 30 to 50 sabins. facilities which are both large and costly. Riverbank The larger reverberationroom is the one regularly and is 27 ft Acoustical Laboratoriescompleted December,1956, used for soundabsorptionmeasurements in long by 19.7 ft wide by 19.1 ft high with a volume of a second reverberation room to be used for sound transmission losstesting. This new facility now meets about 10 100 cu ft. To increasesound diffusion, this explicitlythe standards both the AmericanSociety room contains two 6-by-12-ft steel reflector vanes of for Testing Materials and the American Standards which rotate at a rate of one rpm during the test, and Association 2 for airborne sound transmission loss a number of 12-in.-diam pillars, 4, 6, and 8 ft high,
measurements.
standing the floornear the roomwalls.The absorpon tion of this room below 1000 cpsis 100 to 110 sabins.
This room has 18-in.-thick walls and a 22-in.-thick
ceiling of solid masonry,and is structurally isolated from the sourceroom (and the rest of the building) Figure1 shows plan view of the two reverberation a 20-in. air rooms.The new room, shownon the left, containsthe with a separatefoundationand surrounding
space.
ASTM Designation E90-55,"Standard recommended practice The test panel is positionedin the receivingroom for laboratorymeasurement airbornesoundtransmission of loss of building floorsand walls," American StandardsAssociation, wall, and is installedas shownin Fig. 2. The panel is mended practicefor laboratorymeasurement Airbornesound ards Association,New York. of
999
Copyright 1958 by the AcousticalSociety of America.
New York. This standard is presentlyunder revision. 2 ASA DesignationZ24.19-1957, "American standard recom-
Downloaded 06 Apr 2012 to 58.172.48.46. Redistribution subject to ASA license or copyright; see http://asadl.org/journals/doc/ASALIB-home/info/terms.jsp
1000
I0 FT
R.
L.
RICHARDS
SOURCE ROOM
'i:}
RECEV,NG ROOM
NSULATION \
2" FELT
iI
--TEST
PANEL
", I'
type 633A microphones used, one in each room, are two identicalsystems eachconsisting a voltageamplifier, of attenuator, variableband passfilter (Allison type 2-B), and one channelof a twin-channel graphic levelrecorder (Sound Apparatus Company modelR-Z). A special levelequalizing circuit establishes and maintainsa constantsoundpressure level in the receiving roomby adjusting levelin the the source room.This insures known,constantsignal-toa
noise ratio on the low level side and facilitates reduction
of the data.
3. Technique
,r/l/'/////////////////////////
OFFICES
at a rate of 5 timesper sec.The sweep slowenough is so that the soundpressurelevel in each room is the usually constructed inside a wood frame with inside sameas it wouldbe for steadystate at all frequencies. dimensions 66.5 in. wide by 81 in. high, and is trans- During the sweepthe sound pressurelevels in each recorded.A typical chart ported on a specialcart to the openingwhere it is room are simultaneously rolled up againsta permanent woodframe with inside dimensions sameas thoseof the test panel. Sealing the stripsfacedon two sides with sponge rubber are then placedaroundthe othersideof the panelframe,and a third removable frame is clamped tight against the panel frame. This installationtechnique allowsrapid change panelsand provides solidcoupling of a between the test panel and the surrounding wall, yet is airtight becauseof the spongerubber seals.Also, particular care is taken to sealaroundthe test panel insideof its own frame. Caulking compound very effectivefor is sealing small cracks and is generally used for this purpose. Heavierwalls,suchasbrickor concrete block, Fro. 3. Block diagram of instrumentationfor sound must be built directly in the test opening:A special transmission loss measurements. adapterpanelis usedfor the installation doorsand of panelssmaller than the standardsize. This adapter from the twin-channelgraphic level recorderis shown panelis discussed detail in a later section. in in Fig. 4. The top half is a graph of the soundpressure
2. Instrumentation
Fro. 1. Plan view of the Riverbank reverberation rooms used for sound transmission measurements.
For the test, the frequency the source sweptin of is 7.5 min from 100to 5000cps,andis warbled 4-20 cps
A diagram of the instrumentationused for the the receiving room. The heavy line is a smoothed measurements shown Fig. 3. Two WesternElectric averagedrawn by eye and, in the caseof the receiving is in room side, is straight becauseof the level equalizer. RECEIVING ROOM SOURCE ROOM The rapid fluctuations about the average line are typical of measurements made in a diffusesoundfield.
SEALING FRAME
( REMOVABLE
)
TEST PANEL
BEARING PLATE
If the warbler
SEAL
""-'SPONGE
PIPE
RUBBER
ROLLER
. - '
:'.v.
MASONRY
' '
.o
fluctuations would be lessrapid and much greater in amplitude making a smoothedaveragemore difficult to obtain.After two runsare made,the pickupsystems are completelyinterchanged and two more runs made. The soundreduction,or difference soundpressure in level betweenthe two rooms,is obtainedby averaging
the differences between the smoothed lines of each of
,.
..
the four charts.By interchanging pickupsystems any differences response in between them are eliminated. The soundreductioncurveis usedin the preparation
Downloaded 06 Apr 2012 to 58.172.48.46. Redistribution subject to ASA license or copyright; see http://asadl.org/journals/doc/ASALIB-home/info/terms.jsp
RIVERBANK
TRANSMISSION
LOSS
FACILITY
1001
according Buckingham'sequation to a
(1)
nessof the room walls in retarding the transmission of soundbetweenroomsis to build in the test opening
the same structure as that of the walls. This was not
where L=sound pressurelevel in source room, L. done becauseof the immensity of the walls, but a was =sound pressurelevel in receivingroom, S=area of measurement madeof the TL throughtwo plaster test panel, fff', and A =absorption of receivingroom, partitions of identical construction(2-by-4-in. wood 16 -in. gypsum boardlath, sabins. Since the absorption of the receiving room studsspaced in. on-centers, remains virtually constant the year around and is and -in. plaster each face), one installed in the checked repeatedly sound for absorption measurements, source room wall and the second in the receiving this measurement is not made for each TL test unless room wall (standardposition),which providedstructhe test panel itself adds significantlyto the room tural isolationbetween the two partitions. The partiabsorption.The test report includesa descriptionof tions were separatedby a 12-in. air space.The results are shownin Fig. 5. The upper the test panel, the measurementtechnique, and a of this measurement graphof the TL vsfrequency. Also, the TL valuesare curve is the TL of both partitions; the lower curve is tabulatedat eachof the nine standardfrequenciesand the TL of one of the partitions tested alone, and is ' the arithmeticaveragegiven.
8O
III.
CHECKS
ON AND
MEASUREMENT REPEATABILITY
ACCURACY
DOUBLE ARTITIO'N
loss
60
are not available,it is impossible checkdirectly the to accuracyof the system. We do know, however, of
certain conditions which must be fulfilled if the measurement is to be correct. Checks on how well these conditions are met are discussed below.
5o
SINGLE
PARTITION
3O
1. Flanking Paths Specialcare is taken to seethat no significantpaths of flanking sound transmissionexist. How the test panel is sealedinto the test openinghas already been discussed. suggested One method to test the effective.'
1:>5 i?
:>50 350
oo
700
Iooo
:>ooo
4000
FREQUENCY, C/S
Fro. 5. Sound transmission loss of singleand double plaster a E. Buckingham, "Theory and interpretation of experiments partitions. For the double partition test, the partitions were on the transmissionof sound through partition walls," Sci. structurallyisolatedwith a 12-in. air spacebetween.TL values greaterthan thoseof the upper curve can be measured. PapersNatl. Bur. Standards 193 (1925), S506. 20,
Downloaded 06 Apr 2012 to 58.172.48.46. Redistribution subject to ASA license or copyright; see http://asadl.org/journals/doc/ASALIB-home/info/terms.jsp
1002
R.
L.
RICHARDS
missioncoefficientof adapter panel; and r,.=sound transmission coefficient test door. After solvingfor of
the sound transmission coefficient of the door
FIXED FRAME
r,.= (S/S,)(r-r)+r,
(3)
2" x 12"
WOOD STOP
LUMBER FRAME
the true TL of the dooris obtainedby multiplyingthe logarithim of 1/r by 10. This adjustmentbecomes
large and uncertain if the measuredTL of the door is within about3 db of that of the adapter.Adapterwalls of masonryhave beenusedfor measuring doorswith a very high TL.
2. Diffusion
/4" BOLT
6 LAYERS OF .3/8"
GYPSUM BOARD SPONGE RUBBER,
and receiving roomsoundfieldsare reasonably diffuse. The special means used to increasesound diffusion
EST DOOR
dicates sound diffusion is the effect of microphone position on test results. A seriesof sound reduction
Fxo. 6. Crosssectionshowingconstruction one end of adapter of panel. This adapter is usedfor installingdoorssmallerin sizethan the regulartest opening.
o z
50 40
$0
20
u
z
55 db in the low-frequency range and 80 db in the high range. A special adapterpanelis usedfor the installationof 36-by-80-in.doorsand panels,which are smallerthan the standardsize opening.This adapter panel consists of a frame of 2-by-12-in.lumber divided vertically into 3 sections. The middlesection 36 in. wide for receiving is the test door, and the two end sections,each 13 in. wide, are filled with gypsumboard sheetsas shownin Fig. 6. If a doorwith a largeTL is testedin the opening, the sound energy transmitted through the adapter panel itself becomes important. The TL through the adapter panel was determinedby filling the opening solidlywith brick (12 in. thick). The TL of the brick was krown from previous tests and was' negligible comparedto the adapter at most frequencies. The TL of the adapter panel, measuredin this manner, is shownin Fig. 7. If the measured TL of a door being testedin the adapterpanelcomes within about 12 db to of the TL of the adapterpanel,then the true TL of the door is found from the relationshipthat the total sound energy transmitted betweenroomsis the sum of each contributingelement.In this case,
10
0
125 175
250
.350 500
7001000
2000
4000
FREQUENCY, C/S
curves were run on a test panel at several different microphone positions both rooms,with the reflector in vanes in the receivingroom stationary. These curves were all drawn on one graph and the mean line drawn throughthe enclosed area. Deviationsfrom the mean wereasmuchas 2.5 db at a few of the lowerfrequencies, but usuallywerelessthan 1.5 db. In fact, deviations of about the same size are noted when comparingruns made with the microphone positionunchanged in (as
the normal test procedure)and the reflectorvanes started at differentpositions the beginning each at of run. All microphone positions were chosen least4 ft at from any room surfaceand at a distancefrom the soundsource where the direct soundpressure level was
at least 10 db below that of the reverberant level.
3. Reversal of Sound Flow
Sr= Slrl-Jr-S,.r,.,
(2)
Measurements
test
whereS= total area, Sq-S,.; r= measured soundtrans- panels, first in the normalway, then by movingthe mission coefficientof door and panel; S=area of soundsourceinto the large room and using the small adapterpanel; S2= area of test door; r = soundtrans- room to receive the transmitted sound. The results of
Downloaded 06 Apr 2012 to 58.172.48.46. Redistribution subject to ASA license or copyright; see http://asadl.org/journals/doc/ASALIB-home/info/terms.jsp
RIVERBANK
TRANSMISSION
LOSS
FACILITY
1003
thesechecksare shownin Fig. 8. The top graph is the TL of a 2-in.-thickpanelwith -}-in. plywoodfacesand glasswool packed in the core space.The middle graph represents wood stud panel with plywood facesand a the lower graph a wood stud, gypsumboard lath, and plaster partition. These curvesare smoothedaverages of the original TL curves,so minor variations are not shown.Except for valuesat a few isolatedfrequencies, the curvesare within the systemmeasurement error. An additionalmeasurement the wood-stud, of plywoodfaced panel was made after the absorption of the receiving room was increasedto about 200 sabins,
almost double the normal
50 40 30
Frequency,
cps
1000
2000 4000 av
-0.5
- 1.5 - 3.5 -0.4
+0.5
-- 1.o - 3.0 -0.6
amount.
/,'
2" PANEL
_ /E
error. The tendencyfor the TL to increasewith decreasing frequency below 250 cps is due to the inefficiency the openings soundradiators.Conof as siderably higher losses noted thelowfrequencies were at with openings smallerthan 18 sq ft.
5. Repeatability
J 40
z
30
20
< IO
,,-., 50
z
II
40
30 20
,../ .....
1.5 2 3 4
I PLASTER
5 6 8 i I000
WOOD} STUD
PARTITION
i.5 2 3
quencies because wavelength the approaches room the dimensions. is possible It that somevariationis due to changes the panel with time and installation in differences, althoughtheseeffectsare believedto be
small becausethe panel had aged for over 3 years
before the first measurement and the installation was
IO
I I00
FREQUENCY,
C/S
accuratelyreproduced.
IV. CORRELATION OF TEST RESULTS
NORMAL
TEST
SOURCE ROOMS
FzG. 8. Effect
The facilityfor sound transmission measurements loss at Riverbank use in from1949until July,1956, consisted
II. Repeatabilitycheckconsisting 5 separate of measurementson a woodstud panelwith plywoodfaces.
Sound Transmission Loss, db
Frequency
cps
Test 1
Test2 a Test3
Test4
Test5
Range
is an opening. Measurements were made with two different-sizedopeningsand the results are given in
Table I. In cases where the sound reduction between
roomsis 15 db or less,Eq. (1) takesthe form TL= 10 log(R-- 1)+ 10 logS- 10 logA, (4)
4000 av
37.0 26.5
37.5 25.9
19.0 20.0 19.5 21.5 26.0 28.5 30.0 32.0 37.5 26.0
37.0 26.0
36.0 25.3
Part 3).
1.5 1.2
Downloaded 06 Apr 2012 to 58.172.48.46. Redistribution subject to ASA license or copyright; see http://asadl.org/journals/doc/ASALIB-home/info/terms.jsp
1004:
50
R.
L.
RICHARDS
40
30
I , ,...
,
20
while too high on an absolute basis,did put panelsin their approximate rank order and can be used with reasonable accuracy(with a 5 or 6 db subtraction) in situations wherea precise knowledge TL vsfrequency of is not required.
2. Correlation with Other Laboratories
/y
o
:'
o
0
40
30
'
:o
// / /
'-
At this time there is very little informationavailable correlatingRiverbank test results with those of other laboratories.In fact, the only material which has recentlybeenmeasured eachof the principaltesting at laboratoriesis an accordian-typefolding door. The
results of these measurements
Txm:. III.
III.
Io
o
n 40
30
2.0
//
1.5 2.
:3 4 5 6
AVERAGE
DIFFERENCE 6.DB
test laboratories.A different folding door was tested at each laboratory, although all were standard No. 228 Hufcor doors, described follows: full accordiandoors,with 55 oz per sq yd as 5-ply, laminated covers eachsideof enclosed verticalpantograph frame; sweepstrips top and bottom of one side to provide continuous contact with the floor and ceiling. These tests were sponsored the Hough ManufacturingCorporation, by Janesville,
Wisconsin.
Sound Transmission Loss, db
IO
IOO
I I
8 I
IOOO
1.5
Frequency
cps
Lab. As
Lab. Bb
Lab. Co
Lab. D d
FREQUENCY,
C/S
NEW
..... OLD
TEST
TEST
FACILITY
FACILITY
2000
4000 av
Lab. A: Riverbank
25.0
28.5 16
26.0
28.0 16
29.0
35.0 20
27.0
32.0 20
of a reverberantsource roomand a special nonreverberant receivingroom with certain known characteristics. Inherent in this measurement technique were several assumptions about the receivingroom soundfield not susceptible experimentalproof, and the difficulties to in obtaining an accuratemeasurement TL by this of methodwereneverfully resolved. It wasnecessary destroythe old measuring to facility in buildingthe new one, and unfortunatelyonly a few test panels could be kept for correlation purposes. Enough informationhas been gatheredon similar constructions now, however, to give a fairly reliable
indication of the difference in results between the old
b Lab. B: National Research Council, Ottawa 2, Canada, February, 1958. * Lab. C: Geiger and Hamme Laboratories, 3250 East Morgan Road, Ann Arbor, Michigan, June 1957. (Reported as sound reduction since no
adjustment is made for 10 logs and 10 logA.) d Lab. D: National Bureau of Standards, Washington 25, D.C., October,
1957.
and the new. In Fig. 9 are shownthe resultson three panels which were measuredwith the old facility and then remeasured after the change.It can be seenthat the differenceis a function of frequency and is not consistent differentpanels,so correction the old for of measurementscan be applied only to the average sound transmission loss value. Average differences of all the panels whichwereretested, and thosenew panels which were similar in constructionto older panels, ranged from 4 db to nearly 10 db, but most were between 5 and 6 db. This meansthat the old results,
It is recognized that this type of construction not an is ideal choice make sucha comparison to because the of difficulty in obtaining a uniform sealingcondition. Also,a differentdoorwassupplied eachlaboratory, to although eachdoorwasa regularproduction modelof identicaldesign, and eachinstallation was supervised by a representative the doormanufacturer. of
The author is deeply indebtedto his colleagues at Armour Research Foundation, Mr. Ralph Huntley,
Supervisor RiverbankAcoustical of Laboratories, and Mr. FranklinTyzzer, Senior Physicist, their aid in for preparing materialfor this paper. the
Downloaded 06 Apr 2012 to 58.172.48.46. Redistribution subject to ASA license or copyright; see http://asadl.org/journals/doc/ASALIB-home/info/terms.jsp