Room Acoustics@DeKalb High School

Results of a project during the summer of 2003
 
 
Sound Samples: 10 seconds in duration
125 Hz 250 Hz 500 Hz 1000 Hz 2000 Hz
4000 Hz Brown Noise Pink Noise White Noise Silence



Sound Absorption Coefficients



Reverberation Time (RT or T60)

For many years I have discussed the concept of reverberation time with my high school physics classes. Although I have played them recordings of similar sounds made in rooms with different reverberation times and showed them a typical decay curves, I have never been able to measure the T60. These activities will allow a group of students to measure the T60 for a room by two methods. The first method is the Sabine method of absorption which involves only the size of the room and the absorbing materials. The second method is actually two similar ways of measuring the sound decay in the room, the Impulse method and the Tone Cut-Off method. Not only can these methods be used by students to calculate the T60 but they can see how well the different methods compare and they can make changes to the room and predict the effect it will have on the T60.

Sabine Method of Absorption

The reverberation time of a room is the time needed for the sound level in the room to decrease by 60 dB, hence the name T60. Wallace Clement Sabine developed a method to relate the T60 to the size of the room and the amount and type of absorbing material in the room.

T60=(0.161s/m)(V)/(S1a1+S2a2+S3a3....)

V - volume of the room in m3.
S - surface area of each of the absorbing surfaces in m2.
a - absorption coefficient for each surface. (see charts)

The product Sa is referred to as the absorption A of a material and is sometimes expressed in sabins or metric sabins. The sabin is named for professor Sabin and is equal to the absorption of one square foot of open window. The metric sabin is equal to one square meter of open window.
In other words a wall with an area of 20m2 and absorption coefficient of 0.3 will absorb as much sound as an open window with an area of 6m2.

Sample Problem


A rectangular room is 10m wide, 15m long and 5m tall. The room has carpet on concrete with no padding on the floor, painted concrete block walls and a wooden ceiling. Using the foll
qowing absorption coefficients, find the T60 for the room at 1000Hz.
afloor = 0.37
awalls = 0.07
aceiling = 0.08

Volume = (10m x 15m x 5m) = 750m3

Safloor = (10m x 15m x 0.37 ) = 55.5 m2
Sawalls = (10m +15m) x 5m x 0.07 x 2= 17.5 m2
Saceiling = (10m x 15m x 0.08 ) = 12 m2
SSa = 85 m2

T60 = (0.161s/m)(750m3)/(85 m2) = 1.4 seconds

Absorption Table

This data table was sent to me by Dave Moyer from Riverbank Acoustical Labs.
(See second packet)


Sound Measurements:

Equipment:

Extech Sound Level Meter 407740
LabPro Interface with appropriate cables
Computer or TI Graphing Calculator
Existing Sound System with CD player
Sound Samples on CD*
12" latex balloons
Logger Pro 3.0 software

*In order to make these measurements as simple as possible a CD of tones was used rather than a signal generator. The intent is that anyone who has the ability of
down load sound files from a web page and transfer them to a CD will not need a signal generator. The tones may be down loaded from our web site at the top of this page.
They include 10 second samples of the following tones.
250Hz
500Hz
1000Hz
2000Hz
4000Hz
Brown Noise
Pink Noise
White Noise
Silence -I found it helpful when using the CD to put a track of silence between each of the tones.


Impulse method

The impulse method is a quick way of making T60 measurements. It has the advantage on not needing a frequency generator (CD player). However, the drawback is that measurements in specific frequency bands cannot be made unless a filter is used on the receiver. The filter is more likely to be more difficult for the average high school teacher to get than a CD player. Another problem associated with the impulse method is that of early reflections. Although these reflections are an important characteristic of the room and should be observed, their spikes can make T60 measurements more difficult.

To create a large sound impulse a large balloon is popped in the room while a sound measurement is made. Not only will a large balloon tend to be louder than a small balloon but it will also tend to excite more of the lower frequencies.


Tone cut-off method:

The sound absorption coefficient for a material can be different for different frequencies of sounds. Thus, a material which may be a good absorber of high frequencies may be a poor absorber of low frequencies. In many applications only a
certain frequency range is of interest. For example, most speech is in the 500-2000 Hz range. Thus, for a room to deliver speech clearly it would be best to keep the T60 for these frequencies low.
An acoustic engineer can easily measure the T60 for a specific frequency by using equipment capable of analyzing different frequencies at the same time. However, this equipment will cost thousands of dollars. The tone cut off method is a less expensive but more time consuming way of examining the absorption of specific frequencies. The procedure is quite simple. A pure tone is played in the room at a relatively high sound pressure level for a second or two and then turned off. The sound pressure level vs time graph is then analyzed to find T60.


Triggers:

Because of the delay between the interface and the computer/calculator, using a trigger is helpful. I had the least problems
using an increasing trigger with both the impulse method and the cut off method. During the impulse method the trigger was set, then the sound of the balloon activated the data collection. During the cut-off method the trigger is set, then the tone is played. A short time after the trigger has been tripped the tone is stopped. The trigger, although helpful, is not necessary and quality data can be obtained without it's use.


Sampling Rates:

The calculator allowed a total of 100 samples. I chose to collect for 4 seconds. This will be long enough for most rooms of interest. The sampling rate would then be 25 samples/sec. The computer allowed for much higher sampling rates. I chose two different rates. 25 samples/sec for 4 seconds to match that of the calculator and 2500 samples/sec for 4 seconds. The higher value was very near the highest the hardware/software would allow.

Data Analysis:

I used two similar methods to find T60. Since it is often difficult to measure a 60dB drop, most acousticians will find T60 by doubling their T30 measurement. When using the calculator the time for the peak was subtracted from the time of 30dB less than the peak. This T30 was doubled to find the T60. Since the time interval between readings was relatively large at 0.04 sec and the sound level was changing rapidly, the time recorded for the 30dB drop was often an approximate linear interpolation of the two nearest points to the nearest 0.01 seconds. The second method was to divide 60dB by the slope of the linear regression line through the decay.

Conclusions:

SLM

The SLM has several settings from which to choose. "A" weighting was chosen rather than "C" weighting because of the target frequency range. The 50-100 dB setting was used to best match the sound lev
els. Both the Fast and Slow response settings were tested. In every case the T60 was lower when the SLM was in the Fast position. Usually the values were 10-20% lower. Since these lower values were more consistent with the results found by a local acoustic engineer, I suggest the Fast response setting be used when measuring T60.

25s/s vs 2500 s/s

While using two very different sampling rates, certain trends became clear. The T60 values were similar. The differences between showed no signs of a consistent bias. However, it was much easier working with the lower sampling rate. The data files were much smaller and thus loaded more quickly. The analysis was faster and easier. At the lower sampling rates the graphs forming in almost real time made the collection process easier. The main advantage I could see for the higher sampling rate is the improved definition of shape in the decay graph.(see graph 1) This is of special interest when studying the resonance of a room at a particular frequency. I suggest sampling rates between 25s/s and 2500 s/s but much closer to 25 s/s. I would suggest higher sampling rates when room resonance is of special interest.


T30 vs Slope

When finding T60 by the two different methods I found that the T30 time gave a slightly lower value for T60 then by using the slope. This was not true in every case but an overall result. This is an expected result since the graphs curve and the range over which the slopes were taken usually extended past the T30 time.


Clap, Balloon, Tone and Noise

I tried several ways to produce a testable sound. Clapping is the easiest, but it is difficult to make a clap loud enough except when the clap is very near the SLM. A large balloon is very good but does not yield information about specific frequencies. It also introduces detectable reflections which may or may not be of interest. Pure tones are good for specific frequencies but they may introduce room resonance which may or may not be of interest*. The Brown, Pink and White noises were without a doubt the most consistent.(see graphs 2 & 3) The decay graphs were very linear thus the difference between T30 and slope methods almost vanished.
*Acoustics labs use rotating deflectors to eliminate resonance within the room.


Room Resonance

I tested three large rooms. A 2500 seat High School Gymnasium and a 600 seat High School Auditorium both built in the late 60's. I also tested a 300 seat church sanctuary built in the mid 90's with very little concern for acoustic properties. The gym and sanctuary both have large parallel walls. The auditorium had several common acoustical features including sound absorbing material and few parallel wall surfaces. The gym and Sanctuary both showed a great deal of room resonance while the auditorium showed almost none.(see graphs 1,3 & 4)


Loudness and Decay

When the tones were played at different sound levels the T60 was slightly different. In almost every case the louder the sound the shorter the T60. This was to be expected because of the slight curve of in the graph.(see graph 5)


Room changes

It was quite easy to make changes to the room that would cause measurable changes in the T60. One of the easiest is to open doors and windows. One obvious problem with this is that it could allow unwanted sounds to enter. Another simple change is to add/remove sound absorbing materials. These could be acoustic tile, people, carpet, pieces of foam, etc.


Background Noise

Background noises are a great concern when doing studies in room acoustics. These noises often go unnoticed until you start collecting data. It is important when choosing a room to study to consider when these noises will be at a minimum. Try to find a time when no other people will be around. Make prior arrangements for air moving equipment to be turned off. These problems were of special concern to Wallace Sabine while he was doing studies in Boston during the early 1900's. So much so that he did many of his studies during the wee hours of the morning. These were major factors leading him to a new acoustics lab in rural Geneva, Illinois. History buffs would find the circumstances surrounding his move to Geneva quite interesting.


Correlation between the different methods

It is somewhat difficult to draw definite conclusions when comparing my results using the SLM with the results from the acoustical engineer. Although he had standard professional grade equipment he only collected one sample from three locations and he only used the impulse method. He also only took measurements in the church sanctuary. My 500Hz average value is 8% lower than his. The 1000Hz value is 8% higher. The 2000Hz value is less an 1% lower. Thus, over this frequency range my average results are very close to the results given by his small sample.

When comparing the measurements taken with the SLM to the Sabine Method of Absorption the results are not as close. The Sabine method gave T60 times which were 20-50% lower than the values I measured using the SLM. I attribute this difference to the uncertainty in the coefficients. The church sanctuary had very large and uniform surfaces. This made the calculations of the area much easier. However, since I somewhat unsure of the precise coefficients to use for some of the surfaces, the T60 values are more uncertain than I would hope them to be.


Future Studies

An area of interest that I did not have time to investigate was the correlation between the regular "bumps" in the decay graphs and the room dimensions and location of the SLM. I believe the data I already have might be able to shed some light on this relationship. The fact that the geometric shape of the gym and sanctuary are so simple should make the analysis easier.


Final Note

I have found that although the equipment and procedures that were used were not intended to give the type of results required by acousticians, they are quite capable of giving the results needed to teach some basic concepts of room acoustics. The significant advantage is that these results can be obtained with equipment costing thousands of dollars less than that used by the professionals. Another advantage is that these procedures could be preformed by a small independent group of students.

Resources
The Physics Teacher.
Main article: Jesse, K.E. (1980). "A Classroom Acoustical Absorption Experiment,"
TPT 18:41
The Science of Sound Thomas D. Rossing Ph.D
Andrew Morrison NIU Physics Department
Dave Larson S&V Solutions, Inc. Sycamore IL
Dave Moyer Riverbanks Acoustic Laboratories Geneva, IL