Вы находитесь на странице: 1из 19

The Loudspeaker Study

Hannah Knorr
With: Jay Bliefnick, Michael Hansen, Matthew Johnson, Miles Possing, Ming Yu
Acoustical Testing I at Columbia College Chicago
November 27th, 2013
Attn: Dr. Dominique J. Chenne, Dr. Lauren Ronsse

Table of Contents

Abstract.3

Introduction...3

Equipment and Signal Chain.3

Anechoic Testing
o Equipment Setup...4
o Frequency Response..5
o Time Response..8
o Crossover Investigation9
o Polar Directivity..12

Non-Anechoic Testing
o Equipment Setup.13
o Frequency Response...14
o Time Response14

Extra Credit: Microphone Comparison15

Conclusion...17

Abstract:
In the Loudspeaker Study, students were required to analyze the frequency response, the
crossovers cut-off point, and the polar directivity of a loudspeaker using the anechoic chamber
and compare the found results to those derived in a non-anechoic environment. At the end of this
project, the students were able to successfully complete these tasks using TEF. The frequency
response, time response, and polar directivity were established, and the crossover was found to
be around 2.2 kHz. Additionally, a bonus microphone comparison was conducted to compare the
original results with.
Introduction:
The Loudspeaker Study was completed as a requirement of the Acoustical Testing I
course at Columbia College Chicago. This report was compiled by Hannah Knorr; however, the
project was completed in coordination with group members Jay Bliefnick, Michael Hansen,
Matthew Johnson, Miles Possing, and Ming Yu. In this study, students were challenged to
analyze the frequency response, the crossovers cut-off point, and the polar directivity of a
loudspeaker using the anechoic chamber in the basement of 33 E. Congress Parkway at
Columbia College Chicago and compare the found results to those derived in the non-anechoic
environment of the LL01 classroom in 33 E. Congress Parkway at Columbia College Chicago.
Equipment and Signal Chain:
The speaker analyzed in this study was an Event TR8XL. The microphone used for
testing was an ElectroVoice Re-55 omnidirectional dynamic microphone. The primary software
used for analysis was TEF. The basic signal chain for this study can be viewed in Fig.1.

Electrovoice
RE55

Input 3

TEF20 Box

PC

Snake

Output 1
Station #2

Event 8XL

Figure 1: The basic signal chain for the loudspeaker testing.

Anechoic Testing: Equipment Setup


For the anechoic testing conditions, the speaker was placed on a table around the center
of the anechoic chamber. The microphone was placed one meter away from the speaker.
Absorption was placed around the base of the microphone stand in order to eliminate potential
reflections from the stand. A picture of this setup can be viewed in Fig. 2.

Figure 2: The equipment setup for anechoic testing.

Anechoic Testing: Frequency Response


In order to analyze the frequency response of the speaker, a series of sine sweeps from 20
20000 Hz was run through TEF and the response was recorded. Initially, the group decided to
break the frequency spectrum down into octave bands and test each band with increasing
resolution. This was done in order to shorten the necessary testing time as higher frequency
resolutions take more time, allow for better resolution in the highest frequencies where it is most
important, and reduce the likelihood of TEF crashing. The results of this test can be seen in Fig.
3.

Figure 3: The composite frequency response of the speaker. Each individual color represents an octave band. As the octave
bands increased, so did the resolution, i.e. the rightmost blue line has the highest frequency resolution.

The dips seen in Fig. 3 around 500 Hz, 1000 Hz, etc., represent where the next octave band was
transitioned into. However, the dips themselves are pretty negligible as the group ran the test
parameters for each octave band a little below and above the desired values in order to ensure a
proper volume level was reached when TEF began the sample. Table 1 shows how the frequency
resolution increased for each octave band.

Frequency Band (Hz)

Frequency Resolution (Hz)

0 500

2.7

500 1000

3.3

1000 2000

5.1

2000 4000

7.1

4000 8000

10.0

8000 - 16000

20.1

16000 -

20.1

Table 1: The frequency used for each octave band in the composite frequency response test.

The interesting part to note about the frequency response was the overall low response in low
frequencies, mostly below 50 Hz. It was theorized that this was possibly a function of the
microphone and not the exact response of the speaker.
Additionally, a frequency response was taken at basically the highest resolution possible
for TEF. The frequency resolution was 2.0 Hz, with 8192 samples in 5000 seconds. This high of
a resolution was most likely overkill; however, the group was curious to see the results. These
test results can be found in Fig. 4.

Figure 4: The frequency response graph of the speaker at an incredibly high resolution.

This graph allowed a smoother overall picture with a clearer picture of the lower frequencies.
The most interesting thing about this result was the large dip around 400 Hz and the small dip
around 1500 Hz. This dip in frequency response was unexpected, but the group theorized that it
could be a modal effect within the room itself as the anechoic chamber being used was not a
100% perfect anechoic chamber. To investigate this modal effect, the speaker and microphone
were moved around 35 towards the corner of the room and a response test with a 5.0 Hz
resolution was run. These results can be seen in Fig. 5. Only the response between 100 2000
Hz was taken as this range covered both dips being investigated.
(This space intentionally left blank for formatting purposes.)

Figure 5: The investigation of the room mode effects. The white line is the normal frequency response, the yellow line is with
the speaker off-axis, and the red line is with the speaker and microphone off-axis.

As the graph shows, moving the microphone and speaker off-axis towards the corner of the room
eliminated the 400 Hz dip, proving modal effects were initially taking place in the room.
However, the 1500 Hz dip did remain, meaning either this is part of the actual frequency
response of the speaker or there was a vertical axis room mode taking place that was not
accounted for.
Anechoic testing: Time Response
Next, the time response of the speaker was taken in order to check for accuracy within al
the tests. The time response results of the speaker in the anechoic room can be seen in Fig. 6.

10

Figure 6: The time response of the speaker in the anechoic environment.

The primary impulse of the signal begins around 3 milliseconds after the test begins. This is
expected as the speaker is around 3 feet away from the speaker and the general rule is 1 foot per
millisecond response time. Therefore, the group was able to conclude a generally high accuracy
for all testing done in the anechoic chamber, i.e. there was nothing manipulating the responses in
a major way.
Anechoic Testing: Crossover Investigation
Two methods were used to find the cut-off frequency of the crossover. In the first,
absorptive material was taped over either the tweeter or the woofer in order to isolate each driver
and a frequency response was taken. The microphone was also moved directly up to the driver
being tested in order to minimize time effects, especially concerning high frequencies in the
tweeter. This setup can be seen in Fig. 7.

11

Figure 7: The setup for the driver isolation tests. The left is the tweeter-only test and the right is the woofer-only test.

It was theorized that where the major volume increase or decrease occurred, the cutoff would be
in that general area. Both isolation tests were taken and a general region was identified as
containing the crossover. The tests were then rerun from 1000 5000 Hz at a higher frequency
resolution and overlaid for comparison. This can be seen in Fig. 8.

Figure 8: An overlay of the driver isolation tests. The top line represents the tweeter-only test and the bottom line is the
woofer-only curve.

12

From this, the crossover was found to be between 2100 2900 Hz as this is where the two driver
frequency responses begin to cross.
As the range found in the first method was kind of wide, a second method was used to
find a more specific place for the cut-off frequency. In this method, the initial testing conditions
were reinstated, but the microphone was moved directly up to the speaker and the head was
placed directly between the two drivers. The hope was that by allowing for some driver phase
interaction, a more accurate response could be found. The frequency response was once again
tested; however, it was at a very low resolution (~100 Hz). This was in order to smooth out the
curves and find a definite crossing point. The results of this test can be found in Fig. 9.

Figure 9: The frequency response of the close-mic'd speaker. The red line is most likely the woofer and the white line is the
tweeter. The definite dip in each line gives the crossover cut-off frequency.

As expected, a definite dip was seen in the graph. This gave the group a better idea of the
specific cut-off. Here, the dip seems to be around 1800 Hz, which is lower than what was

13

expected based on the range found in the first test. It is interesting to note that the reported
crossover for the speaker is 2600 Hz. This is within the range found using the first method, but
800 Hz off from the results of the second method. This disagreement would be interesting to
further investigate.
Anechoic Testing: Polar Directivity
Another required parameter to analyze was the polar directivity of the speaker. To find
this, the speaker was placed on an ET2-ST2 turntable and rotated 10 each test. Then, the
directivity at each 1/3-octave was recorded by TEF, Eventually all these tests were compiled by
TEF and the polar directivity was found. This result can be viewed in Fig. 10.

Figure 10: The polar response for each 1/3-octave band as recorded by TEF.

14

From this, the general behavior of the speaker across all frequencies was derived. At low
frequencies, the speaker was almost perfectly omnidirectional. As the frequency increased, the
speaker became more and more directional, taking on more of a cardioid shape. This is expected
as high frequencies tend to be directional than low frequencies due to their short wavelengths.
Non-Anechoic Testing: Equipment Setup
A comparison of the speakers behavior in a non-anechoic environment was also
required. For this period of testing, the microphone and speaker were moved outside the
anechoic chamber into LL01. Once again, the microphone was placed one meter from the
speaker. The first tests were run with nothing around the setup. Second, absorption was placed
on the floor between the microphone and the speaker. Finally, a reflector panel was placed on
one side of the setup. Pictures of the configurations of the latter two tests can be seen in Fig. 11.

Figure 11: The setup for the non-anechoic testing. The left photo shows the placement of absorption between the
microphone and speaker. The photo on the right shows the setup with the reflector panel only.

(This space intentionally left blank for formatting purposes.)

15

Non-Anechoic Testing: Frequency Response


Using the first testing conditions only, the frequency response of the speaker in LL01
with a frequency resolution of 11.0 Hz was taken. These results can be found in Fig. 12.

Figure 12: The frequency response of the speaker found in LL01, a non-anechoic environment.

Obviously, the frequency response here is not as smooth. This is due to the effects of the room
on the speaker, i.e. modes, reflections, and signal interference. As in the non-anechoic
environment, the low frequency response is very low. There are many more harsh peaks and dips
than in the first tests, especially around 500 Hz, 1400 Hz, and 5000 Hz.
Non-Anechoic Testing: Time Response
The time response was also investigated in the non-anechoic environment. For this, all
three conditions (direct path, with absorption, with reflector) were utilized and compared with
one another. These results can be found in Fig. 13.

16

Figure 13: The time response of the speaker with no changes (black), absorption (blue), and a reflector panel (orange).

For all 3 tests, the direct sound attack was around 3 milliseconds after the test began. As in the
anechoic environment, this is expected due to the distance between the speaker and microphone.
In the first condition, no absorption or reflection, a second spike around 4.5 ms was noticed. It
was theorized that this was a reflection off of some surface. After absorption was placed on the
floor between the microphone and speaker, this spike does not occur. Thus, the first strong
reflection must have been from off the floor in front of the speaker. Lastly, the reflector panel
result was very close to the no condition result; however, a few more small peaks are seen with
the reflector, meaning some reflections were added in by the panel.
Extra-Credit: Microphone Comparison
In order to gain bonus points, the group decided to conduct a microphone comparison. To
do this, the ElectroVoice microphone used in all the testing was swapped out for an Earthworks

17

M50 condenser microphone. Then, the frequency and time response of the speaker were once
again taken and the results were compared to the ElectroVoice results. Fig. 14 shows the
frequency response of the speaker according to the Earthworks microphone.

Figure 14: The speaker's frequency response using the Earthworks microphone.

It can be seen that the Earthworks results are overall much flatter than the ElectroVoice results.
This is most likely due to the fact that the Earthworks is a Type I testing microphone specifically
designed for as flat a response as possible. Also, it was noted that the overall level of the results
was a little below the ElectroVoice levels found. This could be contributed to the fact that the
Earthworks is a condenser microphone and between needing a preamp and having to route the
signal further through the APHEX patch bay, some level was probably sacrificed. Finally, the
time response for the speaker was taken with the Earthworks microphone. These results can be
seen in Fig. 15.

18

Figure 15: The speaker's time response with the Earthworks microphone.

This test was in accordance with the ElectroVoice tests in that it took around 3 ms for the attack
the register after the test began. It was noted that the attack with the Earthworks seemed a little
bit sharper and negligibly faster. This could be contributed to the Earthworks being a condenser
microphone in that its transient response would be better than the ElectroVoice because its mass
is lighter.
Conclusion:
Using TEFs analysis software, the team was successfully able to derive the frequency
response, time response, polar directivity, and crossover cut-off frequency (2100 2900 Hz) of
an Event TR8XL loudspeaker. Also, these parameters were found in both an anechoic and a nonanechoic environment for comparison. Finally, a microphone comparison was conducted and
compared with the results found throughout the study.

19

If this study was to be repeated, there are many other aspects of the speaker that could be
analyzed. For example, the entire study could be completed with the Earthworks microphone for
comparison instead of only the frequency and time response. The crossover could be further
explored by varying the distance between the speaker and microphone to see if the specific dip
would move higher or lower in frequency. The speakers low frequency port could be
investigated to discover which low frequencies it is aiding. Also, the anechoic chamber and nonanechoic room conditions could be explored in order to determine their effect on the found
results.

Вам также может понравиться