Non-Anechoic Testing Environment

Transcription

1 Non-Anechoic Testing Environment Acoustical Testing 1 Dr. Lauren Ronsse, Dr. Dominique Chéenne November 15 th, 2014 By: IRINEO JAIMES TEAM: CHRIS KEZON, HAYDEN JUBERA, NATHAN NEGRU 1

2 Table of Contents Abstract.. 3 Introduction....3 Measurement set up and Equipment Test Method Understanding Room Effects.4 - Non-Anechoic Environment Test..7 Results Conclusion 10 2

3 Abstract Using a non-anechoic room, the team develop methods of finding a reasonable flat frequency response for a test microphone. Using Gold-line s TEF software and hardware as the main measurement tools, the team was able to filter out unwanted room excitation created from the sine sweep test implemented. Strong first order reflections were treated manually by placing absorption panels in the areas where chance of reflection were high. Taking quick measurements in the time domain gave the best representation of an anechoic environment because the microphone picked up less of the room s excitation. A close to flat response was measured by using this method. Introduction The purpose of the non-anechoic loudspeaker experiment was to understand how anechoic measurements could be made outside of an anechoic environment. By understanding the principles of 1 st order reflections, any image sources effecting the data that would not show up in an anechoic environment got manually treated using Auralex studio foam absorption panels. The test loudspeaker and microphone both have their own limitations, an anechoic measurement was done to take those faults into consideration during testing. Figure 1 shows the frequency response using a sine sweep, in the anechoic chamber. The frequency response is flat as expected from Hz, after 8000 Hz the responses seem to Figure 1: Anechoic sine sweep test 3

4 vary drastically. If such peaks/dips in amplitude appeared in future test, it was due to the speaker and microphone configuration and not the room environment, and for the purpose of this experience could be ignored. Measurement set up and Equipment Using TEF analyzing software it was possible to 1 m. investigate strong 1 st order reflections, created by nearby surfaces. The loudspeaker was set up 1 m. away from the microphone. The loudspeaker and microphone were both 2 ft. away from the white board. In the first test a sine sweep was used to produce a frequency response graph and decay curve graph. The team made assumptions, the white board was going to be the contributor for the strongest first order reflection TEF measured, followed by the either the floor or light fixture above (out of frame). Figure 2: Initial test set up Then the second test repeated the same process with the exception of adding absorption panels to the strong suspected first order reflectors. The absorption panels used were 2 x 2 Auralex Studio foam absorption panels. Additionally the speaker was QSC model and the microphone was a TEF testing microphone. Standard issued TEF interface hardware was used for analog and digital interaction. Testing Method Understanding Room Effects Running the first test yields the Figure 3. Focusing only on the data below 10 milliseconds, there are three distinctive peaks, the first peak (initial signal) was measuring the sound energy the microphone was picking up directly from the speaker. Reason for only 4

5 focusing on the time below 10 milliseconds was because the longer the measurement went on the more of the room excitation the microphone picked up. The other two peaks are strong 1 st order reflections. The first reflection occurring at ~ 7.5 milliseconds is assumed to be the whiteboard. The second reflection occurring at ~ 8.1 milliseconds is assumed to be the floor or light fixture in the ceiling. #1) Strong 1 st order reflection #2) Strong 1 st order reflection Initial Signal Figure 3: Sine sweep test in untreated room For the second test, absorption panels were placed on the surface of the whiteboard and right below the light fixture as depicted in Figure 5. Running the same sine sweep test the 2 nd and 3 rd peak decrease in amplitude. Figure 4 shows the two strong 1 st order reflections almost completely disappeared. Confirming the whiteboard and the light fixture were the contributors of the early reflection. LEFT BLANK INTENTIONALLY 5

6 #1) Strong 1 st order reflection disapeared #2) Strong 1 st order reflection disappeared Initial Signal Figure 4: Sine sweep test in treated room The next set of tests looked into the frequency response of the microphone. Knowing what surfaces are the most reflective it becomes easier to do a frequency response with absorption vs. without absorption of the microphone. In order to get the best resolution for a frequency response piece meal measurements were made, and is represented by different colors in figure below. Then comparing that with a frequency response test without taking measurements in a piece meal manner. Across the the frequency spectrum there was a lot of variences in amplitude, which was expected in the untreated room Figure 5: Set up with absorption panels LEFT BLANK INTENTIONALLY 6

7 test. When the sine sweep test is done with absorption added the graph below, there is a much flatter frequency response. Still a various frequencies experience dips in amplitude, yet the room is having less of an effect on frequency response. Figure 6: Piece meal frequency response of untreated room Figure 7: Piece meal frequency response of treated room The piece meal frequency response test provides important pieces of information that helped with further testing. The amplitude dips at ~ 9000 Hz and various other dips above 9000 Hz is caused by the loudspeaker and the microphone set up, simliar dips were present when an anechoic measurement was done. Low frequency absorption is hard to implement due to the characteristics of low frequencies. When comparing Figure 6 with Figure 7 the frequencies < 200 Hz were not effected by the absorption panels at all, and for the purpose of further testing measurments of 1000 Hz 20,000 Hz will be the focus. 7

8 Non-Anechoic Environment Test Understanding the effects of the room, the TEF filter can now be introduced to the testing. The TEF filter allows higher order reflections to be ignored when taking the frequency response, decreasing the amount of phase cancelation. Figure 8 shows decay curve the test without absorption (white) overlaid with another test that used absorption panels on the reflective surfaces (red). The red curve is what would be expected to be seem in an anechoic chamber, and was achieved by treating the whiteboard and light fixture above. Doing so allows a wider bandwidth filter to be used, yielding more accurate results. Figure 8: Time response of treated room (red) and untreated room (white) with 45 Hz filter Prior filter was set up to measure the frequency response from 0-2 ms then filter out the rest, trying to avoid the 1 st order reflections, depicted in Figure 9. Proceeding with this method would have filtered out more useful data than desired. Instead treating the reflective surfaces and widening the bandwidth of the filter to about 0-8 ms, yield better results with higher resolution, demonstrated in Figure 8. LEFT BLANK INTENTIONALLY 8

9 Figure 9: Time response of untreated room, with 15 Hz filter Results Proceeding with the wider bandwidth filter a frequency response was done in the nonanechoic room, Figure 10 shows the result. The results are promising, the graph looks very similar to Figure 1 that was tested in the anechoic chamber. Figure 10: Frequency response from 1 khz 20 khz, in treated room Removing the absorption panels and testing the frequency response when 1 st order refection are present yields Figure 11, shows a lot more comb filtering happening. Yellow curve 9

10 is an overlay of no absorption test. Figure 11: Overlay of both frequency response with absorption and without Reviewing the results, it is possible to test the frequency response of a microphone using a non-anechoic environment. However, it becomes difficult to determine whether the microphone is creating amplitude dips toward the higher domain in frequency content if an anechoic assessment on the equipment cannot be done. Understanding what surfaces in the room will create strong 1 st order reflections is important if the TEF filter method will be used. Conclusion In conclusion, measuring an anechoic like responses from a non-anechoic environment is possible by eliminating strong first order reflections and taking quick time domain measurements of the test signal. The method is great for quick estimation when an anechoic environment is not readily available, however, the decay curve using such method involves measuring some of the excitation of the room, and it would be difficult to get a flat response, short of placing absorption panels on all the walls in the room. Using TEF filters and absorption panels can minimize the room effects. 10

11 Future studies would involve testing more than two (2) locations in the room. Doing so would give insight into how different boundaries effect frequency response. Testing at different distances from the loudspeaker were avoided due to time constraints, future testing will include such data. Testing with various microphone and loudspeakers were also avoided due to time constraint, but anechoic testing clearly showed amplitude dips due to either the microphone or loudspeaker. 11