Acoustical Testing I Loudspeaker Study

Transcription

1 Acoustical Testing I Loudspeaker Study Submitted to: Dr. Dominique Chéenne & Dr. Lauren Ronsse Columbia College Chicago November 19 th, 2014 By: Andrew Hulva Cody Elston, Daniel O Brien, Erich Pfister

2 Table of Contents Abstract... 3 Introduction... 3 Methods & Results... 4 Frequency Response... 5 Crossover Frequency... 7 Polar Directivity

3 Abstract Students were to study the Event TR8XL powered loudspeaker in a semianechoic environment (assuming all measurements to be free-field), obtaining the frequency response, polar directivity, and crossover frequency. The caveat was using only the TEF-20 software package, a suite foreign to all members of the group. With an imperfect test chamber, all aspects of the testing process had to be considered, which lead to a more thorough understanding of the testing process. Introduction An Event TR8XL was the loudspeaker subject to analysis. This two-way, 150- watt, powered loudspeaker had driver diameters of 8 and 1, and enclosure dimensions of x x The group undertaking this analysis consisted of Andrew Hulva, Cody Elston, Daniel O Brien, and Erich Pfister. The anechoic chamber at Columbia College Chicago was imperfect, and with that came a set of limitations that had to be understood. The first limitation was the bandwidth of measurement, and this was limited by the size of the chamber. With approximate dimensions 9ft x 15ft x 11ft, modal influences imbued variations in the data below the 250 Hz band. This frequency band was the nearest above the calculated resonant frequency (248 Hz) of the smallest dimension, from equation 1 which found the primary axial mode of a dimension (derived from the resonance of an open tube). f r = 2c L where, fr = resonant frequency (Hz) c = speed of sound ( ft/s) L = room dimension (ft) Equation 1 The second limitation was the fact that the chamber was not absolutely anechoic. Common modern anechoic construction consists of a massive wall material covered with absorptive pyramids whose depths exceed 3 ft. Their room 3

4 volume is approximated to be twice that of the chamber at Columbia College Chicago. The chamber at Columbia College Chicago is constructed of the common massive wall material (concrete), but the walls are covered with layers of differing density Roxul. While the Roxul in the installed configuration had a NRC value greater than 1, it was its depth (approximately 6 in.) that kept the chamber semianechoic. Methods & Results Prior to any data collection, proper signal flow had to be established as there was no previous configuration left for the group. Connecting the two rooms (classroom and chamber) was an analog snake with patch points at the teaching station and inside the chamber. The test station equipped with the TEF-20 hardware and software was connected to the chamber via this snake. A visual representation of the signal flow is shown in Figure 2. Figure 1: Signal flow showing microphone and preamplifier model (courtesy of Erich Pfister) For all tests, the loudspeaker was positioned on top of an electronically controlled turntable approximately 3.28 ft. (1 m) from a Behringer ECM

5 omnidirectional, condenser microphone, nonparallel to the chamber walls. The microphone, vertically, was positioned equidistant from the bottom of the tweeter and the top of the woofer. This is shown in Figure 2. Event TR8XL Turntable Behringer ECM 8000 Figure 2: Arrangement of speaker and microphone Frequency Response A broadband (20 Hz 20 khz), relatively low-resolution (50 Hz) test was first executed to obtain data that could be compared to a higher resolution (constant 1/10 of lower bound per octave), piecewise, dataset. This is shown in Figure 3. Notable deficiencies Figure 3: Broadband frequency response (20 Hz 20 khz) of Event TR8XL The low-resolution lead to an -18 db/octave low-end roll-off beginning at approximately 100 Hz. There are also two notable deficiencies at approximately 450 5

6 Hz and 650 Hz. There is no immediate explanation as to why these appear in the data. The piecewise test was conducted in octave pieces from 20 Hz to 20 khz, with a frequency resolution of one-tenth a given piece s lower bound. This higher resolution resulted in greater accuracy at lower frequencies. The data from this test is shown in Figure 4. Notable deficiencies Figure 4: Piecewise frequency response of Event TR8XL While the two deficiencies at 450 Hz and 650 Hz still appear in the data, the low-end roll-off begins at approximately 60 Hz and has a slope of -24 db/octave. An overlay of the two responses is shown in Figure 5. Figure 5: Overlay of the piecewise and broadband frequency responses 6

7 Crossover Frequency The most common way, in previous studies, to test a single driver was to dampen it by attaching a fiberglass pad to the diaphragm. While effective, there still existed the possibility of bleed from the damped driver. To mitigate this, the driver not being testing was disconnected from the amplifier. There was initial concern of the digital crossover providing a correction for the disconnection, but the sophistication of such a mechanism deemed it unlikely to have been engineered into this particular loudspeaker. Each driver was tested for its frequency response and the resultant responses overlaid onto a single graph. The frequency resolution of the woofer (8 ) and tweeter (1 ) tests was 125 Hz. From the loudspeaker s specifications that are readily available online, the crossover was reported to be of fourth-order at 2.6 khz. The results of the tests are shown in Figure 6. Figure 6: Individual woofer (red) and tweeter (tweeter) responses The intersection of these two curves occurs at approximately 3 khz; however, this intersection occurs approximately -1 db from the beginning of the downward slope. A cutoff frequency of a high/low pass filter (a crossover utilizes both) is defined as the frequency that has a response -3 db from unity. A crossover is designed such that the high and loss pass filters have this cutoff point at the same frequency so their combined responses are 0 db. The error here is from the 125 Hz resolutions. To correct for this, the sample size for estimating the crossover was 7

8 increased from one to two, reducing the error in by a factor of 1/ 2. Instead of using the intersection of the two curves, the -3 db (cutoff) frequencies of both were arithmetically averaged, as shown in equation 2, and a value of 2.7 khz was calculated. This compares well with the published value of 2.6 khz. f crossover = f woofer + f tweeter 2 where, fcrossover = crossover frequency (Hz) fwoofer = cutoff frequency of woofer (Hz) ftweeter = cutoff frequency of tweeter (Hz) Equation 2 Polar Directivity The polar directivity of the loudspeaker, in third-octave bands from 16 Hz to 8 khz, was obtained using the TEF-20 system in conjunction with an electronically controlled turntable. The frequency resolution was 50 Hz, with a 10 rotation between tests. The octave-band results from 32 Hz to 8 khz can be seen in Figure 7. Figure 7: Polar directivity plots The 16 Hz plot was not shown due to the published response to be 35 Hz - 20 khz, and the resolution being low relative to such a frequency. A frequency resolution of 50 Hz was accepted in this case due to the size of the drivers. At low frequencies (such as 32 Hz and 63 Hz shown above), the corresponding wavelengths are much larger than the diameter of the drivers (the most important factor in 8

9 hornless driver directivity). From the plots, one can see distinct lobes forming at 1 khz, with the directivity increasing as a function of frequency. Conclusions The accuracy of the data obtained from these tests is congruent with what was expected from the group. With the published value of the crossover frequency similar to the value obtained in this study, credence is lent to the accuracy of the frequency response and polar directivity, for which there is no direct comparison. In future tests, an increase in frequency resolution would improve the accuracy of all parameters tested; however, it would increase the time required, something that was of the essence in this study. The chamber itself could be improved upon with an increase in absorption depth. Since this is unlikely, a more practical approach would be to determine the lowest frequency for which the chamber is useful, and provide confidence levels for reproducibility below this value. 9