CM1 Measurements and Analysis
Measurements were performed using the Sample Champion program developed by Paolo Guidorzi of Purebits. The program is a Maximum Length Sequence (MLS) program that can generate an impulse response from this pseudorandom signal. The Fast Fourier Transformation is applied to the impulse response to generate the system frequency response. The operating principle is the same as that of the well known MLSSA program.
I received speakers with the sequential serial numbers: 6507 and 6508. I have used these serial numbers to identify each test measurement taken.
Test samples were taken using 16 bit resolution with a 48 kHz sampling frequency. Fast Fourier Transformations were performed at 64K allowing for a frequency resolution of .73 Hz, which is more refined than even 1/24th octave resolution at the lowest octave; the A0 key on the piano has a fundamental frequency of 27.5 Hz. Quasianechoic response was compiled using a Blackman-Harris, half right window based on crossover frequencies provided in the owner’s manual.
Figure 1: Impulse response of unit 6508.
The first figure is the normalized impulse response captured from unit 6508. A theoretical impulse response should appear as a single pulse up from zero and back for a single driver. This is a composite response for a two-way speaker system, and while the multiple drivers each have their own peak, a speaker with well damped behavior should have its dynamic response die out with few additional oscillations. The plot shows behavior typical of a two way system, the initial spike from the tweeter followed by a smaller hump from the mid/woofer, which has a slower rise time than the tweeter. There is rapid decay for both drivers, indicating a high level of damping. This is consistent with a design that uses an inverse horn loaded tweeter that damps out back reflections and the Kevlar mid/woofer; a material with has a high internal damping.
Frequency response plots were generated for 1 meter impulse response measurements time windowed over a 4 millisecond span. Bass response was measured from summed woofer and port near field measurements corrected to the far field.
Figure 2: Frequency response of unit 6508 on axis and 30 degrees off axis.
The second figure is of the frequency response for 6508 for on axis (green trace) and 30 degrees off axis (red trace). Some non-smooth frequency separation begins between 3000 and 4000 Hz, near the crossover frequency, while lobbing of the response begins around 8000 Hz.
Figure 3: Comparison of units 6508 and 6507.
I also compared frequency response for the two units, 6508 (green trace) and 6507 (red trace), shown in the third figure. Response variations between the two units show fairly good correlation over the quoted frequency response. The largest variations occur near 100 Hz and 1500 Hz, in the range of the mid woofer where response will be more sensitive to any geometric variations in the Kevlar weave.
Figure 4: Cumulative spectral decay of unit 6508.
The next figure shows the cumulative spectral decay based on the impulse response shown in figure 1 windowed over 4 milliseconds for quasi-anechoic response. The plot shows decay over a 50 dB range for frequencies in the midrange and treble regions. The speaker shows good decay with fairly smooth behavior over the time interval.
Figure 5: In room response of units 6508 and 6507.
The final plot is of in room response. Unit 6508 (green trace) was measured from the left stereo position while 6507 (red trace) was measured for the right stereo position. From the plot, significant in room roll off can be seen occurring just below 50 Hz.
