It is unfortuate that the measurement software does not allow for more than five plots at a time in an overlay. Otherwise the inclusion of the gym-derived frequency response would be a fruitful addition to the plots. Additional plots will be displayed below that will include that reference.
A reminder before discussion begins, 1/12 octave smoothed plots may not be as precise a match to human hearing perception as 1/3 octave smoothed plots, but they do provide substantially greater 'engineering' data than do 1/3 octave smoothed plots. Given that we are looking at a dipole line array in-room response, the effects of 18" lateral microphone position shifts are much more evident with this level of smoothing than with a greater level.
What points of interest can be seen in these plots?
1. Each plot set bears a 'family' resemblance across the five microphone positions for it's experimental condition. There are no significant differences in frequency response below 100Hz across the conditions. Significant amplitude disparities occur between conditions above 100Hz.
2. The octave band response between 50-100Hz is the same across all three conditions, and is at a substantially higher amplitude than the anechoic reference response in that range. This is significant, and indicates strong room effects.
3. The frequency band 100-300Hz contains the bulk of the significant frequency response changes based upon the distance from the front wall. The response in the 44" from the wall condition is the worst of the three conditions, and has a deep dip and a peak in this range spanning an amplitude range of approximately 24dB.
4. The most linear amplitude response overall is seen in the 22" from the front wall position. This is the condition in which the baffle's inner edge is on line and in contact with the hanging foam sheet. The amplitude span in the 100 - 300Hz range in this condition is approximately 12dB.
5. In the 33" condition, there is an extremely linear amplitude response (by comparison), across all microphone positions, from about 225Hz on up to the measurement limit of ~650Hz. By comparison, each of the other two conditions show significant peaks and dips in that range.
6. Across the three condition, from 22" to 44", the extreme right microphone position (red traces) from 200Hz on up shows a fascinating progression of increasing frequency response ripple width. An open question will remain from this study as to what is the relation between those differences and theoretical dipole behavior?
Further discussion will ensue after presentation of the following alternative displays of the data. Below are shown completely unsmoothed plots (brace yourself) of the 'right', 'middle', and 'left' microphone positions compared with the anechoic plot from the gym. It is an illuminating tutorial to see such comparisons of unsmoothed frequency responses.
Comparison between anechoic and 22" from wall reverberent condition.
Comparison between anechoic and 33" from wall reverberent condition.
Comparison between anechoic and 44" from wall reverberent condition.
Bracing, aren't they? Reverberent plots are so much more messy than we like to believe. The highly smoothed plots from popular audio publications do tend to encourage a 'smoothed' view of loudspeaker responses, and, in my opinion mislead the naive about expectations for in-room response.
With the inclusion of the anechoic 3m/1m response from the gym in the above overlays we can immediately begin to discern features of the speaker and room interaction:
1. The DynaPleat's response below 100Hz is trivially affected by the baffle's distance from the wall in the three experimental conditions! On the other hand, the response below 100Hz is dramatically affected by room eigenmodes across all conditions.
2.There appear to be about six room eigenmodes below 100Hz that are almost equally excited by the DynaPleat array in the three different baffle distance-from-wall positions. Centered at approximately 37.5Hz, 45Hz, 65Hz, 75Hz, and 85-95Hz they are dominant features of the frequency response below 100Hz across conditions. The 6dB to 9dB increase in amplitude in the 65-95Hz range above the reference level is highly significant to the low frequency performance in the listening room. The 45Hz and 75Hz modes are nulls at the listening position.
3. The range 100Hz to 300Hz is greatly affected by the baffle distance from the wall. This is a bad range in which to have so much potential variance.
4. The 44" from the wall experimental condition is also seen again as the worst position of the three for the DynaPleat array for frequencies below ~650Hz. There is a significant reduction in amplitude between 100Hz and ~150Hz, and a peak at ~225Hz in this condition compared with the two other conditions.
What is to be learned from all of this?
Standard dipole placement wisdom, 'as far from walls as possible', may not be the best tact for optimum frequency response in range below ~650Hz. Room eigenmodes dramatically alter the anechoic low frequency response of a speaker, even a dipole. Acoustical damping treatment in the room maybe of significant benefit, especially in close association with a dipole.
Let us ask the question, based upon these results, of what would be the situation in the listening room had the DynaPleat array been equalized to some level of low-frequency flat response based upon the anechoic response? For example, what if a +6dB boost had been applied at 70Hz to nominally bring that level up, based upon the anechoic response, to the approximate level of the frequency response above 300Hz? The range between 60-100Hz would then have an amplitude above that of the higher range in the listening room, potentially leading to boominess or fatness in that range.
Of the three responses associated with the three distances from the front wall I would choose to use the 33" position. Since the response below 100Hz is the same in all conditions, other criteria must be used for selection. The in-room frequency response linearity above ~225Hz for the 33" condition just cannot be ignored. That will be the critical crossover region for this driver, and linearity through the crossover region will be highly beneficial. The frequency response depression between 100Hz and 225Hz is relatively benign, and could be easily equalized with a parametric section.
In closing, I emphasize my astonishment at the effect of the low frequency room eigentones. During the course of the RD75 Dipole Baffle Study I played the RD75 driver in many different dipole baffles in essentially the same position(s) as the DynaPleat array was measured at in this study. An equivalent 'in-room' study for that driver was not performed, but based upon listening perceptions and crude measurements with warble tones and a Radio Shack sound level meter I became convinced that the standalone RD75 dipole baffles, with a 150Hz 2nd order high pass active filter in circuit, were generating significantly greater low frequency performance in the listening room than implied by the anechoic gym measurements.
This study seems to provide some confirmation for those past perceptions and simple measurements. There is inspiration in this study for further work.
John Whittaker
January 1999
Page 1
Page 2
Page 3
Acoustic Line Source Research - Table of Contents.
