Quote:
Originally Posted by tINY 
In the original drawing from Germany, the 4 woofers on the front wall are shown to be about 1/4 of the way in from the ceiling/floor and side walls. If you only consider frequencies below where the distance between the drivers is 1/2 wavelength, then most of the room sees no destructive interference from the multiple drivers (at least until the wavefront bounces off the back wall). Frequencies higher than that (where 1/2 wavelengths are less than the driver spacing) will not be coherent and you get positional frequency response variation.
|
I am very confused by this 'figuring'.
Any frequency where the spaced sources are greater than 1/4 wavelength results in comb filtering and polar lobing. I am not sure what 1/2 wavelengths have to do with this, except that they will result in comb filtering and polar lobing as well.
“1/4 distance to a boundary” and ½ wavelengths do not see interference until the signal ‘bounces off the back wall’? I'm not quite sure of the reasoning here regarding 1/4 distances and 1/2 wavelengths bouncing off walls..., but dependent upon the gain of the reflected signal (which will be less with significant spacing and the inverse square law) and the timing/distance, yes, the signals, regardless of frequency or total wavelength, will/can still act as virtual sources and combine via superposition and contribute still more to the total superposition.
Based upon the time or distance offsets, the initial null frequency and the null frequency interval at which the comb filtering will repeat is easily calculated.
With multiple spaced sources, you are subject to comb filtering and polar lobing due to the combination of sources from ALL spaced sources, be they real drivers, or virtual sources such as walls.
In this room we have the spacing from each driver to each boundary (up down, sideways) where the effective wavelength is 2X the separation distance – meaning that if the distance to the surface is 4 foot, then 2X = 8 foot. And 8 ft is the quarter wavelength corresponding to a 32 foot wavelength, or a frequency of ~35 Hz!
With a time differential of 7.1 ms.
And the initial null frequency is 71 Hz.
And the null frequency interval is 141 Hz, with the first three null frequencies at: 70.6 Hz, 212 Hz and 353 Hz.
If the horizontal and vertical driver spacing is, say, 6 foot between the 2 top horizontal speakers, then the 1/4 wave frequency is 24 foot, or ~46.9 Hz.
With a time differential of 5.3 ms.
And the initial null frequency is ~94 Hz
And the null frequency interval is 188 Hz; with the first three nulls frequencies at: 94.2 Hz, 282 Hz, 470 Hz.
Thus, the diagonal driver spacing would be 8.49 feet.
With a time differential of 7.5 ms.
And the initial null frequency is 66.5 Hz.
And the null frequency interval is 133 Hz; with the first three null frequencies at: 66.5 Hz, 199.5 Hz, 299.5 Hz.
The reason they are worse than with simply one driver or two closely packed drivers where the reproduced passband is within ¼ wavelength and all effectively sum is that you will have multiple ‘initial’ null frequencies! And you will have multiple repeating comb frequency intervals as well as corresponding polar lobing intervals. In other words, the direct signal propagated from the speakers will exhibit much greater comb filtering and greater polar lobing anomalies. And the more sources sharing the same offsets located on the same plane, the more the response is reinforced. This can become manifest in a significant impact upon the locational cues.
Combine this with the modal response, and you have a mess... MUCH 'larger' than if you did not have the multiple spaced drivers reproducing the same passband. Hence, far from being an optimal design, this configuration simply exacerbates the response anomalies in an attempt to increase gain. There are much better ways to increase gain without the attendant anomalies.