Dick Pierce wrote:
The ONLY way possible to avoid comb filtering effects is to have a
single loudspeaker placed in an anechoic chamber.
As soon as you introduce multiple sources, whether real or virtual,
you WILL have comb filtering. (A "virtual" source is a reflection
of a real source. For example, place a loudspeaker 2 feet above the
floor: at frequencies where the floor is reflective, which is most
frequencies, there is an apparent or "virtual" source 2 feet BELOW
the floor, giving a doublet source with a separation of 4 feet).
That means that ALL stereo systems, ALL home theater system in ALL
rooms suffer from pretty massive amounts of comb filtering due to
path length differences to multiple coherent sources. It is completely
unavoidable in practice.
That being said, it can be argued that the more real sources you have,
especially sources that are placed different distances apart from room
boundaries and each other, the more evenly distributed the nulls of
the comb filtering will be and thus the smoother the overall response
due to comb filtering will be.
I think it can be argued, that, at its WORST, and additional center
channel speaker will NOT make things worse, and may potentially make
The arguments of the people you cite might be strictly true in a
first-order analysis but fail or at least loose much of their strength
when the entire boundary and multiple source considerations are taken.
One can further suggest that two competely coherent sources, placed
vertically, restrict radiation in the vertical plane, but not in the
horizontal plane, and thus can arguably REDUCE comb filtering effects
due to interference effects with the virtual sources under the floor
and above the ceiling.
Another point: the demon "comb filtering" itself is something of a red
herring in reality. Comb filtering can only occur if the size of the
radiating sources are very small compared to the wavelength while at
the same time the separation between them is large compared to the
wavelength. For something like the dual center-channel approach,
the ONLY frequencies where this is relevant is high frequencies, and,
within the listening window, the path length differences are simply
not big enough to cause the problem AND where they are big enough,
you're sufficiently far enough off axis that you don't care.
I believe the comb-filter argument against using dual center-channel speakers is specious and unsupportable