Did some back-of-the-envelope math, assuming that the diaphragm was
1"x1" and that the plates were 1mm apart, and came up with a
capacitance of something like 5.4E-12 farads per mic. I think that
would require a rather high bias voltage to yield a workable signal.
Gain can be improved somewhat with thinner spacing. A business card
is approximately 0.25 mm thick, so that would give you 4 X the
capacitance per unit area. Doubling the width of the strips would
quadruple the capacitance of a single element (and quadruple the
overall area of the array).
Also, after doing some drawings, it became apparent that the size of
the sensor influences its directional response at any given frequency
-- so each element does have an effective, frequency-dependent comb
filter-ish "aperture". The net effect of all apertures at all
frequencies is a little obscure ...
On Oct 11, 1:02 pm, pezman <mikehoga...@gmail.com> wrote:
> Just wanted to extract this one item from the acoustic camera thread.
> I think that you could make a 32x32 array by arranging 32 parallel
> strips on a background and then suspending 32 parallel strips that are
> perpendicular to these "background" strips. If you record all
> signals, I believe that you can isolate the signal arriving at any one
> element using a suitable linear combination of the resulting signals.
> You should also (in principle) be able to beam-steer the sensor and
> post-process signals after the fact in much the same way as a
> plenoptic camera. In fact, you have the enormous advantage that you
> can preserve temporal information at pretty high resolution, so you
> can do all sorts of post-processing to manipulate spatial, phase and
> directional characteristics of the array (again, this is all "after
> the fact" processing of the signals).
> Copper tape on masonite is probably a decent way to build the
> background. You would need something very low-mass for the
> "suspended" strips, so that they can move with relative ease. And you
> would probably need to calibrate each microphone in order to linearize
> it for post-processing.
> At any rate, not tough to build such a critter. However, it might be
> challenging to collect signals at adequate resolution -- and there
> might be something fundamental that I have overlooked. Also, storing
> 64 signals at high resolution is no trivial task. You could make a
> smaller array, of course. For example, a 4x4 should be pretty
> feasible.
