naturerecordists
[Top] [All Lists]

Re: the nature of parabolic reflectors

Subject: Re: the nature of parabolic reflectors
From: Eric Benjamin <>
Date: Wed, 25 Feb 2004 10:45:46 -0800 (PST)
I must start out by confessing that I've never used or built a parabolic mi=
crophone.  In fact I've never even seen one.  But all of this discussion ha=
s piqued my interest.

In order to achieve the theoretical gain, the actual recording circumstance=
s must result in the waves from various directions arriving at the focus wi=
th phase errors conforming precisely to the expectation of theory.  Put ano=
ther way, everything must work exactly right, or the gain will be less than=
 predicted by theory.  This can be seen in Wahlstrom's figures 12, 13, and =
14.  The measured gain is always below the theoretical gain with the except=
ion of one data point in figure 14.

If follows that Errors of the following types may occur in practice:

Focal length error
Non conformance to exact paraboloid
Aiming error
Violation of the plane-wave assumption

Focal length error: If the microphone is not located precisely at the focal=
 point, then it will be progressively further away from the point where the=
 high-frequency maximum is, and the high-frequency output will be less than=
 the theoretical optimum.  This is shown quite clearly in Wahlstrom's figur=
es 9, 10, and 11.  One practical consideration is that the designer might l=
ike to use a tripod arrangement to support the microphone capsule from the =
rim of the dish, but for the case where =E1 =3D l this results in a non-rig=
id support system

Non-conformance to an exact paraboloid results in waves from one direction =
arriving with positive or negative phase shifts relative to other direction=
s of incidence.  Variations of 1/2 wavelength, which is only about 8.6 mm a=
t 20 kHz, will result in complete cancellation.  Not only can there be erro=
rs due to manufacturing tolerance, but for flexible paraboloids it is extre=
mely unrealistic to expect that they will snap back to exactly their origin=
al shape.  Assuming that the errors are not systematic, the result is merel=
y a failure for the gain to continue to increase at higher frequencies.

A side implication is that a spherical reflector may work nearly as well in=
 practice, and be easier to fabricate.

Aiming error when measuring (or using) the microphone will result in a roll=
ed-off frequency response.  Because the polar pattern of a parabolic microp=
hone becomes progressively narrower with increasing frequency,

A violation of the plane wave assumption will occur if the source of the so=
und is not located at infinity.  Again, in order to achieve the theoretical=
 gain from the paraboloid reflector it is necessary that the planarity of t=
he wavefront not be in error by more than about 1/4 wavelength.  Note that =
this effect increases with the size of the dish.  A very large paraboloid i=
s not appropriate for recording near sources.  Assuming that the dish is 0.=
5 meters in diameter, and that the distance is 10 m.

So for a dish of diameter 0.5 meters, 5 meters from the source, the "height=
" of the wavefront entering the dish is about 1.2 cm.  This will result in =
cancellation for a source at approximately 14 kHz.

In looking at Wahlstrom's analysis in his appendices, I see that the figure=
s do not precisely portray the shape of the gain curve.  Using Wahlstrom's =
analysis and the result in his Equation 14, I have calculated the theoretic=
al gain curve for a parabolic microphone of dimensions the same as that of =
the Telinga microphone.  The horizontal axis thus takes on the dimensions o=
f Hz.  The principle thing that can be seen that is not visible in the figu=
res in Wahlstrom's paper is that the oscillating part of the sound field af=
fects the response up to high frequencies.  The implication of this is that=
 the sound is dispersed in time, and indeed this would be expected given th=
at there is both a direct sound and focused sound component to the micropho=
ne signal.

If there is interest on the part of the group, I could expand upon this in =
more detail, and put the results into the files section of the Nature Recor=
dist group pages.

Eric









[Non-text portions of this message have been removed]



________________________________________________________________________
________________________________________________________________________

<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU