From: Gregory Kunkel <>
> Here is a quote from a paper previously mentioned in this forum:
> ACOUSTIC PROPERTIES OF PARABOLIC REFLECTORS
> Randolph Scott Little
> Laboratory of Ornithology
> Cornell University
> Ithaca, New York*
> "In order to avoid deep cavity resonances, as indicated by the 200 Hz
> peak, the microphone should lie outside the plane of the edge of the
> reflector. On the other hand, in order to suppress sounds coming
> from directions behind the reflector, the microphone should be well
> inside the rim. A suitable compromise is to place the microphone at
> the plane of the rim. This means that the focal length must be
> one-quarter of the diameter in order to satisfy the formula for a
> parabolic curve."
> Greg Kunkel
In light of more recent work, a good deal of what's in that paper is
very misleading, if not wrong. It is interesting from a historic
perspective as it represents the beliefs about parabolics at the time.
Probably partially as a result of the sort of parabolic designs used
A parabolic, by it's very nature does not have deep cavity resonance's,
that requires multiple reflections, and sound traveling inside the dish
is quickly reflected back out of the dish. The only resonance we have to
deal with is the dish material itself, which when hit by a outside force
resonates. Sound itself is not generally enough force, insects, brush,
movement, wind, are the more common ones.
The 200 Hz "peak" is a anomaly of the particular parabolic design
tested. It's not the same in all parabolics. And it's not resonance, but
a interference pattern between direct and reflected sound.
As far as ideal depth, best gain and smoothest gain increase with
frequency occurs at a l/a ratio of 4 (where l is dish overall depth, and
a is focal length). This has the mic well inside the dish. Sten
Wahlstrom, in the paper that detailed all this stated that a l/a of 1
was a practical compromise as the difference from ideal was not that
great. As the focal length increases beyond the plane of the front of
the dish the mic is moved out into the interference patterns produced by
the interaction of the direct and reflected sound. This results in a
distinct decrease in gain in the lower frequency area (the 200 Hz is
part of that). At a l/a of 1/4, there may even be a specific frequency
in which the gain of the reflector is negative. Sten's paper has some
graphs that are better than words at explaining this.
The Telinga dish has a focal length well inside it's dish depth. Which
partially explains it's good gain performance at low frequencies.
Someone asked about putting a mic off axis. If the mic is a little off
the central axis, there will be a irregular falloff of gain in the
higher frequencies. There will still be plenty of gain, but it won't
increase with frequency. And the farther off axis you get, the worse it
Back when we had a previous discussion of parabolics I put up a pdf of
Sten' paper with the warning it would not stay up on my website forever.
By some quirk of fate it's survived a couple housecleaning's of my site
and is still there:
(a note to our moderator, it might be worthwhile putting this paper on
Doug's nature recordist's binary area, for future discussions, it really
won't stay on my site forever.)
Sten's paper is now getting old too. But I don't know of anything more
contemporary that really addresses audio recording with a parabolic as
we do it. Sten's info agrees fairly well with my field experience with
Sten, for instance, set a practical limit for a parabolic at a l/a near
1. I wondered about that for a while, but I think I have a explanation.
During his time parabolics were pretty much made by a process called
spinning, of sheet aluminum. The spinning process is a neat way to make
a bowl, but as you make the bowl deeper it get's to be less stable to
do. (I've done spinning) A l/a of 4 would be tricky to do with this
process, thus Sten's practical limit. Nowadays we have many more methods
of making a parabolic reflector, and it would not be a problem to make a
dish with a l/a of 4. I'd really like to try one sometime and see how it
did. It is on my far too lengthy potential project list.
Sten, also did not discuss the use of mic baffles, which can greatly
improve the performance of parabolics with a l/a of less than 1. You can
see a simple baffle on my old homemade parabolic at the bottom of this page: