--- Walter Knapp <> wrote:
> From: Klas Strandberg <>
> > > > I have done lots of experiments with both the Telinga and other
> dishes. A
> > loudspeaker 30 meters away, - outside - a controlled sine sweep,
> and a
> > printer at the output of the mike. Realistic conditions!
>
> I like to experiment at much greater distances as well. At least 100
> meters if not farther. Also realistic conditions. One of the problems
> is
> that most detailed lab testing of parabolics is done at very close
> range. It's hard to have a soundproof chamber that's long enough.
> While
> those doing the math development think they understand the effect of
> this, there is a uncertainty involved.
Not having the parabolic mic at great distance from the sound source
will yield results that do not match the predicted response of the
parabolic reflector.
Dr. Swenson in his USACERL technical report "Reflector Microphones for
Field Recording of Natural Sounds" says his simulations used 1000
wavelengths as the distance from sound source to reflector mic.=20
http://www.cecer.army.mil/TechReports/pat_mike/pat_mike.post.pdf
Juha Backman in a paper presented at the 112th AES convention in 2002
attempted to measure the frequency response of parabolic mics, but
encountered problems due to the size limit of their anechoic chamber.
They measured at 3.6m from sound source to focal point. This paper is
AES convention paper 5499.
> > My findings are that if I make 10 similar measurements, the result
> will be
> > different up to =B1 6 db from case to case. This is because of the
> wind
> > (blowing frequencies away) and general acoustics, randomly
> changing. All
> > such changes are taking place all the time when we are "out there".
> >
> > If I make 10 different measurements, like holding the dish 1 meter
> up in
> > the air, or moving the tripod 3 meters aside of the previous
> position, the
> > measurements are VERY different from one another.
> > All such changes are taking place all the time when we are "out
> there". Very
> > few recordists, (they DO exist though!) - look for the best
> position to
> > record, like a photographer sitting down, standing up, moving three
> steps
> > left etc.
No doubt this is true. Even simulations show a rippled freq. response
curve, and considering off axis response varations, repeatable results
would be difficult indeed.
> > I am fully convinced that ALL SUCH ERRORS DESCRIBED, CAUSED BY
> DISTANCE,
> > WIND AND GENERAL ACOUSTICS - ARE MUCH GREATER THAN THE ERRORS
> CAUSED BY THE
> > PARABOL DESIGN (unless is is a poor design, that is.)=20=09
>
> I agree here, and it's not just parabolics, every kind of mic this is
>
> true. And the pre's and recorders. What is also important is to be
> able
> to distinguish the source of some effect (error is too strong a
> word).
> Far too many times something someone points out as a equipment error
> belongs. Even if we don't like it
For a parabolic reflector to work per theory, it is assumed that the
waves entering the dish aperature are parallel. Wind, reflection,
thermals and refraction from the environmental features certainly
affect the parallelism of a distant sound source.
> > So what do we have? Parabols + mike are rather "coarse" designs,
> which are
> > not very precise, not very controlled, but still work when nothing
> else works.
>
> I don't consider a parabolic to be any less precise than any other
> mic
> design. Sure there are irregularities in reflector shape and so on.
> But
> all mics have such things, just on most mics the scale is smaller.
> > The development of the Telinga has taken several "natural" steps,
> as birders
> > have used it, and they have given me valuable feedback. 90% of such
> feedback
> > has been about practical concerns, all from a new design of a
> shoulder strap
> > to which mosquito spray you can use on the dish.
> >
> > "Sound quality" has been judged by my own ears and, if possible,
> inside and
> > outside measurements. Inside measurements are very good when
> designing the
> > mike capsule itself. It took many sleepless weeks to force some
> sense out of
> > 4 electrets working in parallel, towards a separating plate, for
> example.
> > For this I used a sound dampened "box" with a loudspeaker on top,
> making a
> > sine sweep, 300 to 20.000 Hz. Fairly flat. Then I could work with
> the
> > capsules, move them around, try different positions and dampening
> materials
> > and so on, and all the time look at a display showing the frequency
> > response. After some time I got a feeling for what was right and
> wrong,
> > getting more and more acquainted with the mike. I could get rid of
> peaks and
> > gaps and get a reasonably good polar diagram. I have no idea how
> the stereo
> > DATmic connects to Sten's math, however. What happens when such a
> mic is
> > placed inside a parabol's focus?
> > Still, I claim, the errors, which may follow, are lesser that "the
> errors"
> > caused by reality.
Your methods worked, the Telinga Stereo Dat mic is excellant and
useful.
=20
> Sten briefly mentions the effect of mic design on his studies in his
> text, but, as has been pointed out, does not account for the mics in
> his
> math. He gets us as far as what's happening at a focus point, the mic
>
> effects would require a lot more study.
>
> It's fairly easy to understand what might happen with different mic
> diaphragm diameters, much more a problem when multiple diaphragms are
>
> picking up sound modified by a PZM design, and those are at varying
> distances from the dish. And to top it off the end user adjusts the
> mic
> location by ear. Doing it by practical experiment is probably the
> only
> way. I can easily understand why this took some time.
Backman's paper from the AES used 1/2" omni, figure of eight
(unspecified diameter capsule), and a 6mm cardiod in their
measurements. In the end, their conclusion was that the noise floor of
the mic is usually the most important criterion in practical
applications. Lower noise floors are usually in omni mics, pressure
transducers.
> On a practical level understanding the behavior of that focus and
> it's
> effect on mics is very useful. I often use fine aiming control to
> mask
> out unwanted callers who are on a different frequency but in line
> with
> my target.
>
> > The Twin Science is very conventional, especially when using the
> omni
> > looking out of the dish. I'm sure that Sten's math explains very
> well the
> > Telinga with the omni working. When the directional mike is used,
> facing
> > inwards into the dish, one can expect that some of the sounds
> reflected by
> > the dish also hits the back of the membrane and cancels some of the
> gain.
> > Perhaps you also get other effects. As I mentioned above, it is
> almost
> > impossible to make accurate measurements under realistic
> conditions, outside.
> >
> > I have thought of, though, to try to make a curve over "average
> errors
> > caused by distance" - as the Telinga picks it up, and then build a
> filter to
> > be used at replay. A common filter, like 6db/octave doesn't work.
> It should,
> > theoretically, but it doesn't sound good.
>
> Actually, because only part of the frequency range of the dish
> follows
> the 6dB/octave rule, it's not surprising that applying a filter
> across
> all frequencies at that level would not work well.
The parabolic reflector will have a 6db per octave increase in gain
with frequency increase where the focal distance is of the same order
as the wavelength, that is, where the Focal length/wavelength > 1
(Wahlstrom).=20
Where focal length/wavelength <1/64, gain is 0db (Wahlstrom).=20
Between these 2 regions, where 1/64 < focal length/wavelength < 1 is
the most variable gain (most ripple freq. response) for a parabolic mic
due to the interference of the direct sound to the mic element, with
the reflected sound off the dish (Wahlstrom, Backman). It is this
range of frequencies that are most affected by the focal length
compared to the depth of the dish (Wahlstrom, Backman). The flatter
the dish, where depth of reflector/focal length < 1, have more
pronounced freq. response and gain variations.
As noted before, the deviation from 6db per octave at higher
frequencies will be caused by a mic diaphragm larger than the 'globe'
of focus, unable to be fully illuminated by the focused waves.
> > Please go on with the present discussion!! Here is a combination of
> skills
> > in math added to overwhelming practical experience! As a matter of
> facts, I
> > think this is the first time in history that such a discussion has
> taken
This is a wonderful thing. Thank you all.
bret
__________________________________
Do you Yahoo!?
Yahoo! Mail SpamGuard - Read only the mail you want.
http://antispam.yahoo.com/tools
________________________________________________________________________
________________________________________________________________________
|