On Wed, 25 Jun 2008 21:15:10 -0700, "Jerry Steiger"
wrote:
It's been quite a while since I stopped following this group. I stopped
being the recorder for the local youth symphony and ran out of time to keep
up with audio matters, bu now I want to build a small audio test chamber for
work. I'm hoping the good folks here can help me a bit.
We make rugged hand-held computers for outdoor use. Our products are all
water proof. One of the challenges we always have is getting good speaker
output and microphone input. Good for the speaker means easily heard and
understood in somewhat noisy environments, meaning we need something like
80-90 dB SPL output. Good for the microphone means that we can record easily
intelligible voice input. Both are at a distance of a foot or two or three
with a frequency range of about 300 to 3400 Hz.
In the past we have tested speakers and microphones very simplistically,
using a sound level meter to monitor speaker output in a normal office
environment, but we would like to get a little more professional about it.
We had a summer student do some preliminary work, so we now have a reference
speaker (Event Tuned Reference 8) and microphone (Audio-Technica AT2020), an
Alesis RA 150 amplifier, an Edirol UA125 USB Audio Capture System, a small
box to link them all together, and some Matlab programs on the computer to
drive the system and plot the data.
I was thinking that it would be best to have a small enclosure in which I
could mount either a speaker (either the reference speaker or a test speaker
or one of our hand-held units acting as a speaker) and a microphone (either
the reference microphone or a test microphone or one of our hand-held units
acting as a microphone). I would line the inside of the enclosure with sound
absorbing panels so that we would be measuring mostly the direct response of
the speaker or the microphone, since that is what we would mostly be getting
in our outdoor environment. Am I on the right track? Is this something
reasonably easy for me to do?
How big should the chamber be? I was thinking of making the inside
dimensions 25"x59"x96". I chose those dimensions so I wouldn't have any
double or triple resonances but could use standard sizes of sound absorbing
panels. How much effort should I put into making the side dimensions to
spread the resonances? How close together can resonances be before they
become too problematic?
I would line the walls with 4" thick 24"x48" cotton panels filling most of
the sides and ends and 12"x12" foam corners on two of the long edges to
cover the gaps. I was thinking that I would put a hole in the long side
roughly 3 feet from one end that I could put removable plates in that hold
the reference speaker or the speaker under test. I would put another hole
and plate one meter from that hole on which I could mount my reference
microphone or the microphone under test. That would put my microphone
roughly 2 feet from the other end. Is that a reasonable way to position the
pieces? Should I fill the space behind the speaker and microphone with more
sound absorbing material? Would stacking sections of the pads work? Would
stacking the pads on the sides to leave just a narrow space a bit bigger
than our hand-held be worthwhile?
Would 1" MDF be stiff enough for the walls? Would I need to stiffen it with
some outside ribs? I can readily calculate the resonant modes of rectangular
plates. Do I need to worry much about the resonances? How many modes would I
need to look at?
Will the cotton panels (B.A.P. Bass Buster from www.acousticalsurfaces.com)
absorb enough sound to make this very roughly like an anechoic chamber at
the frequencies of interest? I don't need super precision, but I would like
to have measurements that correlate fairly well with what the user would
hear in actual use.
You're going to read much more useful and generally more positive
comments than my current one, so please forgive me if I jump into
the cold water with some of the "gotcha's":
First, just as a matter of good old Science, you'll need *something*
to act as a reference, if you really want to investigate.
Could be a reference speaker, or could be a reference mic, but neither
is cheap - and worse, neither is a "primary" reference. Without a
reference, everything you measure is only relative to ... something
else. Does that make sense?
Primary references tend to be impulse noise generators: pop a balloon
(not actually too wonderful), starter pistol (get your insurance in
order first), or Don Pierce's spark gap with camera flash discharge
(see the r.a.p thread - top recommendation, very elegant).
Second, if your box is dead enough, you don't care about dimensions.
If it isn't dead enough, you don't care about its results.
Third and lastest, above 300 Hz, measurements today are usually done
in a live room with a time window. Because impulses are challenging
dynamically - the ideal is an infint(insert some appropriate vowel)lly
short time of infinit(same damn vowel)lly large amplitude - judgement
calls must be made. Hence the strong recommendation to research
Don Pierce's recent posts, here on r.a.p IIRC.
Commercial implementations allow for a noise source with the
mathematical equivalent of an impulse, called an "MLS" (a
"maximum length sequence", not so helpfully named...) The
resulting hardware devices are generically called Melissa,
after my sister.
Fourth and post-mortem, it's voice range, so why do you care? Arf!
All the best fortune,
Chris Hornbeck