[coffeedj]

I have a problem which this group may be able to shed light on.

I'm currently working on the design of a medical product which could best be
described as an electronic stethoscope on steroids.

One of the problems to solve is how to accurately profile the system
response at each point--which includes microphones, low level amplifiers,
13.2mm speakers for reproduction, acoustic tubes, sampling engines, DSP
filters, displays, and so on.

Profiling the electronic components is simple, but creating corrected driver
response for the speaker which puts the test signal into the microphone is
the question. I'm using a heart sound simulator from Andries to drive test
tones, heart sound files, and sweeps. To calibrate the Andries speaker I
used an Exotech 407764 Sound Level Meter data logger in A weighted mode and
coupled the input of the meter directly through a sealed stethoscope bell to
the speaker. This is how the system is connected in the product, and I
duplicated the conditions. To created the proper speaker correction factor,
multiple averaged frequency sweeps were done. When I applied the correction
factor to the system the results were believable as they duplicated the
manufacturers spec sheet for the microphone (Panasonic Electret) response.

However, I'm not yet completely convinced it is correct. The manufacturers
spec sheet is for the mic measured in an anechoic chamber, and I am sealing
it in a bell. Also, the sound meter is being used in a sealed manner as
well--and repeated emails to Extech about my manner of using their equipment
are ignored. I suspect that the sealed method of measuring is OK since
their factory calibration method uses a sealed 94db 1KHz tone generator in A
weighted mode. (although at 1KHz A and C weighting are essentially the
same.)

Almost all sound measurements are done in open air or anechoic chambers--and
I am sealing everthing in a tiny volume.

Any experience or wisdom from the group would be appreciated.

..

[Patrick Turner]

coffeedj wrote:

I have a problem which this group may be able to shed light on.

I'm currently working on the design of a medical product which could best be
described as an electronic stethoscope on steroids.

One of the problems to solve is how to accurately profile the system
response at each point--which includes microphones, low level amplifiers,
13.2mm speakers for reproduction, acoustic tubes, sampling engines, DSP
filters, displays, and so on.

Profiling the electronic components is simple, but creating corrected driver
response for the speaker which puts the test signal into the microphone is
the question. I'm using a heart sound simulator from Andries to drive test
tones, heart sound files, and sweeps. To calibrate the Andries speaker I
used an Exotech 407764 Sound Level Meter data logger in A weighted mode and
coupled the input of the meter directly through a sealed stethoscope bell to
the speaker. This is how the system is connected in the product, and I
duplicated the conditions. To created the proper speaker correction factor,
multiple averaged frequency sweeps were done. When I applied the correction
factor to the system the results were believable as they duplicated the
manufacturers spec sheet for the microphone (Panasonic Electret) response.

However, I'm not yet completely convinced it is correct. The manufacturers
spec sheet is for the mic measured in an anechoic chamber, and I am sealing
it in a bell. Also, the sound meter is being used in a sealed manner as
well--and repeated emails to Extech about my manner of using their equipment
are ignored. I suspect that the sealed method of measuring is OK since
their factory calibration method uses a sealed 94db 1KHz tone generator in A
weighted mode. (although at 1KHz A and C weighting are essentially the
same.)

Almost all sound measurements are done in open air or anechoic chambers--and
I am sealing everthing in a tiny volume.

Any experience or wisdom from the group would be appreciated.

I am not sure what the hell you are actually doing because your long
explanation
doesn't say what you are doing in a nutshell.

But I use pink noise in rooms in which sound systems are set up for
test.

Testing in a shoe box would perhaps restrict the range of frequencies
possible,
and inside a tennis ball the range would even be less.

Using sine waves whose frequency is gradually adjusted gives a totally
meaningless
indication of the level of audio ebergies liberated into room from a
transducer
because of reinforcing and nulling resonances, hence the use of pink
noise
and filtering of bands of frequencies.

Patrick Turner.

.

[Phil Allison]

"coffeedj"


Almost all sound measurements are done in open air or anechoic
chambers--and I am sealing everthing in a tiny volume.


** I got news for you - Mr Caffeine Addict.

Microphones do NOT respond to the SPL in a room or an open space !!!!

An omni mic ( as used for SPL meters and speaker testing ) responds to the
actual SPL impinging on its diaphragm at any instant - the output signal
results only from oscillating air pressure in that tiny region of space
adjacent to the diaphragm.

Maybe that answers your ambiguous Q and maybe not.

It's your problem to get your head around the physics.




........ Phil

[Chris Hornbeck]

On Fri, 27 Apr 2007 15:26:37 -0700, "coffeedj"
wrote:

Profiling the electronic components is simple, but creating corrected driver
response for the speaker which puts the test signal into the microphone is
the question. I'm using a heart sound simulator from Andries to drive test
tones, heart sound files, and sweeps. To calibrate the Andries speaker I
used an Exotech 407764 Sound Level Meter data logger in A weighted mode and
coupled the input of the meter directly through a sealed stethoscope bell to
the speaker. This is how the system is connected in the product, and I
duplicated the conditions.

A question arises: Are the results intended to be slavishly
fidelic or to be medically useful?

In neither case does anyone here have *any* competence.

Ignore anything you read here. ("Including this." - Mark
Twain)

Seriously, you need better council. This group is, within a medical
context) for bull****. Not for medical instruments.

All good fortune,

Chris Hornbeck
"But of course, when you need it, it ain't headroom any more."
- Don Pearce

[coffeedj]

Well, from the comments recieved it looks like I'm on my own. Tried to keep
it short enough to be readable, but evidently more information would have
helped some.

The size constraints convert this to a system that measures pressure, not
sound per se. That is how standard stethoscopes work. A doctor listening
for an S3 gallop (very bad news indeed) at 30Hz can't hear the sound--he
must feel the puff of air in his ear canal. None of the standard pink
noise/white noise tests are useful. Unlike music--stethoscopes are really
listening for only a few very narrow bands of frequencies.

Fidelity is never the concern for medical products--only diagnostic yield
matters. And that has to be proven in clinical trials.

Thanks anyway.

"coffeedj" wrote in message
. ..
I have a problem which this group may be able to shed light on.

I'm currently working on the design of a medical product which could best
be described as an electronic stethoscope on steroids.

One of the problems to solve is how to accurately profile the system
response at each point--which includes microphones, low level amplifiers,
13.2mm speakers for reproduction, acoustic tubes, sampling engines, DSP
filters, displays, and so on.

Profiling the electronic components is simple, but creating corrected
driver response for the speaker which puts the test signal into the
microphone is the question. I'm using a heart sound simulator from
Andries to drive test tones, heart sound files, and sweeps. To calibrate
the Andries speaker I used an Exotech 407764 Sound Level Meter data logger
in A weighted mode and coupled the input of the meter directly through a
sealed stethoscope bell to the speaker. This is how the system is
connected in the product, and I duplicated the conditions. To created the
proper speaker correction factor, multiple averaged frequency sweeps were
done. When I applied the correction factor to the system the results were
believable as they duplicated the manufacturers spec sheet for the
microphone (Panasonic Electret) response.

However, I'm not yet completely convinced it is correct. The
manufacturers spec sheet is for the mic measured in an anechoic chamber,
and I am sealing it in a bell. Also, the sound meter is being used in a
sealed manner as well--and repeated emails to Extech about my manner of
using their equipment are ignored. I suspect that the sealed method of
measuring is OK since their factory calibration method uses a sealed 94db
1KHz tone generator in A weighted mode. (although at 1KHz A and C
weighting are essentially the same.)

Almost all sound measurements are done in open air or anechoic
chambers--and I am sealing everthing in a tiny volume.

Any experience or wisdom from the group would be appreciated.

.

[coffeedj]

An omni mic ( as used for SPL meters and speaker testing ) responds to the
actual SPL impinging on its diaphragm at any instant - the output signal
results only from oscillating air pressure in that tiny region of space
adjacent to the diaphragm.

That is correct. Turns out the biggest physics problem to solve is the
relationshipof the mass of vibrating air with respect to the diaphragm
(weight and material) used. Littman used this concept to create a "tunable"
stethoscope back in the '40s.

[Phil Allison]

"coffeedj"

An omni mic ( as used for SPL meters and speaker testing ) responds to
the
actual SPL impinging on its diaphragm at any instant - the output
signal results only from oscillating air pressure in that tiny region of
space adjacent to the diaphragm.

That is correct. Turns out the biggest physics problem to solve is the
relationshipof the mass of vibrating air with respect to the diaphragm
(weight and material) used.


** The diaphragms of condenser / electret mics is so thin it weighs less
than a mm or two of air each side.

Total non issue and irrelevant to my post.



........ Phil

[Patrick Turner]

coffeedj wrote:

Well, from the comments recieved it looks like I'm on my own. Tried to keep
it short enough to be readable, but evidently more information would have
helped some.

The size constraints convert this to a system that measures pressure, not
sound per se. That is how standard stethoscopes work. A doctor listening
for an S3 gallop (very bad news indeed) at 30Hz can't hear the sound--he
must feel the puff of air in his ear canal. None of the standard pink
noise/white noise tests are useful. Unlike music--stethoscopes are really
listening for only a few very narrow bands of frequencies.

Well OK, the doc listens and feels to understand. The best docs hear and
feel more than the worst,
and kill less patients.

But say you know the bandwidth you need to know about, then why not have
a spectrum analyser
which gives a bar graph display of all F present, and then all docs
don't need to be so great,
and can see the display beside a bed without using a stethoscope because
there is a mic on the patient's chest.


BTW, sure, there is a lotta BS on any news group, with a large amount of
immaturity
displayed by men who should know better.

But all gatherings of men display this aspect of human behaviour from
Parliament
to medical conferences to employer groups or trade union meetings.

So when at first its seems we are of no help, try again, expect the
noise, but ignore it, OK,
and persuade ppl to see your problem. Perhaps someone with a small
amount of intelligence
will say something useful.

This group is supposed to be about tube craft, and once tubes were the
basis of electronics in
medical diagnostic work, but not any more, and solid state and computer
gear dominates
the entire medical gear scene, and nobody here has any experience with
its special design requirements.

Acoustic phenomena is little known amoung us. Most ppl here stay to
learn to build a power amp
at age 50, then leave, never to be seen or heard of again; they don't
have a deep interest
in electronics and the related fields, such as speaker design and
acoustic, or microphone use.

Some universities have private discussion groups operate away from the
inane behaviours of unqualified
BS artists, and crazy/difficult ppl who abound in public access
unmoderated Usernet groups.

If you want answers, look further afield.

Grandpa told me, "If yer don't look, yer won't know!"

Patrick Turner.



Fidelity is never the concern for medical products--only diagnostic yield
matters. And that has to be proven in clinical trials.

Thanks anyway.

"coffeedj" wrote in message
. ..
I have a problem which this group may be able to shed light on.

I'm currently working on the design of a medical product which could best
be described as an electronic stethoscope on steroids.

One of the problems to solve is how to accurately profile the system
response at each point--which includes microphones, low level amplifiers,
13.2mm speakers for reproduction, acoustic tubes, sampling engines, DSP
filters, displays, and so on.

Profiling the electronic components is simple, but creating corrected
driver response for the speaker which puts the test signal into the
microphone is the question. I'm using a heart sound simulator from
Andries to drive test tones, heart sound files, and sweeps. To calibrate
the Andries speaker I used an Exotech 407764 Sound Level Meter data logger
in A weighted mode and coupled the input of the meter directly through a
sealed stethoscope bell to the speaker. This is how the system is
connected in the product, and I duplicated the conditions. To created the
proper speaker correction factor, multiple averaged frequency sweeps were
done. When I applied the correction factor to the system the results were
believable as they duplicated the manufacturers spec sheet for the
microphone (Panasonic Electret) response.

However, I'm not yet completely convinced it is correct. The
manufacturers spec sheet is for the mic measured in an anechoic chamber,
and I am sealing it in a bell. Also, the sound meter is being used in a
sealed manner as well--and repeated emails to Extech about my manner of
using their equipment are ignored. I suspect that the sealed method of
measuring is OK since their factory calibration method uses a sealed 94db
1KHz tone generator in A weighted mode. (although at 1KHz A and C
weighting are essentially the same.)

Almost all sound measurements are done in open air or anechoic
chambers--and I am sealing everthing in a tiny volume.

Any experience or wisdom from the group would be appreciated.

.