Acoustic Array Hardware?

Hi All,

As some of you might know, I'm generally a theory/SW kinda guy.  I'm
quite interested in sensor array processing, in particular acoustic
sensors (standard microphones or ultrasonics).

I was wondering if anyone out there can suggest a rig that is capable
of acquiring 8 to 16 channels of audio data, from 8 to 16 microphones
(ultrasonic receivers), with a bandwidth of at least 8kHz?

That's only a data rate of 64 to 128kHz, so I can't see that it's a
big problem.  I'm interested in whether anyone sells the entire rig,
rather than just the components.

I'm currently mainly interested in the acquisition only; doing any
processing is not (currently) necessary, but it might be in the future.

There are some people like Acoustic Magic that have an 8-sensor array,
but I gather from their technical dudes that it's hard to get access
to the individual sensor signals.

Any pointers gratefully received!

Ciao,

Peter K.



-- 
Peter J. Kootsookos

"Na, na na na na na na, na na na na"
- 'Hey Jude', Lennon/McCartney

"Peter J. Kootsookos" wrote:
> 
> Hi All,
> 
> As some of you might know, I'm generally a theory/SW kinda guy.  I'm
> quite interested in sensor array processing, in particular acoustic
> sensors (standard microphones or ultrasonics).
> 
> I was wondering if anyone out there can suggest a rig that is capable
> of acquiring 8 to 16 channels of audio data, from 8 to 16 microphones
> (ultrasonic receivers), with a bandwidth of at least 8kHz?
> 
> That's only a data rate of 64 to 128kHz, so I can't see that it's a
> big problem.  I'm interested in whether anyone sells the entire rig,
> rather than just the components.
> 
> I'm currently mainly interested in the acquisition only; doing any
> processing is not (currently) necessary, but it might be in the future.
> 
> There are some people like Acoustic Magic that have an 8-sensor array,
> but I gather from their technical dudes that it's hard to get access
> to the individual sensor signals.
> 
> Any pointers gratefully received!

Hi Peter,

We have a card that might well do what you're after:
http://www.bittware.com/products/PMC/AUPM/aupm_desc.stm

You'd need two to get 16 channels though.
It includes two ADSP21065L's, so you could do some processing later if
you wanted to.

-- 
Jim Thomas            Principal Applications Engineer  Bittware, Inc
jthomas@bittware.com  http://www.bittware.com          (703) 779-7770
Visualize whirled peas.

Look at the "Gina" and "Laaya" by a company called Echo.  They are
professional audio systems, not expensive, and work quite well.


Maurice Givens


p.kootsookos@remove.ieee.org (Peter J. Kootsookos) wrote in message news:<s68of02gam0.fsf@mango.itee.uq.edu.au>...
> Hi All,
> 
> As some of you might know, I'm generally a theory/SW kinda guy.  I'm
> quite interested in sensor array processing, in particular acoustic
> sensors (standard microphones or ultrasonics).
> 
> I was wondering if anyone out there can suggest a rig that is capable
> of acquiring 8 to 16 channels of audio data, from 8 to 16 microphones
> (ultrasonic receivers), with a bandwidth of at least 8kHz?
> 
> That's only a data rate of 64 to 128kHz, so I can't see that it's a
> big problem.  I'm interested in whether anyone sells the entire rig,
> rather than just the components.
> 
> I'm currently mainly interested in the acquisition only; doing any
> processing is not (currently) necessary, but it might be in the future.
> 
> There are some people like Acoustic Magic that have an 8-sensor array,
> but I gather from their technical dudes that it's hard to get access
> to the individual sensor signals.
> 
> Any pointers gratefully received!
> 
> Ciao,
> 
> Peter K.

I forgot to add that Gina or Laya coupled with Cool Edit make a very
nice audio capture/editing/mixing/generation system (in fact, I used
Gina and Cool Edit with a Matlab program to perform the complete ITU
G.168 Echo Canceller test suite).

Maurice Givens


p.kootsookos@remove.ieee.org (Peter J. Kootsookos) wrote in message news:<s68of02gam0.fsf@mango.itee.uq.edu.au>...
> Hi All,
> 
> As some of you might know, I'm generally a theory/SW kinda guy.  I'm
> quite interested in sensor array processing, in particular acoustic
> sensors (standard microphones or ultrasonics).
> 
> I was wondering if anyone out there can suggest a rig that is capable
> of acquiring 8 to 16 channels of audio data, from 8 to 16 microphones
> (ultrasonic receivers), with a bandwidth of at least 8kHz?
> 
> That's only a data rate of 64 to 128kHz, so I can't see that it's a
> big problem.  I'm interested in whether anyone sells the entire rig,
> rather than just the components.
> 
> I'm currently mainly interested in the acquisition only; doing any
> processing is not (currently) necessary, but it might be in the future.
> 
> There are some people like Acoustic Magic that have an 8-sensor array,
> but I gather from their technical dudes that it's hard to get access
> to the individual sensor signals.
> 
> Any pointers gratefully received!
> 
> Ciao,
> 
> Peter K.

Peter,

You might want to check out Sonorus (http://www.sonorus.com) 8-channel
24-bit audio A/D boxes (they also have similar D/A boxes). They have a
fairly high sample rate (suitable for music studios), no additional input
filtering required. They connect by optical cable to a card that plugs into
your computer. I can't speak to suitability of channel matching, sampling
synchronization, interface, ... for your application, but it might be worth
a look.

Dirk

Dirk A. Bell
DSP Consultant


"Peter J. Kootsookos" <p.kootsookos@remove.ieee.org> wrote in message
news:s68of02gam0.fsf@mango.itee.uq.edu.au...
> Hi All,
>
> As some of you might know, I'm generally a theory/SW kinda guy.  I'm
> quite interested in sensor array processing, in particular acoustic
> sensors (standard microphones or ultrasonics).
>
> I was wondering if anyone out there can suggest a rig that is capable
> of acquiring 8 to 16 channels of audio data, from 8 to 16 microphones
> (ultrasonic receivers), with a bandwidth of at least 8kHz?
>
> That's only a data rate of 64 to 128kHz, so I can't see that it's a
> big problem.  I'm interested in whether anyone sells the entire rig,
> rather than just the components.
>
> I'm currently mainly interested in the acquisition only; doing any
> processing is not (currently) necessary, but it might be in the future.
>
> There are some people like Acoustic Magic that have an 8-sensor array,
> but I gather from their technical dudes that it's hard to get access
> to the individual sensor signals.
>
> Any pointers gratefully received!
>
> Ciao,
>
> Peter K.
>
>
>
> --
> Peter J. Kootsookos
>
> "Na, na na na na na na, na na na na"
> - 'Hey Jude', Lennon/McCartney

Peter J. Kootsookos wrote:
> Hi All,
> 
> As some of you might know, I'm generally a theory/SW kinda guy.  I'm
> quite interested in sensor array processing, in particular acoustic
> sensors (standard microphones or ultrasonics).
> 
> I was wondering if anyone out there can suggest a rig that is capable
> of acquiring 8 to 16 channels of audio data, from 8 to 16 microphones
> (ultrasonic receivers), with a bandwidth of at least 8kHz?
> 
> That's only a data rate of 64 to 128kHz, so I can't see that it's a
> big problem.  I'm interested in whether anyone sells the entire rig,
> rather than just the components.
> 
> I'm currently mainly interested in the acquisition only; doing any
> processing is not (currently) necessary, but it might be in the future.
> 
> There are some people like Acoustic Magic that have an 8-sensor array,
> but I gather from their technical dudes that it's hard to get access
> to the individual sensor signals.
> 
> Any pointers gratefully received!
> 
> Ciao,
> 
> Peter K.
> 
> 
> 
Not meaning  to be overly judgmental, but IMHO this is a piss poor spec 
;)  If anyone tries to sell you anything with this paucity of detail, I 
would be suspicious.

You might want to give some thought to dynamic range.  Real sensors also 
get thunked, so things like saturation recovery time can be important.

Back in the old days, a lot stuff was handled in analog, like AGC. 
Converters have gotten relatively cheaper so things like channel to 
channel phase matching tends to be handled on the digital side after 
oversampling.

  You need to look at how they do sampling over channels.  A lot of 
boards will sequentially sample over the channels.  Most folks really 
want simultaneous sampling but the sequential stuff can work if you know 
the times well, i.e. you can compensate in post processing.

Thre remains much that is done in  analog.  Most people have a natural 
ability to create oscillators when they don't want to.  This is an area 
that is hard to spec before hand. You typically have to physically move 
things around to make them work acceptably.  Isolation issues in biomed 
applications don't make things easy either.

If you plan on using the data in a way that you can trace back to 
physical quantities, you might have to do an actual physical 
calibration. If you buy, the seller should not give you a strange look 
if you ask for a calibration curve and not some average curve, but "the 
actual" curve.  A band width of 8KHz is narrow band to some folks and BB 
to others.  If you fall in the BB category, the sensor is not likely to 
have a nice flat response.

If you can afford it, get electronics that is shock and temperature 
rated, even if you think you don't need it.  You have to consider the 
value of your own time.  You might have spent a lot to just to get the 
point in the field where you can acquire your data. If you then have a 
10 cent OP amp fail on you so that the trip becomes a waste, it can be 
highly frustrating. Your colleagues will think you're a pud. If you 
don't want to do that then buy a spare system or two and bring them along.

Stan Pawlukiewicz <stanp@nospam_mitre.org> writes:

> Not meaning  to be overly judgmental, but IMHO this is a piss poor spec 
> ;)  If anyone tries to sell you anything with this paucity of detail, I 
> would be suspicious.

Yup, I haven't put together a HW spec for a while, so I'm just
starting by gathering information.

> You might want to give some thought to dynamic range.  Real sensors also 
> get thunked, so things like saturation recovery time can be important.

OK, thanks.

> Back in the old days, a lot stuff was handled in analog, like AGC. 
> Converters have gotten relatively cheaper so things like channel to 
> channel phase matching tends to be handled on the digital side after 
> oversampling.

Yup.

>   You need to look at how they do sampling over channels.  A lot of 
> boards will sequentially sample over the channels.  Most folks really 
> want simultaneous sampling but the sequential stuff can work if you know 
> the times well, i.e. you can compensate in post processing.

Yes, I've seen quite a bit of stuff on compensating for sequential
sampling of the channels as opposed to synchronised sampling.

> Thre remains much that is done in  analog.  Most people have a natural 
> ability to create oscillators when they don't want to.  

"If you want to design an amplifier, design an oscillator first.  If
you want to design an oscillator, design an amplifier first."

> This is an area 
> that is hard to spec before hand. You typically have to physically move 
> things around to make them work acceptably.  Isolation issues in biomed 
> applications don't make things easy either.

Yes, I've been talking with some Biomed dudes (mainly MRI) recently;
they have some entertaining problems.

> If you plan on using the data in a way that you can trace back to 
> physical quantities, you might have to do an actual physical 
> calibration. If you buy, the seller should not give you a strange look 
> if you ask for a calibration curve and not some average curve, but "the 
> actual" curve.  

Thanks, that sounds like good advice.

> A band width of 8KHz is narrow band to some folks and BB 
> to others.  If you fall in the BB category, the sensor is not likely to 
> have a nice flat response.

I'm still "sounding" out the sensors, I suspect it will be relatively
broadband, though.

> If you can afford it, get electronics that is shock and temperature 
> rated, even if you think you don't need it.  You have to consider the 
> value of your own time.  You might have spent a lot to just to get the 
> point in the field where you can acquire your data. If you then have a 
> 10 cent OP amp fail on you so that the trip becomes a waste, it can be 
> highly frustrating. Your colleagues will think you're a pud. If you 
> don't want to do that then buy a spare system or two and bring them along.

Yes, a former colleague of mine had a submarine and a frigate tied up
for three days taking towed array measurements.  For some reason, this
person thought they were an engineer.  It turns out that the three
days worth of data they had been collecting was all `0' due to a break
in the connector from the sensors to the acquisition gear.  They had
neglected to do a quality check on the data at all while at sea and
only noticed when they were off the boat.

I wasn't there when they found out, but I heard that there were some
choice words said by someone who is usually immaculate in their clean
language.

Thanks again for the (long) useful post!

Ciao,

Peter K.

-- 
Peter J. Kootsookos

"Na, na na na na na na, na na na na"
- 'Hey Jude', Lennon/McCartney

Jim Thomas <jthomas@bittware.com> writes:

> We have a card that might well do what you're after:
> http://www.bittware.com/products/PMC/AUPM/aupm_desc.stm
> 
> You'd need two to get 16 channels though.
> It includes two ADSP21065L's, so you could do some processing later if
> you wanted to.

Thanks, Jim, I'll include it in the required reading before I figure
out exactly what's needed.

Ciao,

Peter K.

-- 
Peter J. Kootsookos

"Na, na na na na na na, na na na na"
- 'Hey Jude', Lennon/McCartney

maurice.givens@ieee.org (Maurice Givens) writes:

> Look at the "Gina" and "Laaya" by a company called Echo.  They are
> professional audio systems, not expensive, and work quite well.

Hi Maurice,

That looks very sweet.  Thanks for the pointers; I've also added it to
the recommended reading (before I actually narrow down what I'm after).

Ciao,

Peter K.

-- 
Peter J. Kootsookos

"Na, na na na na na na, na na na na"
- 'Hey Jude', Lennon/McCartney

maurice.givens@ieee.org (Maurice Givens) writes:

> I forgot to add that Gina or Laya coupled with Cool Edit make a very
> nice audio capture/editing/mixing/generation system (in fact, I used
> Gina and Cool Edit with a Matlab program to perform the complete ITU
> G.168 Echo Canceller test suite).

Cool!  Thanks!

Ciao,

Peter K.

-- 
Peter J. Kootsookos

"Na, na na na na na na, na na na na"
- 'Hey Jude', Lennon/McCartney