room response exitation source

[Magnus Danielson]

From: Martin Czech <czech at Micronas.Com>
Subject: room response exitation source
Date: Mon, 3 Apr 2000 10:20:51 +0200 (MET DST)

> How can one obtain a room response, or more general the transfer function
> of an acoustic environment?
> 
> Balloons and starting pistols are easy to carry with, and usually have
> loud enough output. Balloons may have considerable different pulse forms
> from one explosion to the next.
> 
> I think that the basic characteristic of a room still comes out,
> even if the impulse response is not deconvoluted (no, I know that!),
> or statistically varied.
> 
> I think I get myself a new gun (starting pistol, also good for signal
> ammunition).
> 
> Professional people use spark gaps, or speakers with noise sequences.
> 
> If such a noise sequence is generated in a digital fashion (famous shift
> register), why use a linear amp?

1) Still a linear system - easier compensation
2) Output level control
3) Analog amp isn't terribly expensive anyways

>  I could think of an bass speaker chassis (maybe also tweeter) with a
>  switching power stage, i.e. a mosfet pair following the logic level of
>  the shift register. All running from 12V battery (automotive type). Of
>  course the dc bias has to be kept away from the speaker via caps.
> 
> Is there any problem with such an approach?
> 
> -no loudness control, always 100%
> -perhaps higher frequencies cause coil trouble
> -perhaps higher frequencies have inter modulation
> -difficulties with recording the shift register output (alias)

You would certainly like to have a amplitude/effect control (loudness is not
a good word to use here).

Also, you do want to care about what energies you run into a speaker. No
speaker element is really capable of handling frequencies as well as you like.

> You have to record the electronic shift register output, because
> you need the reference signal for deconvolution (shift register frequency
> and dat-recorder frequency will drift).

Right. You would not only like to have the frequency of the sampling and
generator equal, but you would also like to know the phase of the generator
noise in order to make good cross-correlation.

> If this works, this would be a portable excitation source,
> the thing that Magnus asked for...

No, I rather proposed it. I proposed the creation of a device that also did
the cross-correlation (deconvolution in your terms) on the spot.

Cheers,
Magnus