[sdiy] About that spherical impulse generator...

[Czech Martin]

I think the discussed approaches will have problems.

Starter pistols, sparc gaps etc have problems:
Every event sounds different. Air turbulence,
moisture, pressure, light, there are many factors
that influence the impulse. There are tons of literature
about spark gaps, this is by no means a simple thing.

But there is a much more critical thing:

To have good S/N you need a loud impulse.
This is a problem if the microphone is relatively close,
because it will lead to massive distortion.
The air compression will be non linear anyway.

Perhaps you can use the method for situations where
the diffuse field dominates, i.e. microphone very far away,
so first wave and reverb have about equal level.


Because of these reasons, multiple pulses are used.
To avoid resonances, the pulses are random distributed.
There is another name for this: noise.
In order to predictable energy, pseudo noise with a flat
spectrum is used, i.e. cyclical noise.
Do to the cyclical property, one can sum up multiple
runs to have better S/N. Also, cyclical deconvolution
can be used. This method is very susceptible to distortion.


Anyway: the sound source must show equal level from 20-20000Hz.
A piezo will have no output below -say- 1000 Hz.
However you will use it, you get no information below that.
Remember that a single ideal pulse has a completely flat spectrum
from DC to infinity. You can not play this with a tweeter only.

I think that only a wide band dynamic chassis can do this.

m.c.