room responses etc.

[Magnus Danielson]

From: Martin Czech <czech at Micronas.Com>
Subject: Re: room responses etc.
Date: Wed, 9 Aug 2000 15:35:18 +0200 (MET DST)

> Yes, the ground reflections...  I observed quite long impulse responses
> so far, it would mean quite a large construction...  A tower would be
> nice, but the construction elements would reflect, too!  Perhaps a dive
> tower in a swimming pool?

Well, you don't really need to have a longer responce than what makes the
tail of the speaker/mike responce to go quiet. Now, there are things to
consider when doing this. The mike's impulse responce will be much quicker, so
it will be preatty dead when you have good distance, but the speaker and the
cross-over filters contribute to this lowfrequency aspect. Another things is
that what still is "alive" after some time is really just a subset of the full
set of poles/zeros, so you should be able to do an rather accurate compensation
of these properties so that you end up with a quieter responce.

One trick is to measure up the speaker element by element, both the free-space
responce as well as the impedance. The impedance measurements let's you tap
into the poles of the speaker. The surroundings is only contributing zeros in
a free-field measurement anyway. Now, if you are able to measure up the poles,
you just take that and compensate it away and you can have a much quiter
responce to work with.

Swimming pool jump-towers are nice, since you get out from the actual tower
and has little direct reflections from that. However, watch out so your
impulse responce doesn't become too "wet" ;)

> A dear, getting arrested again. My first time beeing arrested was
> because of list members suggesting the use of a cap pistol in the
> Cologne Cathedral.

WHAT!??!? You have been arrested? And for doing measurements? Hmmm... you
really should have taken a diffrent set of tools with you....

> Now I fear getting arrested when climbing the fence at my local public
> swimming pool...

It is really starting to sound like a new form of protestmovement or something,
working against bad sound or something...

You seems to really go a long way to get your audio stuff to the top notch! ;D

I have taken many impulse responces, but I have not yeat been arrested!

> Perhaps I should not try to eliminate the reflections at all?  What if I
> LAY the speaker on ground, so that the chassis main axis points towards
> heaven? Say on a great agricultural area, no buildings around. If I
> build a wooden "collar" for the speaker, basically a 2x2m area with a
> hole inside to fit the speaker just on the surface, I'd have a controlled
> reflection from that collar but certainly not from heaven.

Well, this sounds all very nice... but putting the speaker on an infinit baffle
or something is not going to help you, you want to know how your speaker
behaves in a free-space environment. Mounting the speaker in a large baffle
will alter the speakers natural zero's and thus will give a diffrent responce.
If you want a better responce from the speaker, then please go ahead - it's a
great thing to do, but if you want to know the impulse responce of the speaker
you are going to make measurements with, then you should know that this will
make the measurement not fully equalent.

The longest responces components comes from DC blockers in curcuitry,
lowfrequency parts of the cross-over filter, and the resonant setting of
bas speaker, volume and possibly the port size and port length.

> And OTOH , would it be possible to remove the reflections by comparing
> different elevations above ground in case of a "tower" ?
> Basically, if the speaker is above a concrete plain (road),
> i can say that only one reflection comes up, with a predictable
> delay, that should be equal for all frequencies (I hope).

In theory you would have an almost perfect mirror of the wave, in practice you
will have a number of extra reflexes forming a tail from the initial reflex.
These comes an expanding ring around the point of the initial reflex, this is
really what you get when the (theoretically spherical wave) hits the surface
and gets reflected back. The smoother surface, the better it is. For low
frequency sound it is behaving much more like a perfect mirror than for higher
sounds, the reason is that the errors in distance must be compared to
wavelength.

> Can't we apply a mirror image speaker below the road, this kind of
> model usually does the trick for field problems...
> Should give characteristic peaks and notches...

Well, yes, you will give the typhical phase in and phase out things. But how
can you predict those? Your speaker is _NOT_ an even source, the energy
distribution is not even as you turn your speaker. The impulse responce that
your speaker sends away towards your mike though direct radiation is not the
same impulse responce that it sends away towards the reflection. Also, the
same thing applies for the mike, it does not have the same charactersitic
impulse responce from its direction towards the speaker as it has towards the
surface.

With this I want to say that what happends up to the first reflection is we
able to use rigth away, but at the first reflection and onwards it gets very
hairy very quickly. My short story is: you don't want to go there.

> Only if the concrete would give an ideal reflection.
> I know nothing about that.

I think I have covered that by now...

Cheers,
Magnus