[sdiy] OT: Horn Speakers

[Magnus Danielson]

From: patchell <patchell at silcom.com>
Subject: Re: [sdiy] OT: Horn Speakers
Date: Mon, 19 May 2003 07:05:33 -0700

>     I found a book once that had plans for all kinds of speakers....however,
> that was back in 1970 and I am afraid I don't remember the title, let alone
> the author.  However, they don't call them exponential horns for
> nothing....(the shape of the horn is an exponential)...and it turns out that
> the size of the mouth determines the frequency response....but, that is about
> all I can remember...I would think a search of the internet would turn up
> something...well, maybe not.

A quick dig came up with

http://melhuish.org/audio/horninfo.htm

(Lots of horn-info if you dig around from there it seems)

The AES Journal articles of Earl Geddes is a must-read experience since they
cover the field in a much greater theoretical foundation than any previous
horn theory (which is really just a big mess IMHO). If you want to do horns,
then doing it according to those shapes will get you much closer to an optimum
horn than any of the old stuff. The trick is to "terminate" the horn mouth.
I've done experiments which where very succsessfull in reducing any standing
wave flaws. If you load a TAD driver or a EV driver on one of these you can
reach way up in the screaming sky for the top part.

Traditional horns are bitchy like hell in the response, partly due to impropper
impedance matching of the shape themselfs and partly due to impropper
termination. To fix the first you need to use a propper shape, to fix the later
you can run a handfull of tricks-of-trade.

The point which Geddes have is that the horn-shape should support the natural
shape of the expanding wave. Since the shape at the throat of the horn is a
flat surface which is all oscillating, it is the shape of the expanding wave
from that surface which we need to consider. The horn shape should always be
orthogonal (i.e. in 90 degrees) from the wavefront shape. By a lump of math I
suspect most people here doesn't really feel is very healthy he comes up with
a very simple formula for the shape. I have to dig up the actual AES paper for
it, but it was dead-simple as I recall it.

Whenever you deviate from the 90-degree slope of the waveshape then you get a
reflection backwards due to the impedance error. This is the fun stuff about
3D design issues for wavefronts, it is the 3D shape of the wavefront/impulse
and it's matching to the 3D shape of the design which interacts instead of the
1D properties in traditional transmission lines of RF type. Besides those
aspects you run into the same troubles basically. 3D waveguides for microwave
designs have about the same properties since the math work the same way.

Check out these links:
http://melhuish.org/audio/links.htm

Cheers,
Magnus