Originally Posted by catapult
Brandon, as far as measuring reflections (HOM) inside the horn, looking at the impulse response is pretty good. Paul W did some before and after adding fish foam and you could see that it cleaned up the impulse some. I suspect but don't know for sure that the screw-in adapter he was using, with its long straight section, was creating some weird reflections at the throat and mucking up the impulse.
I think I sort of remember that now. Yeah I agree, this is why I mentioned that a CSD of the MLS might be ok, plots the time element of the impulse, which just are not that intuitive for me, unless the reflections are very bad.
On a couple of Penn's points> why don't more people make a "proper" Geddes type WG? I think the biggest thing is 18 months ago, not much more than a few industry pro's knew Geddes work very well. After he came to diyaudio I think a lot more people are becoming familiar with his work. So I think you will start seeing more people use his principles to build real WG's. Just a matter of time I think.
On Wayne P's point about vertical directivity, I remember the whole original debate and to me, it was sort of a mountain out of a molehill, in the discussion as a whole, and the actual technical facts. ANY speaker with two non-coincident drivers, say a woofer and tweeter in a simple 2-way, will form a null in the crossover area as you move vertically from about the mid point of the speaker to above the speaker.
Say you are facing the speaker and at a 30 degree angle above it's mid point looking down at the speaker with your microphone in hand. Basically you have flat response of the woofer at lower frequencies, then as you get around the crossover area a null forms, then the response bounces back as the tweeter takes over. Pretty simple concept right? Well, Wayne uses a horn that is very rectangular, for illustration, say 12 inches wide, but 4 inches tall. This has a very high directivity in the vertical plane due to its small height (versus the deep wall angle). Wayne would rather that we have the flat response of the woofer, then the null forming in the crossover region, then due to the high directivity, a tweeter response that doesn't "bounce back". Why this is better is beyond me. In one you have full response, a null, then full response again. On Wayne's you full response, then a null, and then more of the null all through the tweeter response. In neither example do you have flat response through the whole spectrum, but at least with the standard first example you have a null of a certain width-that the designer can adjust and minimize. With Wayne's it just full response then nothing. How do you balance that? How is that smooth from a power response perspective?
Which brings me to this post:
Originally Posted by LTD02
the first is the on-axis hole that occurs with axial symmetric waveguides. you'll notice that the on-axis measurement has about a 5-7db suckout around 6khz. geddes' answer is to listen off-axis. that is fine, but not being able to listen to speakers on-axis is a little strange to lots of folks.
a second reason is that for a vertical arrangement of speaker and waveguide, an axial symmetric waveguide cannot be matched in directivity vertically.
a third, but minor, reason is that large axial-symmetic horns push the drivers quite far apart, which limits the vertical listening window.
The first is a real concern. The third is a real concern. But the second is bunk, not just because of my above explanation either. How can a round, 12" waveguide be a poorer response match in any direction
to a round, 12" woofer than a rectangular, 12" x 4" horn? It doesn't make sense.
How important are the other two? Well the first doesn't concern me much personally. There will always be uneven frequency response with nulls in some areas and peaks in others. Where they are, what direction, what frequency, and how big they are determine whether they demand a solution or not. The on axis hole is quite small, even if we look at it as power radiated versus area, as it's really a narrow cone shaped dip right? The majority of speaker designs make much bigger and more detrimental sins regarding power response (frequency response plotted over many angles in front of a speaker).
The third issue is a real issue and demands a solution. One solution is to cross low, and this is what Geddes does. I must say though, that it's funny this argument relies on this issue causing high vertical directivity and it being bad. Yet Wayne's argument is that Geddes design isn't directive enough vertically. Hmmm which is right, can't have both...
Now an elliptical waveguide would solve both problems, Geddes understands this would be the ideal form of his idea, but he doesn't have enough time/interest/money to fabricate it. Maybe AE will.
One last point I'd like to make concerning all this directivity and smooth response. For the most part, the width of the mouth determines the horn/WG's cutoff frequency. In general, at frequencies below the cutoff frequency the response flares dramatically. In other words, it's best to have the response down quite a bit before you hit this point. This is one reason why I don't care much for these high aspect rectangular horns, with the 15" width and 4.5" height. I don't have a calculator, but at 4.5" it probably loses pattern control -flares- at 3-4khz. If you are crossing lower than this you could actually have quite low vertical directivity in this area, and the response would be very ragged to boot.