JBL M2 for The Poors

[plasnu]

Patrick, I finally concluded that JBL waveguides are NOT an improved version of their horns after listening to JBL horns/waveguide for 2 years. Those JBL waveguides have very little relationship to their previously designed horns, culturally, philosophically, nor technically.

I strongly believe that those new JBL waveguides were designed by a person who dislikes the sound of horn, for people who dislike the sound of horn, and they do not sound like a horn indeed. Still, they do sound like a compression driver, though.

[Patrick Bateman]

Originally Posted by Patrick Bateman

I made something similar to the JBL Image Control Waveguide. I set the size so that it's footprint would be identical to the Elliptical Oblate Spheroidal waveguide from yesterday. From the side, they look similar...





From above, you can see the "beaks"


The model in ABEC





I thought this graph is REALLY interesting : though the footprint is *identical* to the EOS waveguide, the beamwidth is about 20-30% wider! Very interesting. Now I see why JBL stated that the I.C. Waveguide allowed them to cram a compression driver into an in-wall speaker.


Here's the output on axis. Above 4khz, the I.C. waveguide has more output than the E.O.S. waveguide. I believe that this is because the ICW is radiating into a narrower angle, and therefor the output level rises. This waveguide is a study in contradictions: on-axis, it's radiating into a narrower angle, raising output. Somehow, directivity is *wider* than a EOS waveguide. I assume this has something to do with the diagonal diffraction slots.

The obvious thing to do is blend the two and get a compromise between both: the smooth polars of the EOS with the wider directivity of the ICW.

footnote: if you look at the graph of the EOS response from yesterday, and the ICW response from today, the latter looks "lumpier." Part of the reason for this is that I doubled the resolution of the sim. As I understand it, ABEC only simulates at the frequencies that you tell it to simulate. IE, if you tell it to simulate eight frequencies, that's all you get; everything else is interpolated. So I cranked up the resolution on today's sim to 64 frequencies, whereas yesterday's was only 32. My laptop only has four cores, so I'll need to set up a proper desktop so that I can start doing sims with 100+ frequencies.

I'm doing a call back to a previous post, because I've been doing some more sims of something similar to an Image Control Waveguide

[Patrick Bateman]

I'll cut to the chase. My sims seem to show two things:

1) The horizontal diffraction slots of the JBL ICW seems to outperform the vertical diffraction slots of a conventional diffraction horn.

2) The sims seem to indicate that the larger your waveguide is, the worse it performs. This is something I've long noticed, anecdotally, so it really requires some research. On the forums, a few people have noted that the much less expensive JBL LSRs sound competitive with the much larger and more expensive M2s. Some have speculated that the difference is the compression driver, but perhaps the difference is simply the waveguide.













Here's a sim of a 'conventional' waveguide with a diffraction slot, similar to the 18Sound XT1086. The performance isn't great, and I think the issue is that the diffraction slot is screwing up the pathlengths. In the ABEC sim, note how lobes start to appear as low as 2500Hz, and the lobes become more severe at higher frequencies. Geddes has done a ton of research on diffraction slots, and has argued that the discontinuity in the slot creates higher order modes. This ABEC sim seems to show that the diffraction slot also creates comb filtering due to the pathlength differences caused by the slot. If you look at a diffraction waveguide as two segments, there is a diffraction slot that 'feeds' the larger waveguide bell. For the waveguide to work properly, it must be fed with a wavefront that matches. For instance, if the 'bell' of your diffraction waveguide measures 120 degrees by 60 degrees, then the wavefront that exits the diffraction slot must match that, or you're going to get discontinuity.



I threw together this pic to illustrate what I mean. In the pic, note how there's a mismatch between the wavefront shapes on the vertical and the horizontal. The mismatch is due to the varying pathlengths.

Basically there only seems to be a couple of ideal solutions here:

1) You could drive the waveguide bell with a ribbon. But this isn't a perfect solution, because the ribbon isn't a perfect match for the waveguide. But it might be better than a compression driver because...

2) If you use a compression driver, you are probably going to have astigmatism. Basically the wavefront that exits the diffraction slot will be bent on the vertical axis but mostly flat on the horizontal axis.

I think that if one had a LOT of time to kill, they could probably make about a hundred different combinations of diffraction slot and waveguide bell and you could probably come up with something that works halfway decent. Of course, this would only work for a single compression driver, because I haven't even touched on the fact that the phase plug of the compression driver itself will alter the results of this!

[whgeiger]

avoid discontinuities in the second derivative of the of the curves, both longitudinal and transverse, that define all horn boundaries. This tack should minimize the horn artifacts otherwise attributable to its size and geomerty. WHG