Theory in Dunlavys & vertical dispersion..

[Virtalahde]

I'm trying to wrap my brain around the idea I've read from various places: In case of large Dunlavy speakers (SC-IV etc), you need to listen to them from a hefty distance or else the stereo image falls apart.

Now, playing around with the Xdir simulator from the Tolvan website, I am thinking that this is true with time-aligned MTM/WMTMW configurations that use 1st order cross-overs - lobing seems pretty bad one octave above the cross-over frequency when the woofers don't really roll-off fast enough. Sit further and you won't have such a severe comb-filtering on your forehead.

Is the claim based on the lobing or am I completely off the track? If you had similarily time-aligned drivers with steeper cross-over slopes (LR2 for example), things would look much easier in my simulations.

[dlr]

Dunlavy designed for a listening position of about 10 feet with his large designs. It takes that kind of distance for proper driver integration. Closer than that and the lobing due to center distances becomes a serious issue. Likewise, a large woofer off-axis response would complicate things.

Steeper slopes reduce the overlap region, but will still have lobing problems, just less pronounced. Dunlavy used first order for a transient-perfect target to get close to ideal square-wave response. You won't get that with higher order slopes unless using one of the few higher order transient-perfect types, such as a filler driver. But those fall apart even faster off-axis.

Dave

[cptomes]

Anyone know if Duntech is going to be imported? I had picked the SC-III as my first new expensive speaker purchase and before I could get the money salted away the company was sold and then closed. SC-III (or any Dunlavy for that matter) just doesn't appear very often on eBay or in the classifieds... Audio Video Logic in Des Moines was the biggest dealer, so shipping for me was a nonissue and they had good rotating stock of trade-ins.

I would love to find a set of SC-III or SC-IV. Truly beautiful speakers, very well made with no black magick or voodoo marketing BS.

[DSP_Geek]

Originally posted by Virtalahde
Now, playing around with the Xdir simulator from the Tolvan website, I am thinking that this is true with time-aligned MTM/WMTMW configurations that use 1st order cross-overs - lobing seems pretty bad one octave above the cross-over frequency when the woofers don't really roll-off fast enough. Sit further and you won't have such a severe comb-filtering on your forehead.

Is the claim based on the lobing or am I completely off the track? If you had similarily time-aligned drivers with steeper cross-over slopes (LR2 for example), things would look much easier in my simulations.

You have learned well, young Jedi.

Seriously, though, that's exactly the problem with low-order crossovers, particularly with symmetrical designs such as WMTMW featuring two mids with considerable distance between them. Woofers aren't such a problem because typically the crossover frequency is low enough that lobing isn't quite as severe.

A Stereophile review of the large Dunlavys said: "the Signature VIs are the largest pair of headphones in the history of audio". I have no reason to doubt it, especially when they were quite flattering otherwise: http://www.stereophile.com/floorloud...62/index3.html

[LineSource]

Rockport offers a modern interpretation. Very well reviewed.

[dlr]

Originally posted by LineSource
Rockport offers a modern interpretation. Very well reviewed.

Looking at that photo, it's not the same, even an approximation. The physical offset of the drivers is the inverse of Dunlany and Duntech, that much is obvious. To be a first order time-aligned system requires the smaller drivers to be offset forward of the next larger driver. The one pictured is the inverse for the sake of aesthetics, I'd say. Unless it's got some kind of the time delay in the crossover sections (as in DSP), it's not the equivalent.

dlr

[LineSource]

Originally posted by dlr

Looking at that photo, it's not the same, even an approximation. The physical offset of the drivers is the inverse of Dunlany and Duntech, that much is obvious. To be a first order time-aligned system requires the smaller drivers to be offset forward of the next larger driver. The one pictured is the inverse for the sake of aesthetics, I'd say. Unless it's got some kind of the time delay in the crossover sections (as in DSP), it's not the equivalent.

dlr

This photo of the old school Dunlavy shows the same relative physical offsets with the low frequency drivers physically forward of the tweeter in an effort to get a time aligned wave launch from all cones despite the different attachment point of the voice coil and cone. There is also some delay in the crossover elements and from phase plugs that complicates a simple physical alignment.

The old school Dunlavy used heavy felt in the recessed cavities to absorb side energy before it could reflect off the adjacent bumped-out speaker, plus to marginally reduce cabinet edge diffraction. The new school Rockport uses organic diffraction reducing curves.

Naturally a digital or analog active crossover could add delay to speakers on a flat baffle, but some designers favor high end passive components over ALUs and Opamps. In addition, the heavily rounded edges are necessary to reduce diffraction, so you might as well use the same cabinet technology for time alignment.

[Bare]

Ever actually sit down and Audition a Set Of Dunlavy's??
Despite ALL brochure Babble and self serving 'Audio Hacks' ..
The damned things sound 'average'... at Best.
They look foolish.. oversized goofy things with a multitude of Drivers that barely equal many 2 driver units and their sound is simply mediocre.. they are dead as a product.. no loss, IMO.
If yr gonna Lust after something, at least make it worth dreaming over.

[dlr]

Not sure where my head was at the time.

dlr

[Virtalahde]

People,

thank you for your answers on the topic! Much appreciated!

Originally posted by dlr
You won't get that with higher order slopes unless using one of the few higher order transient-perfect types, such as a filler driver. But those fall apart even faster off-axis.

Ah, filler drivers. Read about that in the John Kreskovsky site.. I think I need to read that again.

Originally posted by DSP_Geek
Woofers aren't such a problem because typically the crossover frequency is low enough that lobing isn't quite as severe.

Yep, I haven't quite figured it out what really is the advantage of a WMTMW vs. MTMWW. Sure, in a speaker like SC-IV I can understand it, you have the woofers operating at well above 200Hz and beyond with 6dB/oct slopes. But in a giant like the one posted above
(thanks for censoring the half naked guy out of it, I unfortunately had seen the photo before..) I'd think the benefit is more trivial.

One nice thing though - woofers spaced like that would most likely excite room modes in a different way, smoothing the in-room bass response.

Originally posted by LineSource
Naturally a digital or analog active crossover could add delay to speakers on a flat baffle, but some designers favor high end passive components over ALUs and Opamps. In addition, the heavily rounded edges are necessary to reduce diffraction, so you might as well use the same cabinet technology for time alignment.

Hehe, which opens up a nice can of worms.

To align for time & phase.. Or phase only? I'm not so sure nice looking in-room square wave response is the key, nor the 6dB/oct cross-overs, but for time alignment seems like a wise thing to do to me.

I am probably correct when I assume that you can time align the drivers physically (by voice coils or however you approximate it), but when you muck up a cross-over there you lost your alignment, more so with steeper cross-over slopes? You could correct this with an all-pass filter of course.

Not that good speakers haven't been made with no time alignment..

I wonder if this post made any sense. I need some sleep.