Horizontal widely spaced driver arrangements for stereo

[Tim Link]

About 15 years ago I recall seeing a pair of B&W 801 speakers that were disassembled and arranged so that the tweeters were pretty close together in front of the listener, the mids were out at about the normal 60 degree triangle position, and the woofers were way out wide, to either side of the listener. This had been set up somewhere as a demo, with some explanation that the lower frequencies required more separation to achieve a similar stereo effect. I may have some details wrong but I recall the picture with a young woman sitting in the listening chair and the drivers all separated around her.

Since seeing that, I've not seen more about that kind of arrangement. I experimented with it and found it very intriguing sounding, creating an illusion of sounds actually occurring in the room in a way they simply would not when all the drivers were stacked up on top of each other in the usual, sensible arrangement. Now, through happenstance I'm back to that arrangement to some degree, with my bookshelf speakers fairly close together, but crossed over at 300hz to corner horns that are out in the corners. I normally run this in a 3 speaker up-mix for the bookshelves to create some crosstalk reduction that widens the stereo sound stage and solidifies the center image. Out of curiosity I decided to turn off the center speaker up-mixing and listen to the setup in regular stereo. To my surprise, it continued to produce a wide soundstage, with clear high tones seeming to be coming from well beyond the bookshelf speakers. The little speakers were doing a fine disappearing act. I also tried running R.A.C.E. through the stereo setup to actively reduce crosstalk. It widened the soundstage more to like what I hear with my normal 3 speaker up-mix, but not a whole lot wider than the straight stereo arrangement was creating.

So, it seems there still is something compelling I'm hearing with the horizontal arrangement of high and low frequency drivers, spread apart at considerable distances. This is definitely not a point source, and would be a real bear to qualify using the speaker testing method popular on this site. I guess you could measure each driver/cabinet separately and add everything up somehow.

Has anybody else read about this configuration, or tried it? Is there a particular name for it?

 

[Tim Link]

I did a little searching and came across the idea of stereo shuffling, to increase the apparent stereo width of lower frequency signals to match higher frequency signals. The speaker arrangement seems to be somewhat related to addressing the same problem that shuffling does with recordings. If they've already shuffled the recording, then the horizontal driver arrangement may exaggerate it unnaturally. If not, it may help. https://www.audiosignal.co.uk/Resources/Stereo_shuffling_A4.pdf

 

[Curvature]

About 15 years ago I recall seeing a pair of B&W 801 speakers that were disassembled and arranged so that the tweeters were pretty close together in front of the listener, the mids were out at about the normal 60 degree triangle position, and the woofers were way out wide, to either side of the listener. This had been set up somewhere as a demo, with some explanation that the lower frequencies required more separation to achieve a similar stereo effect. I may have some details wrong but I recall the picture with a young woman sitting in the listening chair and the drivers all separated around her.

Since seeing that, I've not seen more about that kind of arrangement. I experimented with it and found it very intriguing sounding, creating an illusion of sounds actually occurring in the room in a way they simply would not when all the drivers were stacked up on top of each other in the usual, sensible arrangement. Now, through happenstance I'm back to that arrangement to some degree, with my bookshelf speakers fairly close together, but crossed over at 300hz to corner horns that are out in the corners. I normally run this in a 3 speaker up-mix for the bookshelves to create some crosstalk reduction that widens the stereo sound stage and solidifies the center image. Out of curiosity I decided to turn off the center speaker up-mixing and listen to the setup in regular stereo. To my surprise, it continued to produce a wide soundstage, with clear high tones seeming to be coming from well beyond the bookshelf speakers. The little speakers were doing a fine disappearing act. I also tried running R.A.C.E. through the stereo setup to actively reduce crosstalk. It widened the soundstage more to like what I hear with my normal 3 speaker up-mix, but not a whole lot wider than the straight stereo arrangement was creating.

So, it seems there still is something compelling I'm hearing with the horizontal arrangement of high and low frequency drivers, spread apart at considerable distances. This is definitely not a point source, and would be a real bear to qualify using the speaker testing method popular on this site. I guess you could measure each driver/cabinet separately and add everything up somehow.

Has anybody else read about this configuration, or tried it? Is there a particular name for it?

Would you mind posting pictures of the setup?

 

[Tim Link]

Would you mind posting pictures of the setup?

Here's a picture of my current room setup. You can see a picture of an older picture of the room on the TV screen in the picture! The horns in the corner are only playing from 300Hz down. The midrange and tweeter horns aren't hooked up to an amp.

 

[Chazz6]

Mild version of the same? I recently lay my Focal Aria 906 speakers into horizontal position, tweeters on the inside a little over a meter apart from each other . Subjective report only: I like it. I hear the same or somewhat less stereo effect, but the different players remain distinct on the soundstage, while altogether the music seems to be integrated more, less obvious that there are two sources.

 

[Tim Link]

Mild version of the same? I recently lay my Focal Aria 906 speakers into horizontal position, tweeters on the inside a little over a meter apart from each other . Subjective report only: I like it. I hear the same or somewhat less stereo effect, but the different players remain distinct on the soundstage, while altogether the music seems to be integrated more, less obvious that there are two sources.

Interesting. That's not a huge spacing difference but your impression of it being less obvious that there are two sources is similar to my own impression. On my first experiment I had the tweeters only about 1 foot apart, crossed over at 5K. I had some 2" drivers at about 4 feet apart covering from 5k to 1.25k. I then had some 5" woofers about 10 feet apart covering from 1.25k to about 80Hz, and then a single subwoofer in the middle of the room. I was building a pair of 5 way towers at the time with separate enclosures for each driver, so decided to try setting up like that with the pieces I had at that point after seeing the article. I was using an active crossover with 48dB/ octave slopes, figuring that would minimize any interference patterns from the large center to center spacing between the drivers. Not sure if that was necessary but it did produce a very compelling sound.

 

[Tim Link]

Since I originally posted on this topic I've moved my room around so I'm playing on the long wall. This moves the corner horns much further apart. I've now got some horn tweeters much closer together in the middle, like I had the little Sony's before. I don't have a third matching tweeter horn, so it's just straight 2 channel stereo now with the tweeters in much closer than the mid and bass drivers, although not quite as close together as the little Sony bookshelf were. After a lot of experimenting I've arrived at having the tweeters crossed over at 2k, which is much higher than I expected to sound optimal. The midrange horns are covering the range from 300Hz to 2K. This is very enjoyable, and the tweeters don't call attention to themselves. I'd swear there's no sound coming out of them until I turn them off, or get up close to them. I'm surprised to once again have the midrange horns covering the full range they were intended for. These mid-horns get very beamy up at 2k, which used to bother me, but I'm sitting far enough back from them now that it doesn't seem to matter. They sound really good. I was prompted to try pushing them back up to 2k based on distortion measurements. Those 2 foot long mid-horns with their 10" drivers have less distortion in their bandpass than either the JBL compression drivers or the bass horns can manage trying to cover the same range.

I've been thinking about the coherency issue, which naturally comes to mind when one considers moving tweeters a long ways away from midrange drivers. I learned from previous experiments that steep crossover slopes help to reduce my perception of there being a problem. In my early experiments it was obvious to me that the drivers seemed to integrate better when there was less interference between them. I could get up close to the speaker and still not tell that there was a tweeter and a midrange operating separately, especially if they were well level matched. Coherency in terms of all the sound coming from one place is a mixed bag. It's good in terms of creating excellent realism when trying to make it sound like a single instrument or singer is coming from where the speaker is. When you move to stereo, I'd argue that the point source coherency becomes a problem. It accentuates the fact that all the sounds are actually emerging from two points in the room. Even if the speakers can seem to disappear well enough in that setup, there are telltale signs in our HRTF response that give clues we don't necessarily want to hear. It's like trying to take a picture through a screen window. In this case the screen has only two holes. We want the image in the background to be in focus, but the screen should be blurred as much as possible. We can do this by moving the drivers horizontally apart and then time aligning them for the listening position. You could just add more point source speakers playing the same frequency ranges, and I've tried that. It can sound really good, but the time alignment gets really tricky when you have multiple sources playing the same frequency bands, so small head movements can result in strong tonal changes. I had a small setup for a while with four stereo speakers, two time aligned on each side to the corresponding ear. I could EQ to taste by moving my head back and forth a few inches. Interesting, but tiresome. If you had a big enough room and could sit far enough away from the stereo pairs the problem could be minimized. I might have to try it again.

In my current setup I suspect I'd be better off I had the frequency ranges divided up a bit differently, maybe 5k and up super tweeters closest together, then 1.5 to 5K mids a bit further apart, then 300 to 1.5K wider still, and finally 300 and below all the way out to the corners.

 

[Tim Link]

Patrick Bateman on DIYaudio answered my opening question with this post: https://www.diyaudio.com/community/threads/cbt-with-crosstalk-cancellation.405434/post-7509240

Opsodis!

In that post I see some of the pictures I recall. Optimum sound distribution, Opsodis, is more than just a horizontal array of drivers, I now see. But the concept is similar to what I've been attempting in the simplest possible way with my current 5 speaker array that has mids and tweets in the middle and woofers way out to the sides.