ARDA - Assisted Rectangular Driver Array

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm proposing that a couple major advantages of the Bose 901's be put to use in a more conventional system: the Rectangular Driver Array, where 8 or so full-range drivers are used, but assisted by tweeters covering the high range and a large bass woofer built into the cabinet, so that the system truly covers the full frequency spectrum, while keeping crossover points away from the vocal frequencies. Another advantage of this design is that many separate full-range drivers can have more accurate pistonic motion, because the cone will not deform quite so much, and the small size is made up for in the bass regions by the subwoofer unit(s). And to make up for the high-range frequencies lost, a crossover point could be put at about 10,000 Hz so that the tweeters kick in and fill, up to 40KHz in some cases.

What do you think of this idea?
a couple major advantages of the Bose 901's

Stop your making my sides hurt.


  • laughing.jpg
    15.5 KB · Views: 490
Personally I don't think you will get very satisfactory results, but try it anyway, tho Bose apart others have tried similar arrangements with less than satisfactory results. The main difference with Bose is that they use an angled rear baffle firing 4 drivers each side off the wall (indirect) and one driver in the front direct at the listener for the L & R stereo effect. Not only that but through their amp/processor to equalise everything down to a flat response including I suspect some time delay compensation in the amp as well. Due to driver lobing it will be a nightmare to get working properly. I would say you are on your own.
Joined 2001
Paid Member
comb-filtering... aka interference

Multiple drivers producing the same frequencies where those frequencies are smaller than the space between drivers (actually half-wl of the freq IIRC) will suffer from interference that will "blur" everything. A lot of work has been done on this for line arrays where the problem can be limited to a single axis, a rectangular array would square the problems.

Jim Griffin White Paper

Paul taylor Line-source Paper

For this to have a chance of reasonable performance you need to be in the far-field for the drivers to integrate.

This is not quite as impossible as it seems

There is a system knows as a Bessel array. This can come in the form of a line-array, or as a block of units. The idea is that the drive-units are connected in such a way that they emulate a single point-source unit, and exhibit the same off-axis response as a single unit.

The basic units are the 5-element line-array and the 7-element line-array (which actually uses only six units, the centre one being omitted), as well as the 25-element block-array (which is derived from the 5-element line-array) as well as the 36-element block-array (which is derived from the 7-element line-array). In the case of the 7-element array, one can put a tweeter in the middle for a true co-axial array, but things are not so easy with the 36-element block-array (you'll see what I mean when you see it's design).

The main advantage of the Bessel arrays are that you have higher power-handling. They were designed so that many cheap full-range units can be used for a high-power design that is cost-effective. Disadvantages? Well, the 7-element array has a combined impedance (in passive form) of only 2.3 ohms, but the sensitivity of the array is basically the same as for a single unit. Furthermore, the listening position must be at least a distance of 5 X (times) the length of the array away, and preferably as much as 10 X. So for a 7-element array, using say six 5" units (with a co-axially mounted 1" tweeter with a face-plate outer diameter of say 4", the total length of the line would be a minimum of 34". Allow for some driver spacing, and the line could be say 35". Now, the listening distance must be at least 5 X the length (175" or 14.58 ft.), with optimal listening distances being 10 X the length and greater (350" or 29.2 ft.). As you can see, this is quite a significant disadvantage, unless you have a large listening room. At such distances, the block array might be a better idea, but the cost increases quite a bit.

I have thought of trying this out myself. Stryke Audio ( sell a suitably cheap driver called the SA2-351, a 2" X 3.5" fullrange driver that goes for a low $2.95 each. Use these in a 25-element block-array per side, and the result will be a array measuring 10" wide by 17.5" tall. The optimal listening distance for these will be 17.5 ft, but they will start sounding good from about 9 ft away. I would also omit the central driver (using only 24 elements per side), and put in it's place two CTS 2" piezo tweeters per side. This will make the center line a bit taller (0.5" taller), but this is not a train-smash. Below that one can use a pair of 10" or 12" subwoofers (the units from the Blueprint 01-series will fit nicely and are not very expensive. The XO-points can be at 100Hz (to the subs) and 20kHz to the piezo-tweeters (which are good until 50kHz). I think the results could be quite good as long as the minimum listening distance is observed.

Any comments, suggestions or ideas anyone? Or am I being too ludicrous to even warrant an answer??? :D


Actually Deon, I think it's well worth the minimal investment. I had researched Bessel Arrays about a year ago when I was investigating point source speakers. I have always been a fan of the Quad 63 delay line approach, and this seems like a neat way of doing something similar with dynamic drivers.

I believe that the original intent for these arrays was for sound reinforcement, didn't Disney do some work on them? Can't remember, maybe it was Phillips. I have some notes and web sites stashed away on it, will have to dig it up.

I never got around to trying it due to driver costs, but you've pointed out some good ones.

Let us know if you try it. I think that an active crossover would be the easiest way to attain the different drive levels of the individual speakers, cheap LM3886 opamps one on each driver, thoughts?

Bessel Arrays

Hi Ron

I think it would be Philips, as they hold the patent for Bessel arrays.

I like you ides of going active, as this will definitely be optimal, but I think that LM3875s are a bit overkill. LM1875s would be more cost-effective, and I don't think you'll hit a power or SPL ceiling in a normal listening room very easily (unless one is almost deaf).

As far as trying it out, I unfortunately have so many projects waiting in the wings for the limited funds I can throw at it, that this one will HAVE to wait :) How about you?

Bessel vs Line arrays

Bessel arrays have never proven particularly successful, offering few benefits to justify their complexity. Line arrays work, and currently comprise better than 90% of very high end ($100k and up) pro-sound installations because they work far better than the cluster (rectangular) arrays they are replacing. Be it a Bose 901 or a 32 box pro-touring cluster array that design approach just doesn't work very well.
I've already done the 5 x 5 square bessel array and the 5 and 7 unit bessel bessel arrays, using the amplitude weighted approach. It's actually quite simple, just a wiring configuration. The link I had for it stopped working months ago.

It works perfectly for what it is intended and that is to make an array disperse like a point source. The problem is that imaging is lost. The sound is quite interesting in that with 2 speakers the image is just a wall of sound between the speakers.

If anyone has the book or even just the wiring diagrams, please post the info, as I'd like to try it again for use as a center channel. I think the 7 unit array mounted horizontally may make a perfect center channel for HT.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.