The 'unity' horn is an interesting way to combine radiation from multiple drivers in multiple frequency ranges into a single horn. If designed correctly, the sound produced is all phase-correct, thus producing a coaxial, acoustically coincident wide-range horn capable of correctly reproducing the actual electronic signal sent to it. The 'unity' concept was conceived by Tom Danley, and is patented. To see the patent, go here. To view the actual patent, click on the 'images' link. You will need a special TIFF viewer which is described in the help section. There was a similar, but different design from Renkus-Heinz prior to the unity horn, and several major professional sound companies came out with very similar products following the commercial release of the unity.
In one of my implementations of the unity, built while I was in school in 2002 and shown in the pictures below, I used a 40 degree horn with an 80 degree flair at the mouth. This is narrower than most of the commercial unity horns that Sound Physics Labs produces, but I feel it works well for a home listening environment. The pattern is not controlled to as a low a frequency as a wider angle with the same horn dimensions would allow, though. This horn covers 200Hz to 16kHz using two driver sections. The midrange section covers 200Hz to 1kHz and uses four 5" drivers - MCM 55-1595's. The sensitivity is around 106dB/watt/m. The tweeter is a TAD 2001 compression driver. On this type of horn, it has a sensitivity that varies significantly with frequency which must be accounted for in the crossover design. It runs at about 116dB/watt/m from 1kHz to 3kHz, then rolling off to around 106dB/watt/m at 16kHz.
For covering the range below 200Hz, I used an H-baffle dipole with two EVM12L woofers, one reversed in an effort to cancel even order harmonics. For below 100Hz, I used my 27Hz horn subwoofer. The crossover for this system was active, all filters being custom built. The amps for each driver were a mix. The tweeter was driven by a single-ended class A tube amp with no negative feedback. The output tube was a 45, and the input tube was a 12SN7GT. The midrange and midbass were run from an Adcom GFA-535, and the subwoofer was driven by a home theater receiver.
Update - 12/5/03
I have written software to predict performance of a unity style horn. It comprises programming the standard electro-acoustic model for a driver / horn combination, but with a few added elements to account for the specific midrange loading geometry. For two different midrange geometries, the program's predictions match measurements very well. Using this, I plan to investigate other possible midrange drivers for use in a low-cost, high performance unity horn. If I find a good candidate and build a horn using it, results may be published here.
Update - 5/22/05
I recently spent some time coming up with some driver parameters for a unity horn midrange. Since using a model like Leach's which is solved for optimum driver parameters is not really applicable in this case (the unity is a conical horn, not infinite exponential, and it has extra elements in the equivalent circuit representation of the system as compared to a standard horn), I just went through multiple iterations of driver parameters using my software model. I set constraints of a 60 degree included angle horn with a smooth flair at the mouth, a reasonable mouth size for home use, and small, short midrange holes. If anyone out there knows of a driver that fits these parameters, please email me - I haven't found any (and no, I am not interested in buying custom drivers from Audio Technology). In the meantime, I am working on designing and building my own drivers which should be much better performers than anything commercially available anyway (but also much more complicated to design, build, and assemble, so they would be prohibitively expensive to buy). Of course the problem with building my own drivers is that I also have to design and build all the tooling for a lot of the parts, so the amount of work is pretty large.
Sd: 67.5 cm2
Mms: 10.1 gm
Cms: 0.38 mm/N
Re: 24.7 ohms
BL: 20.4 T*m
Rms: 1.22 N*s/m, but this seems non-critical
Fs: 81 Hz
Vas: 2.42 L
Sensitivity: 88dB/watt/1m in half space (not on the horn)
Xmax would need to be around 2mm.
Note that this driver would need a rear chamber added, and these parameters are for the driver with no rear chamber. Also, probably obviously, this driver is designed so that four drivers can be wired in parallel to give a standard 8 ohm load. Since these likely won't exist in a current production driver, one could also look for a driver with the same T/S parameters, an Re of 6.2 ohms, and thus a BL of 10.2 T*m in order to use series-parallel wiring.
Here are some pictures of the unity-based system I built in 2002:
Here is a link to a 3-way unity system I built before the above system. This picture is of it in the rather dirty basement of a house I lived in during college. The horn was made by modifying a 2-way unity horn I built in 2001. It used a Radian 475 for the tweeter, and two of the MCM midranges mentioned above with small, long entry holes (not ideal, but I didn't know that at the time). These holes are the white spots at the very back of the horn, on either side of the tweeter's throat. For the midbass, I used two EVM12L woofers. The sound from these entered through the holes in the yellow portion of the horn. This all had horribly non-flat response, but after a lot of active eq and level matching in the crossover (which was all active anyway), the sound was actually pretty good. I think the concentric mounting made up for or masked a lot of the other problems in the system.