| Home | About Me | My projects | My wish list |
| Measurements | Tools of the trade | Links | Home Theatre |
Back to Home theatre main page:
12/7/2010: CBT Kit coming soon at Parts Express!
The official web site of D.B. (Don) Keele
In my opinion, I may have about the best center speaker of any home theater in the world. I certainly know of no commercial option that can touch the performance! This is all thanks to D.B. Keele both for developing the CBT technology and for convincing me to build this project and helping me with the design. This is not a trivial design and there are a number of considerations but it is achievable. I was initially convinced that it would present too many problems with image placement but this was only because I was using the wrong arc angle in my initial thoughts. Don showed me my error and explained how it was possible to place the image exactly in the center regardless of listening position. In order to accomplish this however the beam width would be wider than what I wanted so a compromise was reached with the 120 deg arc as opposed to 180. With this design and placement, the image error is only a few inches off center from the worst possible (most off axis) viewing position. Placement below the screen puts the image vertically much farther off axis than this and even this large offset is not really a problem because the mind tends to place the image up on the screen. So, the 4 or 5 inches error horizontally is completely irrelevant. From the two main (center center) viewing positions the error is in fractions of an inch.
I don't know if CBT is the right technology for every application but for a theater center speaker and numerous other applications it is the ideal technology. If only commercial cinemas would adopt CBT technology in a new generation of commercial cinema speakers. Perhaps the 80+ year reign of the horn could end and commercial cinema sound quality could improve to match the capability of a well designed home cinema. Years ago Lucas films introduced us to THX, now perhaps it's time Lucas teamed with Don Keele to usher in a new era of truly high resolution cinema sound! Actually, who needs Lucas Films! This is Don's creation!
4/1/2010: The new center speaker is up and playing mounted under the screen.
Actually it's been up and playing for awhile but I got shipped off to Taiwan right after getting it up and playing.
Click the thumb for a larger image of the graph. These measurements were
made out doors with the speaker laying flat on the driveway upside down so the
tweeters were as close to the ground as possible. Measurements were made
on axis and in 10 deg increments to 70 deg off axis. As you can see the
response stays relatively flat as it falls off in the off axis.
Measurement setup.
I'm using the Emotiva UPA-7 amplifier to drive it.. (Probably the best deal in amplifiers today) The center bank of eight Dayton Reference 6" drivers has its own dedicated channel with the drivers wired series/parallel to 4 ohms. The next channel drives banks 2 and 4 and the 3rd channel drives banks 3 and 5 of the 6" drivers. The fourth channel drives the Aura 2" drivers banks 1, 3, and 5 and the fifth channel drives banks 2 and 4. The sixth channel drives the Tang Band tweeters banks 1, 3, and 5 and the seventh drives banks 2 and 4. The Behringer crossover uses all six outputs so all amplifier channels except the seventh one has its own level control within the crossover. The seventh channel uses the same crossover output as the sixth using a Y and a resistor voltage divider to set the level.
11/6/09: (edited 11/12/09) The new center speaker is almost built...
Drivers:
132 Tangband tweeters: http://www.parts-express.com/pe/showdetl.cfm?Partnumber=264-841
50 Aura 2": http://www.madisound.com/catalog/product_info.php?products_id=8465
20 Dayton Reference 6" drivers: http://www.parts-express.com/pe/showdetl.cfm?Partnumber=295-342
The old Center Speaker ...
4/9/09
The center speaker is up and playing! I have to say it sounds pretty good too! The only thing I might change is that I might build a second one and mount it above the screen. I might omit the 10" drivers for the top one. I probably don't need these to move the image up higher on the screen and it's not like I'm short on output. Well, next to the lines I suppose I might be but still, this is a pretty substantial center speaker. Oh, yeah I know, I'm going to replace those screws in the 7" drivers with black ones. I just didn't have any 10-24 1" on hand at the time.
Click for larger image. Measurement location is in lower left. The
Earthworks M30 microphone is placed over each individual seat.
You can see the unique transfer function targets here. As you can see,
there are never more than two bands or pairs playing at any one time. By
the time the 10" driver pair transfer function reaches the 5" pair, the two meet
at brick walls. At the peaks or, "critical frequencies" only the one pair
is playing. Based on the spacing (in this case 0.55 wavelengths) you get a
single center lobe and two smaller side lobes. Having 0.55 wave length
spacing provides around +/- 37 deg coverage. All the pairs are spaced 0.55
wavelengths at their critical frequencies. The linear slope between pairs
creates a summation that in effect creates moving phantom pairs that maintain
the 0.55 wavelength between the two critical frequencies. Thus the polar
response is maintained across frequency.
The Behringer crossover can't actually do this but it can approximate it. It definitely can't do the required brick wall slopes! Still, it can do 48 dB/oct and that seems to work. I initially dialed it in as close to these curves as possible but this didn't include any correction for driver response or room modes. Then, once it was up and playing I started fine tuning to optimize the response. To get the sharp peaks (critical frequencies) I had used a couple of band pass filters to boost the response to match the sharp point at the top. Interestingly, the actual requirement for these in the room was less than the original setting which had closely matched the targets. I didn't have to reduce them much, but it was necessary to reduce them.
So, no ribbon tweeter ended up in this speaker. It's 100% Dayton Reference. The 10" (regular one, not the high excursion sub), the 7", the 5", the 4" and the tweeter. Increasingly the Dayton Reference drivers prove to be equal or superior to far more expensive options from Scanspeak and Seas. If Parts Express ever realizes how good these drivers are they are going to triple the price...and they will still be a great deal! The little 4" in particular is amazing, probably the best option for the 500 Hz to 2 KHz band in existence. My friend and mentor Brian Elliot has made this conclusion and he test drivers far beyond anyone else around.
3/28/09
The center speaker is under construction. I'm still at a loss for what tweeter to use. I wanted to go with a ribbon in order to have limited vertical off axis radiation but unfortunately I just can't find a good ribbon. I had planned to use an Arum Cantus G2Si but it had too many problems. Like most ribbons it goes into strange breakup modes and starts radiating random tones in addition to the stimulus that are not harmonically related. It went back to Parts Express! The thing was useless! It's a good thing I had held off cutting the through hole for that thing until I received and tested it. I'm probably just going to end up going with a good dome now.
1-27-08: I've changed my mind about placing the center speaker behind the screen. While I can EQ it to be flat, it just doesn't sound right. I'm thinking now I'll just place it below the screen. Ultimately I think this will be a better solution.
This horizontal "M-T-M" style array might at first look like a problematic design. Experienced speaker designers reading this are already thinking what a disaster the horizontal polar pattern must look like. Interestingly, their expectation would be wrong! This will in fact create a very nice +/- 35 deg center lobe from about 100 Hz to over 2 KHz. Beyond that the horizontal polar pattern will be a function of the single 4" driver and the tweeter. There will be no comb filter issues as with a conventional array of this type. By including the side mounted woofer system into this array it can be made the full 19' wide width of the room extending polar control on the low end all the way to the bottom.
Ulrich Horbach and Don Keele have written an article on the use of digital crossovers with symmetrical driver configurations. Basically they uses sets of symmetrical pairs to achieve a uniform polar response in the vertical plane. Where a single pair would have a problematic polar pattern it is combined with another pair such that the combination achieves a uniform pattern. Each pair operates at 100% only at a single frequency creating a very unique pointed top transfer function. I believe this design approach may be a very good solution for horizontal array center channel speakers by simply rotating the idea 90 deg. Of course I will do it as a dipole which will further improve far off axis control.
One issue with the Horback/Keele approach is that in their 5-way example they have chosen ideal driver spacing that is not practically achievable. By rotating it to create a horizontal array I can place the tweeter and upper midrange over under each other. Doing this with the array vertical creates a horizontal polar problem. Another issue is that typical digital crossovers do not have the capability to create the required transfer functions. Ideally they should be able to create what ever transfer function is required and not be limited to conventional analog filter functions. It will be possible to approximate the proper transfer function using the Behringer digital crossover and based on LspCAD simulations, the speaker will still work quite well.
You need to read the above paper to fully understand but notice that the frequencies and spacing distances in the drawing above share common ratios. (310 Hz / 119 Hz = 2.6 and 62.86" / 24.18" = 2.6) The frequencies listed are not crossover frequencies. These are, "critical frequencies". There is no conventional, "pass band" in the crossover transfer function but rather a, "pass point". The pass band is limited to a single frequency. Based on the desired coverage a driver spacing and critical frequency are selected, perhaps 0.55 wavelengths. The next set of drivers will maintain the same relative spacing at their critical frequency. At each critical frequency only one set of drivers are playing and because each pair of drivers have the same relative spacing of 0.55 wavelength, the polar patterns are identical at the critical frequencies. The digital crossover maintains a linear transition at frequencies between the two critical frequencies such that in essence, the sum could be represented by a phantom pair of drivers moving along the baffle always maintaining the same 0.55 wavelength spacing. Once the next critical frequency is reached, the previous pairs crossover becomes a brick wall. The result is a well controlled polar response at all frequencies. The lower limit of control is set by the lowest critical frequency and corresponding spacing in the system. The upper limit is set by the transition to the center tweeter. The polar response of the tweeter band is thus purely a function of the tweeter's polar response.
The tweeter is the only driver that has not yet been chosen. I would like to find a high quality ribbon but I'm not sure any ribbon will achieve the performance I want. Ribbons tend to have a lot of problems but I think there must be a good one out there somewhere. The rest of the drivers will be Dayton Reference. These consistently prove to outperform the vastly more expensive Seas and Scanspeak drivers, especially the new 4" version.