The World's best $200 Speaker?
The Samadhi Kittentm : A DIY Loudspeaker Project
Copyright 1999 (Samadhi Acoustics, Ltd.)
Sometimes inspiration is triggered by the most innocuous of questions. In the case of the Kitten, it was Dr. Scott Julian of Chippewa Falls, Wisconsin, who e-mailed me to ask about resurrecting the Radio Shack LX-4 (see Tip of the Month #3). It's common knowledge by now that the LX-4 is a discontinued model; dealers selling off their existing stock at ridiculous prices. Scott's idea was to purchase the woofer and tweeter from Radio Shack (there's still a plentiful supply of drivers), and roll his own cabinet and crossover. At precisely that moment the proverbial light bulb went off. Why not redesign the LX-4 using a Ceiling Boundary Ambience Enhancement (tm) (CBAE) alignment?
I've spent two years researching and developing this technology, which is now the cornerstone for the Samadhi Acoustics, Ltd., product line. When it comes to recreating a believable spatial impression in a typical room from two-channel stereo with clarity and focus, there's the right way (CBAE) and there's the wrong way (the rest of the market).
Most loudspeaker designers today follow the established engineering paradigm of designing for a flat on-axis frequency response under echo-free conditions. The speaker prototype is introduced into an anechoic chamber and its crossover network is tweake d until it measures flat at the microphone position on axis. These same designers will tell you that they have no clue as to how this speaker might sound in a real-world environment, such as your listening room. The truth is that speakers that measure ali ke on-axis in an anechoic chamber sound much different in a living room. That's why so many audiophiles view the room as the enemy and spend a fortune on absorptive treatments that makes their room acoustics resemble ever so closely the inside of an anech oic chamber.
The often-stated rationale for an acoustically dead listening room is that the recording contains the original ambient signature of the hall and that is should therefore not be "colored" during playback by the listening room reflections. There are two basic problems with this assertion. First, with two-channel audio, both the reverberant cues and the direct sound are reproduced from a plane in front of the listener. This is most unnatural. In contrast, in the concert hall reverberance "bathes" the list ener due to reflections from the sidewalls and ceiling. It is the time signature of the reflections that gives the listener the Gestalt of "being there," of actually being immersed in the acoustic space. Therefore, in a typical room diffusion is much more important than absorption. Some sound energy needs to arrive at the listening seat delayed relative to the direct sound.
Another crucial consideration is that in an average room, the soundfield at the listening seat is a combination of direct and reflected energy. Even at 9-feet from the speakers the soundfield is already dominated by reflected sound. The reflected/direc t ratio is, of course, even greater in the concert hall. Amar Bose came up with an 8:1 ratio in the late 60s based on actual measurements in several halls. His attempt to squeeze a concert hall soundfield into a typical living room led to the Bose 901 lou dspeaker, which radiates 89% of its sound energy from the back and only 11% from the front.
P. H. Chapelle (1973) was probably the first to get it right: "… the traditional on-axis response does not give a good approximation of the spectral balance of a loudspeaker at typical listening positions in a living room… the soundfield has approximat ely the same spectral balance as the power response." This clearly means that careful attention needs to be paid to the power response or the off-axis output of the speaker as a function of frequency. A typical two-way speaker is omnidirectional in the ba ss and starts to beam as the woofer struggles to meet the tweeter in the upper midrange. As a result, reflected sound that arrives at the listening seat is deficient in midrange energy and the soundfield where your ears happen to be is tonally colored by the speaker's power response.
CBAE is founded on a new paradigm of loudspeaker design that takes the listener and room into account and attempts to recreate the most natural soundfield at the listening seat:
The power response is contoured so as to mimic that of musical instrument, being omnidirectional from the bass through the midrange and becoming progressively more directional with increasing frequency. The electrical network and acoustic slopes of the drivers are contoured to achieve the requisite power response. The woofer is pointed toward the ceiling and the tweeter faces the listener. This facilitates several objectives:
- Immerse the listener in a soundfield that provides the correct time signature of reflected energy. This is essential in the midrange, as the midrange is responsible for most of a recording's ambience.
- Enables a 2 - 3 dB boost in the midrange with a 10 to 30 millisecond delay relative to the direct sound. This requires a ceiling height in the range from 8' to 16'. The first 30 milliseconds define the Haas Window: sound arriving outside this time win dow is no longer fused with the direct sound and is perceived as an echo.
- Minimize the generation of early reflection in the upper midrange that color sound and diffuse image focus. Speakers that are omnidirectional through the octave from 2 to 4 kHz have this problem. Omnidirectional radiators in general splash lots of sou nd off sidewalls creating strong early reflections within a time window of 10 milliseconds.
- Distortion products from the woofer beam along the woofer axis toward the ceiling and away from the listener. This promotes a greater sense of clarity.
The Kitten Design
As it turned out, the LX-4 driver complement was not entirely ideal for a CBAE alignment. In particular, I would have wished for a tweeter that could go down to 2 kHz with decent power handling. Also, the four samples of the Linnaeum ET-5A twee ter I tested were not well matched. The variations were on the order of +/- 2 dB. Still, the end result represents a fair approximation of the magic of CBAE. The soundstage is huge and stable. Individual voices and instruments are perceived to be suspende d between the floor and ceiling. Stand-mounted on Target HS-70 (28" tall) stands, the Kitten just doesn't sound like a small minimonitor should. Its tonal balance is surprisingly full bodied with a convincing lower midrange. I opted for a slight midbass emphasis to offset the lack of deep bass. With eyes closed, it would fool most people into believing they were listening to a large floor-standing speaker. If you already own a pair of LX4s, consider the following: the Kitten is so much more enjoyable that the stock or modified version of the LX-4 so that you may wish to cannibalize the drivers from the stock pair and convert your LX-4s into Kittens. Keep in mind that there is in fact no deep bass. This is neither a disco speaker nor a party animal. The Ch ihuahua-like Kevlar 4" woofer is excursion limited and requires care in feeding. Take care not to over drive it with high-power material.
The bass alignment is fairly close to that of the original LX-4. The prototype boxes I experimented with (see photo 1 ) were built by Dr. Julian out of 1"-t hick veneered MDF. The internal dimensions of the box are: 8 5/8" H x 5 1/8" D x 4 1/4" D. The bass reflex vent is made of PVC pipe, 3 1/4" long, 3/4" in diameter and is located on lower half of the back baffle. The top of the tweeter frame is 3 7/8" belo w the top of the box. Note that all of my design work was conducted with the crossover network outside of the box. You should consider adding a false compartment either on the bottom or back of the cabinet to house the crossover network.
The Crossover Network
Fig. 1 gives the crossover schematic. All coils are 18-gauge air core. The power rating for the resistors is 15 watts. As far as capacitor types, I recommend Solen metalized polypropylene "Fast Caps" - including the 25.0 uF value for the woofer. On the board layout, be sure to keep the coils spaced far apart from each other. The following vendors should stock all of the required crossover parts:
Madisound Speaker Components
8608 University Green
P.O. Box 44283
Madison, WI 53744 USA
Tel: (608) 831-3433
Zalytron Industries Corp.
469 Jericho Turnpike
Mineola, NY 11501
Tel: (516) 747-3515
FAX: (516) 294-1943
Note that the drivers are connected in phase. Be sure to obey the polarity markings on the tweeter. Check for proper woofer polarity by connecting a 1.5-volt C or D battery to the woofer's input terminals. Verify that the positive terminal on the batte ry gives a forward cone displacement when it is connected to the "+" terminal on the woofer.
The drivers are available through your local Radio Shack dealer. The part number and pricing are as follows:
Linnaeum monopole tweeter model ET-5A: part #11086147 ($14.99 ea.)
Kevlar 4" woofer: part #11086154 ($19.99 ea.)
Two of the tweeter samples I received had their Mylar vanes dimpled (see photo 2 for an example of a damaged tweeter). As you can imagine, this does impact the tweeter's frequency response for the worse. Be sure to inspect the tweeters for concealed damage.
A complete set of drivers (plus shipping charges from the warehouse in Dallas, TX) should add up to about $80.
Please note that the Samadhi Kittentm and CBAEtm technology are the intellectual property of Samadhi Acoustics, Ltd. This design is not in the public domain. Permission is hereby granted for individual home constructors and hobbyists to duplicate the Kitten design for their own personal use. Use of this design for commercial purposes is strictly forbidden. Licensing rights may be granted. Contact Samadhi Acoustics for details (P.O. Box 4940, Los Alamos, NM 87544 USA).