Thiel CS5 loudspeaker
I've said it before and I'll say it again: a would-be loudspeaker designer shouldn't even start to think about the possibility of maybe designing a full-range, multi-way loudspeaker until he (and they do all appear to be men) has cut his teeth on a small two-way design. There is still as much art as science in designing a successful loudspeaker, even with all the computer-aided this and Thiele-and-Small that, that even a two-way design requires a designer either to be possessed of a monster talent or of the willingness to undergo months, even years, of tedious and repetitive work—or of both. For a would-be speaker engineer to start his career with a wide dynamic-range, multi-way design, intended to cover the entire musical spectrum from infra-bass to ultra-treble, seems to me to be a perfect case of an admittedly well-intentioned fool rushing in where any sufficiently self-critical angel would fear to tread.
Which stops no one! Consumer Electronics Shows are noteworthy for the large number of large, complex, multi–drive-unit first fruits from would-be high-end loudspeaker designers. And when I was Editor of the UK's Hi-Fi News & Record Review, a magazine that regularly featured DIY loudspeaker designs, I noticed that these designs' popularity among readers was in direct proportion to their complexity, not their sound quality. Publish a design for an excellent, musical-sounding two-way minimonitor and only a few readers take notice; but publish a design for a four-way transmission line the size of an armoire that sounds like nothing on Earth—especially music—and you have to take on extra secretarial staff to cope with the demand for plans.
I think this is why I've always had a high regard for that soft-spoken Southern gentleman Jim Thiel. It was more years ago than I care to remember since I first sipped some of his special whisky (the kind the people in Kentucky keep for themselves) and heard one of his loudspeakers at a Chicago CES. I was impressed, however, that all his designs seem to have had their emphasis placed on extracting the maximum performance from a small number of ingredients rather than trying to bully problems out of existence by adding complexity or to disguise them by featuring large amounts of perceived value. (There isn't even a hint of the "Gee, look at all those drivers!" syndrome with a Thiel loudspeaker.) He also seemed determined not to attempt to run before he had duly learned the rules and skills of walking, the result being a small but high-performance product line for the company that he runs with Kathy Gornik and his woodworker brother Tom.
The news in late '88 that Jim was finally working on a "flagship" design was extremely stimulating, therefore. This would be one to watch for. And, indeed, when I first heard the prototype CS5 in the Audio Excellence room at Stereophile's 1989 San Mateo show, a speaker that embodied every little thing Jim had ever learned and understood about loudspeaker design, it is no exaggeration to say that I felt I was in the presence of potential greatness. I was understandably impatient to get a pair into my own listening room.
As large (and heavy) as the CS5s seem when you are unpacking them from their shipping crate, they take up a surprisingly small area in the listening room. Indeed, for such a large speaker, they are a long way from visually dominating the listening room. One reason for this is their slim profile; another is the glossy black, gently curved, sloped-back baffle, cast from synthetic marble and said to be five times as stiff as a conventional fiberboard baffle.
At the base of this marble casting are aligned three 8" woofers, their Kevlar cones painted black. Closer inspection reveals that the top and bottom differ from the middle by having pads of some rubber-like material applied to their cones, presumably to roll off their upper-range response rather earlier than would otherwise be the case. The central woofer is actually loaded by a sub-enclosure separate from that for the two subwoofer units and is intended to cover a different range: the two subwoofers handle just one low-bass octave and start to roll out by 50Hz; the middle woofer starts its high-pass rollout above 50Hz and covers the two octaves to 400Hz. The result is a low-frequency drive-unit that can be "fast" enough to reproduce much of music's fundamental tones without coloration, yet has sufficient support in the low bass to convey all of music's weight. Presumably because the mass added to the subwoofer cones lowers their efficiency (footnote 1), the CS5's impedance is arranged to drop to 2 ohms in the low bass so that more current can be sucked from the driving amplifier to compensate accordingly.
The middle woofer crosses over at 400Hz to a 5" Kevlar-coned unit, this time with the cone left virgin yellow—any coating at all apparently degrades this unit's performance. A mere octave and a half higher, at 1kHz, this unit hands over to a second midrange unit, this a 2" convex aluminum-dome type from MB that uses a wide roll-surround to increase its dynamic range. The final drive-unit carries the acoustic load from 3kHz to above 20kHz, this an aluminum-dome unit custom-made for Thiel by VIFA. (Its shallow-flared magnesium faceplate features the Thiel name badge.) All the CS5 cone drivers use copper rings around either the center pole or the inside of the magnet to reduce distortion due to the modulation of the magnetic drive field by the field generated by the voice-coil. In addition, the pole pieces are shaped to minimize any change in frequency response as the coil moves within the magnet gap.
Why so many drive-units? The answer lies in Thiel's registered "Coherent Source" trademark. The only crossover filter slope that allows natural, in-phase integration of the outputs of two drive-units adjacent in passband is the first-order type. This kind of filter, however, only offers a meager 6dB/octave rolloff of out-of-band signals, which means that, even two octaves away from its passband, a driver's output has only been reduced to a still audible quarter of what it was before. Such crossovers therefore result in considerable overlap between drive-units, which can lead to severe lobing or directional problems, and as they are not effective in suppressing out-of-band signals, they do very little in protecting a tweeter, say, from life-threatening low-frequency garbage.
"Yes, but..." I can hear Jim Thiel arguing (footnote 2), "a first-order system is the only type that can achieve perfect phase coherence, no time smear, uniform frequency response, and uniform power response." Which is perfectly true. However, the drive-units for use in such a system must be heroic indeed, as they must be well-behaved well above and below their passbands if audible colorations are not to be the result. They must also feature high power handling because of the lack of protection.
The CS5 crossover is itself also heroic. Constructed on a single hard-wired board, it incorporates 87 elements realized with 114 components. Only—only—55 elements are directly related to the first-order high- and low-pass filter functions, the rest being used to fine-tune the system's time response. The two midrange units, for example, are electrically "moved backward," by the equivalents of ¾" and 3/8" respectively, to bring their acoustic centers into the correct alignment. All coils apart from one are air-cored, and the capacitors are polypropylene and pure polystyrene types, the latter custom-made with tinfoil plates and copper lead-out wires. The internal wiring is a polypropylene-insulated solid-core type.
It might be expected that a loudspeaker at this price level would be supplied ready for biwiring or bi-amplification. Not so, however, the CS5 featuring just the one pair of five-way binding posts on its base. The reason why is explained in the Thiel "White Paper": "Experiments convinced us that the speaker would sound better with one better amp than two lesser amps...Secondly, if the speaker had been made bi-ampable, the connecting straps required for normal, single-amp operation would result in degraded sound in single amp operation. Thirdly, the amount of sonic improvement that can be attained by using two of the very best amplifiers instead of one is a very poor cost/performance value. For the additional...say $10,000, much more improvement could be had by spending this money on the speaker...much better results can be obtained by using two pairs of CS5s, the second pair placed directly behind the first, pointing backward...Fourthly, providing bi-amp, bi-wire capability would also make possible the improper use of external crossover systems, which would drastically degrade the performance of the speaker and make Jim Thiel sick."
Setting up the CS5s is a two-person event, the speakers being awkward to move easily, particularly once their carpet-piercing spikes have been fitted. (Hex-head bolts are also provided for those who do not want to damage their wooden floors.) My 21' by 16' listening room is of wood-frame construction, with a tile-on-concrete floor (covered with a thick carpet), and has 9' ceilings. The walls are slightly unparallel—all walls in Santa Fe are "slightly unparallel"—and there is a 7' by 6' alcove adjoining one of the long walls. This wall is also heavily windowed (it is normally covered with blinds), so that the only way to obtain a symmetrical environment for the loudspeakers is to position them along the other long wall, which is faced with books and LPs. With the CS5s, this meant that the maximum distance I could sit from the speakers with them placed away from the rear wall was just 8', which is Thiel's minimum recommendation. Even then, I found that slight speaker movements resulted in a drastically different low-bass/upper-bass balance.
In addition, I couldn't rid the sound of a slight chestiness, a feeling of congestion in the upper bass. What was happening, I can now conjecture with 20/20 hindsight, is that with its relatively lightweight construction and large windowed area, my listening room acts as though it is much larger to low- to midbass frequencies, the result being that extreme LF performance is over-damped when compared, for example, with a room that is of solid brick construction. As a result, speakers that have true 20Hz extension and are even slightly too full in the lower midrange can easily sound unbalanced in exactly the manner I noted with the CS5s. (I became convinced of this hypothesis thanks to Paul Hales of Hales Audio, whose excellent full-range, two-way speakers also suffered from the same problem in my room.)
But persist I did in trying to get the CS5 to sing in the bass and lower midrange, because I was experiencing some quite astonishing things, particularly in the area of imaging and soundstaging. On my own recordings, I was hearing—no, seeing—things that other speaker systems had only hinted at. As described in the March issue's "As We See It" column, the CS5s enabled me to perceive images, when appropriate, well to the sides of the speaker positions. During the sessions for the next Stereophile LP (of Canadian pianist Robert Silverman turning in a truly man-sized performance of the Brahms F-minor piano sonata) I had recorded Larry Archibald "mapping" the soundstage by talking and clapping as he walked from one side of the church to the other and from the rear to the microphone position. With Kavi Alexander's Blumlein microphone configuration we were using, when Larry was well off to the sides, the out-of-phase signals he produced in the two microphone capsules should result in an image that was to the outside of the speaker position.
Listening to this recording via my usual Celestion SL700s, as Larry walked across to the side, he could be heard to move along a U-shaped path; ie, as his image reached the speaker positions, it then moved back behind them rather than continuing to their sides. Via the CS5s, however, his image mapped out a wide, lateral path well to the sides of the speakers, as predicted by theory.
Footnote 1: Drive-unit efficiency is inversely proportional to the square of the sum of the cone and voice-coil mass and the air-load mass. Thus, doubling the cone mass in effect lowers the efficiency to one quarter of what it was before.—John Atkinson
Footnote 2: Quoted from the White Paper "Technical Information: Thiel CS5 Coherent Source Loudspeaker," available from Thiel at the address given in the specification sidebar.