Thiel CS3.6 loudspeaker Page 2
Hallmarks of Jim Thiel designs have been sophisticated crossovers, and the CS3.6 is no exception. The 25-element network, implemented with 38 components, not only provides first-order low-pass/bandpass/high-pass filtering for the drivers, but also corrects for driver-response anomalies. The crossover also has a notch filter at 2kHz to attenuate the woofer's fundamental resonance at that frequency. Nearly all the capacitors are polypropylene—the few electrolytics aren't in the signal path—and all are bypassed with custom-made polystyrene and film types. Inductors are air-core types wound with low-oxygen wire.
The enclosure is no less impressive. A 3"-thick front baffle is mated with 1"-thick sidewalls and back panel, reducing cabinet resonances. Five internal horizontal braces further add to the CS3.6's enclosure rigidity. Cabinet vibrations can be a major source of coloration in dynamic loudspeakers; the acoustic output of a vibrating surface is a function of the surface's excursion and area. Given the very large wooden panels in big loudspeakers, it's easy to see how even small excursions can create audible artifacts. The cabinet-radiated energy is not only spectrally colored, it is also delayed in time. These effects induce boxy colorations and smear transient detail. Although not as elaborate as the nearly $2000-more-expensive Hales System Two Signatures (which use a 4" baffle and 2"-thick sidewalls), Thiel has paid considerable attention to making the CS3.6's enclosure inert.
The sloping front baffle aligns the drivers' outputs so that their outputs arrive at the listener's ear at the same time. A major design element of the CS3.6 is the curving baffle and flush-mounted drivers. These techniques eliminate any sharp edges near the drivers, reducing diffraction. Diffraction is a re-radiation of energy, typically from cabinet edges. This re-radiated energy, delayed in time by a few milliseconds, combines with the direct signal to generate constructive and destructive interference. This can be seen as tiny ripples in the loudspeaker's frequency response. Diffraction also seems to blur image focus, reduce soundstage depth, and make the listener aware he is listening to loudspeakers and not music.
The CS3.6 is clearly the result of a Herculean design effort. The custom-designed drivers, elaborate crossover, and sophisticated enclosure all point to a serious effort at advancing the loudspeaker art (footnote 3). The overall result of these techniques—custom drivers, sophisticated crossover, and careful attention to the cabinet—is extremely flat frequency response. Indeed, the CS3.6 claims a typical response of 30Hz-10kHz ±1dB. This is the range of virtually all musical information; if the claim is confirmed, it represents an extraordinary achievement.
Finally, I must comment on the CS3.6's beautiful build, solid construction, and elegant appearance. The high level of craftsmanship in the CS3.6 is readily apparent.
The CS3.6 displaced my reference loudspeaker, the Hales System Two Signature, during the auditioning. I've had the CS3.6es set up for about two months, using them to audition other components and for pure listening pleasure. After some experimentation, I placed them 48" from the rear wall and 34" and 24" from the sidewalls. The listening chair—which puts my ears at 36"—was midway between the loudspeakers and 10' back.
It didn't take long to realize that the CS3.6 needs a beefy solid-state amplifier to sound its best. The VTL 225W Monoblocks sounded extremely good in the mids and treble—to the point of being a synergistic match—but didn't have the low-end control and current output the CS3.6es seemed to demand. I thus ended up using the Mark Levinson No.23.5 for virtually all the auditioning.
Ancillary components included a heavily modified Well-Tempered Turntable and Arm resting on a Merrill Stable Table. The AudioQuest AQ7000 (and, later, the AQ7000nsx) cartridge was amplified by a Vendetta Research SCP2B, which fed an Audio Research LS2B line stage. A 20' run of balanced AudioQuest Lapis provided connection to the No.23.5.
Digital sources included the Mark Levinson No.30 processor fed by the No.31 transport, the Meitner IDAT, and a Meridian 263 processor (to be reviewed next month). I used a single run of AudioQuest Sterling from the No.23.5 to the Thiels.
Footnote 3: There are many more design touches I haven't gone into. For a full technical description of the CS3.6, see Thiel's excellent and comprehensive technical paper, available at no charge from Thiel.