Thiel CS3.5 loudspeaker Page 2
Perhaps my focus on the "problem" in the CS2, now essentially cured, led me to pay inadequate attention to the high-frequency behavior of this tweeter. Its response is essentially flat out to 35kHz (according to information from Thiel), and sounds that way. The upper harmonics of triangle and cymbals are preserved with excellent detail. If I could make a criticism, it would be that as a listener you are made too aware of high-frequency information. This is certainly equipment-dependent. Use anything in the neighborhood of a typically transistory amplifier and you will be driven from the room. Use an old-fashioned tube amp and you will essentially be throwing away the upper range of the tweeter; something in between—a forgiving transistor product, such as the Levinson '20, or an extended tube product, such as the VTL—is called for here.
But the situation—I don't know whether to go so far as to call it a problem—is not simply a matter of level or of accurate reproduction of an inaccurate input signal. Although I don't hear from this tweeter what JA refers to as typical soft-dome spittiness, I do feel there is a mild amount of "splash," where high-frequency transients are spread out in time just a bit, which makes them more apparent than they should be. They don't disappear quite quickly enough, as they do, for instance, on Gordon's Sound Labs or with the Magnepan ribbon. I am not as much of an enthusiast about metal-dome tweeters as some, though the best of them offer very clear advantages over all but the best alternatives in soft-dome tweeters. I'm unconvinced that simple substitution of the metal-dome unit that Monitor Audio uses, for instance, would better the performance of the '3.5, but I do know that it could be improved in this area. I offer this criticism partially to you as users of the '3.5s, so you can carefully match your associated equipment; I also offer it to Jim Thiel as my idea of where his future designs can go.
As would be implied by the flat-to-22Hz spec, the '3.5s are truly impressive at the low end. I would tend to characterize this advantage, though, as one of ease rather than range, and it is true only within some limitations. By ease, I mean that program material without much very low information—which characterizes most orchestral music—doesn't really sound different in terms of tonal balance, but it happens more easily. On the odd occasion that something low turns up, you experience no anxiety and no sense of loss. The limitations have to do with volume. If you want to hear a really loud bass drum, say at 100dB, or a tremendous orchestral crash, the '3.5 runs up against its stops. You will hear it, but there will be strain. Much larger speakers are necessary for this kind of performance with ease.
As far as low end is concerned, the advantage conferred by the equalizer (outside of certain organ and kick-drum fundamentals, which simply aren't there without it) is much more information about the hall character—a kind of "rounding out" of the performing space—and more body to low-sounding instruments: tympani, for instance, or low orchestral strings. There is a significant amount of music, particularly that recorded in a dry acoustic, where inserting the equalizer makes little low-end difference.
Ambience and Soundstage Retrieval: I mentioned above that defining the size and character of the soundstage was one of the places where the CS3.5 betters the '2; it also betters most other speakers in this regard. I think this is due both to better low-frequency extension and to the increased phase coherence of the sealed-box design. If the low-frequency rolloff of the "natural" '3.5 woofer is minimum-phase, which it theoretically is (though I don't believe that complex electromechanical devices ever respond this simply), then the equalizer's effect should be both flat frequency and phase response. In any case, the '3.5's soundstage integrity is excellent.
With respect to the preservation of individual ambiences, direct comparison between the '3.5 and the latest '2s now tells me that perhaps Thiel has done something with the '2 that he would do well to transfer to the '3.5, if possible. The '2 may now be better, though the difference is small, and it would definitely not be sufficient reason to choose the '2 over the '3.5. The '2 practically commands you to pay attention to the space around the performers; with the '3.5, it's just there.
Dynamics: I'm feeling like a broken record, but with the '3.5 you can take my comments about the '2, and simply increase the range by 4–5dB. This is not an insignificant change, however. Now you can listen to orchestral climaxes at levels I would call quite loud; now there are far fewer selections where you feel any limitation. At the same time, Thiel is able to preserve the sonic character of the speaker over the dynamic range. Perhaps you begin to feel a bit of anxiety in the 96–100dB output range, but not much.
This is not to say that dynamics are an area where Jim Thiel can afford to relax in his top-of-the-line product, nor is he. For more than the last year he has been working on the CS5, the "see God" speaker I referred to in a Show report back in 1986. (This odd expression derives from Jim's encounter with one of his dealers who, though happy to sell the '3.5, demanded that Jim work on a product which would "enable me to see God.") One of the CS5's features will be, I think, four woofers of the same approximate size as the '3.5's, but of custom design. I see two challenges for Thiel: manage to retain the phase and tonal balance coherence of his current designs in a speaker that features both multiple drivers and very high output capability (approx. 115dB); and take advantage of this design effort by bringing down to at least the '3.5 the fruits of his efforts, to give it still greater dynamic capability.
This may sound like I'm happy with the '3.5, but can't be pleased entirely—which is true. I think the '3.5 can become a better speaker; a more relaxed dynamic (together with an ultimate tweeter) is the place I'd start.
Putting a line-level equalizer in a high-end speaker is like waving a red flag in front of the audiophile crowd, both press and consumer alike. My response is a bit complicated.
Overall, I have to say that I wish the '3.5 didn't have an equalizer, but retained the satisfying low end the equalizer provides. Part of this certainly stems from prejudice, part from rationality (after all, if I'm unhappy with phono-only preamps that cost less than $1500—half of which are simply equalizers—how can Jim Thiel do one for what must be only a few hundred dollars?), but part of it is a subtle feeling that all is not as well as without the equalizer. Initial listening tests of the equalizer in-circuit and out-of-circuit using the Levinson 26's tape loop led to a null result. I still retained my subtle residual dissatisfaction with the equalizer, but would have been loath to outright condemn it. Further listening on the eve of this writing, though, made it a bit clearer.
Using fairly high-frequency sinewaves (up to 10kHz), I heard some minor alteration of pitch with the equalizer in-circuit, usually a sharpening of pitch, but this would vary with the test tone, and even seemed to vary with whether I was listening on- or off-axis to the tweeter. (Lower test tones made the equalizer easy to hear, because of the bass boost; pink noise was the easiest.) My guess is that 2–10kHz squarewaves would be the best test signal for "hearing" the equalizer, but I didn't have them. On music, the difference could be heard, on occasion, as a slight sharpening of high-frequency tones, almost as if there was a little additional spike added to the spikier sounds. Much music is inappropriate for detecting this difference—which means that the equalizer is innocuous in its HF performance on much music. With other pieces, though, you had to choose between optimum high-frequency naturalness (no equalizer) and optimum hall sound or orchestral fullness (equalizer in).
What should you make of my vague preference for the non-equalized state? Well, choose one of three reasons given above, or listen for yourself. My resolution is to put the equalizer in a tape loop and experiment. In your system or to your ears the equalizer may be completely tame; if so, use it without reservation. Otherwise, use it where it makes the music better. I should add that certainly some of the difference I hear is due to interconnects and the tape loop of the No.26; nevertheless, everyone will have to use interconnects with this device, and mine weren't bad (Levinson and Magnan).
Overall, I have to conclude that the '3.5 is better with the equalizer. I just wish the equalizer itself were a bit cleaner, would let more of the music through; I think it needs to be as good as the electronics probably used in the system: Levinson, Threshold, Krell, Audio Research.
The Thiel CS3.5 is clearly Thiel's best product to date. It is also their best seller, in spite of costing 50% more than the next closest model. If you value the Thiel attention to finish quality and customer satisfaction, and are in sympathy with the Thiel general philosophy about sonic values—neutral, flat frequency response, excellent preservation of soundstage and ambience—the '3.5 is an easy choice.