Floor Loudspeaker Reviews

The other exception I noted to the general euphoria concerns a tendency for upper-range notes in the range of female voice to "pop out," particularly when sung loud. The same characteristic can be heard on orchestral crescendi, particularly if the recording is a bit bright. I'm not complaining, I think, about hearing what's actually on the recording—the CS5s are capable of presenting other recorded aggravations with better evenhandedness. You hear the problem, you note that it's a problem, but this doesn't distract your attention from the music. This upper-midrange, volume-related problem jumps out at you too much to ignore, but it only happens infrequently, requiring the right combination of frequency and volume; the best recording I've found for demonstrating it is the Dorian Greensleeves CD, with Julianne Baird (a favorite test disc of Dick Olsher's, and reviewed in this issue). I would lay the responsibility for this problem at the feet of the metal-dome midrange unit, or perhaps at the midrange cone unit above its crossover point.

Speaking of bad recordings, though, you will be interested by how much information the CS5 provides about the non-musical features that we take for granted in our hobby: tape hiss, record noise, tube noise, record pops and ticks, microphone colorations and distortions. All these accompaniments are laid bare for your examination, exactly as in front of a large lens. You will find that the character of tape hiss varies from recording to recording, for instance; I'm sure if you wanted to take the time you could learn the character of different tape recorders, discriminate between recordings, and learn to pick out tape recorders according to the sounds of their hisses.

Please also observe that the CS5 cannot be characterized as sweet, or even, on its own, as musical—which may be why JGH, auditioning the CS5s in a strange environment and with unfamiliar recordings, found them ultimately underwhelming, despite his being impressed by their neutrality and the excellence of their imaging. His preference is clearly for the Sound-Lab A-3, a speaker with a whole different menu of priorities and excellences (see JGH's reviews of the A-3 in Vol.9 No.6 and Vol.11 Nos.6 & 11). I found the CS5's lack of sweetness and tonal character-of-their-own refreshing—I may be the reviewer at Stereophile most sympathetic to their virtues. My recordings varied all over the place: some were rivetingly dynamic, others were musically compelling, some were dry, others bright. All were enormously detailed; in most cases, the music captured my interest more than the recording and reproduction artifacts, and in every case I came away from the experience with more information than I had before.

Dynamics: This is an area where Thiel expended tremendous effort. One of the purposes of the CS5's total of six drivers per speaker is to ensure that each driver covers a limited enough range to be able to play very loud comfortably, even using 6dB/octave slopes which inevitably stress drivers to the max. The tweeter, for instance, is only 9dB or so down at 1kHz, an area of great musical energy; the midrange driver, whose lower crossover is at 400Hz, still produces quite a bit of the energy from bass guitar below 100Hz.

The result of this work is a speaker whose most notable characteristic is the ability to move effortlessly from soft to medium loud, or medium loud to quite loud, an ability profoundly lacking in most sound-reproduction systems. You end up turning down the volume frequently, since you're used to more muted musical peaks. The CS5 will also play very loud (105dB, say), but without quite the same effortlessness. I would say the IRS Beta betters the CS5 in this area, though marginally. The WAMM more significantly outdoes the CS5, and the IRS V can play a lot louder; these latter two speakers, however, are rare, immensely expensive, and visually overwhelming.

There is a tendency for the CS5s to get brighter at very high volumes, which exacerbates the same tendency in microphones, tape recorders, vocalists, and ears, so that very loud orchestral climaxes (or extra-loud cymbals on rock recordings) are sometimes not entirely comfortable. I'm actually criticizing the CS5s indirectly because of one of their virtues. Their tremendous cleanness, lack of distortion, and effortlessness at lower volume levels lead you to really crank up the volume for maximum effect—you wouldn't want to do it on a lesser speaker.

The CS5's dynamic capabilities are particularly evident in the mid- to upper bass. This is an area where many products are both sluggish and inherently limited; they won't play too loud and they take their time getting even to their limited output levels. Not so the Thiels. They surprised me over and over again with the dynamics available from ordinary recordings. On rock guitar, for instance, they're the only speaker which has reminded me of what good guitar amps can do live. Perhaps in a really loud mix the Thiels would begin to limit (or bottom out), but I couldn't find their limits in my room at volume levels I could stand. I was sufficiently impressed that I insisted Stereophile's Shipping and Receiving Engineer, Danny Sandoval, take a listen to the CS5s. Danny is a bass guitarist himself, and has found every speaker Stereophile has reviewed to be wimpy at the low end, even the IRS Betas; he's our toughest critic of low-end reproduction. Danny really liked the Thiels; they had the impact he was missing from other domestic speakers.

One area of dynamic capability escapes the CS5s: loud volume levels at very low frequencies (below 30Hz). Since my Waveform review last November, I've insisted on hearing how other speakers do on the Mussorgsky Pictures at an Exhibition transcription for organ on Dorian CD (DOR-90117). Played soft, the CS5s do fine; the very low pedal tones are reproduced clearly, and all musical lines are well defined. At anything approaching realistic live levels (100+dB), though, the low notes, which can have a kind of massive solidity, get thick and ill-defined, and the rest of the music suffers as well. In truth, only the Waveforms reproduced this disc realistically, but I hear similar problems from the CS5 very occasionally on other extremely wide-ranging orchestral material.

More typical, at very low frequencies, was its response to the drum drop on Reference Recordings' Däfos. (You know I'm going deep into my record collection when I listen to Däfos—this was my first audition in two years.) D&3228;fos, overall, was reproduced with great realism and sense of being there, with no dynamic limitations on heavy drum passages, except for the moment during "Gates of Däfos" when Mickey Hart knocks over "The Beast," producing a huge low-frequency transient. In response to this, the CS5 woofers simply bottomed out, which Jim Thiel assures me they can do all day with no damage. In my opinion, this is no defect, simply a limitation—one I can put up with when it happens every two years!

I feel some ambivalence comes through the above discussion. In resolution, I will say simply this: from the midbass on up, and at levels below 102dB, the CS5s are dynamically superb, and they will surprise you; I want them to do better at lower frequencies and higher volumes, but I understand why they can't. I will continue to yearn for the same excellence up to 108 or 110dB that they display so convincingly below 102dB.

Soundstaging and Imaging: What can I say—the CS5s are imaging champs. No one who heard them failed to be enormously impressed. Instrumental and vocal images are very sharply defined, both laterally and in front-to-back depth. Imaging position and relationships varied dramatically from record to record.

In terms of soundstaging, which differs from imaging in that you're talking more about the setting than the performers, the Thiels offer precision and, again, lots of information. Some of this you'll find troubling. Most of the speakers I've used in my current listening room have been di- or bipole radiators: for every acoustic wave projected forward, an equivalent (either in-phase or out-of-phase depending on whether the speaker is dipole or bipole) is projected to the back. This delayed rear wave inevitably imposes its own character on the soundstage (if you're lucky, only minimal character is imposed on tonal quality): in fact, the raison d'être of bi- and dipoles is the pleasing nature of this speaker-created ambience. Now, spaciousness is no bad thing—all acoustic events have it, unless you're in an anechoic chamber. And many recordings are made with only a little, which on good systems sounds unnatural. Bi- and dipoles tend to correct for this problem; the penalty you pay is less realistic space, less fidelity, on those recordings which correctly capture the original ambience.

The CS5s suffer not from this problem. You will hear a very detailed character to the ambience on all your records, whether that's reverberant, dry, artificial, multi-miked, or what have you. You will find lots of anomalies, too—strange repositionings of the performers in mid-song or mid-record, lopsided or truncated recording sites (presumably through odd mike placement or pickup patterns), a performer singing in the left speaker and playing his guitar in the right (this from Taj Mahal's De Ole Folks at Home). On good records, such as Stereophile's Poem LP (he said modestly—but, really, the ambience is superbly picked up, even if you don't like flute and piano music), the soundstage reproduction can be uncanny. The CS5s do not defeat the inherent limitations of stereo; the soundstage is in front of you, not around you as in real life, but you observe that soundstage through a huge window—more like a garage door!

This is the first loudspeaker I've heard where the soundstage is frequently smaller than the angle subtended by the speakers. Normally, the soundstage stretches from speaker to speaker, and sometimes outside. With the CS5s, I heard soundstages that were very narrow (usually on mono records), about two-thirds the width between the speakers, and, occasionally, wider than my listening room (sounds would appear to come from beyond the side walls of the room). I can only interpret this much greater level of discrimination as increased accuracy; in any case, I found it very interesting.

Some hints on obtaining this kind of performance: Space the speakers as wide apart as you can, while still maintaining some distance from the side walls; sit fairly close to the speakers to minimize the percentage of sound coming from side-wall reflections (though not closer than 8'—that's the minimum distance at which the drivers' radiation patterns integrate to a coherent wavefront); aim the speakers so their axes intersect about 30" behind your listening position. Thiel recommends aiming the speakers straight ahead, but in my experience this yields a wider, shallower soundstage with less of the interesting record-to-record variation described above (which makes sense: with very flat off-axis radiation it helps to keep the side walls out of the picture by aiming the speakers away from them). As noted under "Spectral Balance," the tonality of the speakers varies remarkably little within reasonable changes of horizontal axis. Of course, if you try what worked for me and don't like it, experiment!

I've heard speakers that provide more realistic soundstages, but only under certain circumstances. The WAMMs, for instance, will more convincingly tell you that you're in a large church, if that information is on the record. The IRS Betas can also do the same thing, but on occasion they will add that large soundstage to recording sites where it's inappropriate.