Floor Loudspeaker Reviews

The same basic tweeter (with a custom rear chamber designed by Franco Serblin) also produced gorgeous highs in the Sonus Faber Stradivari Homage, so I'm at a loss to explain this seemingly bad behavior, or why designer Diestertich appears to have changed what sounded like the ideal voicing I'd heard at CES. Even novice and casual listeners visiting my listening room described the Caldera's sound as "bright"—regardless of which amps, two solid-state and two tubed, I had in the system at the time (though I don't think "bright" is the word an experienced listener would use). I doubt Diestertich used much vinyl to voice the Caldera, though the problem was clearly audible during CD playback as well.

The offending treble peak aside, the Caldera offered a plethora of the outstanding sonic attributes one expects from an Audio Physic speaker. Like the best of the company's designs, it "disappeared" to leave an expansive, well-grounded soundstage capable of producing appropriate image height, width, and depth. Image stability and focus were exceptionally good, but the peak thinned and obscured three-dimensionality and solidity. With familiar live recordings, it overemphasized air, space, and transients at the considerable expense of the performers' flesh and bones and instruments' bodies.

Like other Audio Physic speakers I've reviewed, the Caldera excelled at transient speed, clarity, and event portrayal. Its rhythmic sense was keen, and events moved through the musical pipeline with satisfactory clarity and certainty. Were the offending sizzle to be tamed, I suspect all of these attributes would remain.

Review samples 057A/B
Audio Physic's reaction when they received the preprint of my original review was that they couldn't recognize the Caldera from either my listening notes or, especially, John Atkinson's measurements. They insisted that the measurements proved that the review pair were "defective," but as both of the pair measured almost identically—see the second measurements sidebar—the proposition seemed difficult to sustain. Nevertheless, given Audio Physic's outstanding track record over the years, and the fact that the original review pair were very early samples (serial numbers 002A/B), we agreed to listen to and measure a second, later pair (serial numbers 057A/B),

It's no small irony that, while listening to that first pair of Calderas, I was also finishing up my monthly column for the UltimateAV website. The subject: How does a reviewer know when a product is defective—as opposed to just not being very good—and at what point does he or she contact the manufacturer about the problem? I invite you to read that column to better understand this common dilemma of audio reviewers.

I was very specific about what I thought I heard from the first pair of Calderas in terms of frequency response. The human ear is extremely sensitive to "low-Q" (broadband) frequency-response deviations of as little as ±0.5dB; a "high-Q" (narrowband spike) deviation, however, needs to be in the ±5dB range before it can be heard. This per Dr. Floyd E. Toole, in whose research we trust ("Audio: Science in the Service of Art.")

What I consistently heard coming from the first pair of Calderas sounded a large-amplitude spike of ca 6kHz. As you can see from the first measurements sidebar, it was a combination of measured phenomena, including a 5dB dip near the 2.7kHz crossover to the tweeter, followed by an approximately 10dB rise between there and 4kHz, with the frequencies shelved up from there 3–5dB relative to the midrange—which, I suggested, sounded superbly smooth.

Whatever that response deviation turned out to be, it was audible enough that my ears could not ignore it. And because the two speakers sounded the same, it never crossed my mind that both of the pair might be defective.

Serial numbers 057A/B looked identical to 002A/B but sounded quite different above the midrange. Gone were the unpleasant high-frequency excess and sense of tonal discontinuity that had invaded every familiar recording I'd listened to through s.n. 002A/B. Every speaker is colored one way or another, and it doesn't take that long to hear it. The trick is to be able to ignore it or "forget" such colorations. A great speaker—by which I don't necessarily mean an expensive one—lets you do that. The first samples of the Caldera didn't come close.

The second pair of Calderas sounded quite different on top. With the tweeter's excess output tamed, the smooth midrange was no longer suppressed and the speaker's overall sound more closely resembled what I'd heard at the Consumer Electronics Show: an unusually rich- yet coherent-sounding Audio Physic speaker.

If anything, the Calderas now sounded slightly mellow—as I'd found Audio Physic's Kronos to sound, which uses the same coaxial drive system. Smooth and inviting yet reasonably detailed and spatially coherent, the second pair of Calderas produced a sound that was, for the most part, in keeping with their high price tag.

Replaying most of the LPs and CDs I'd used to evaluate the first pair of Calderas, I heard many of the positive attributes one expects from an Audio Physic speaker, but without that offending sizzle. The smoothed-out response yielded more genuine detail and greater long-term listenability. Strings went from edgy to silky, and the speaker's exceptionally coherent reproduction of voices was no longer marred by accentuated sibilants. Familiar old reference recordings such as Belafonte at Carnegie Hall (LP), and newer ones such as Arturo O'Farrill's stunning Live in Brooklyn (CD, Zoho ZM200507), once again took on their essential tonal characters.

However, as with the first pair of Calderas, I found the second pair's bass performance somewhat disappointing. While I was willing to allow for speaker/room interaction being the cause of the first pair's somewhat ripe bottom end, now that I've seen the first pair's measurements, I suspect that the second pair, whose bass performance sounded identical to the first's, will measure no better in that area: with the same peak around 40Hz and a rapid rolloff below (–10dB at 30Hz). A noticeable "weight" masked the acceptable but less than impressive extension—especially for a $30,000/pair speaker.

Compared with the first pair's tonal discontinuity, the second pair of Calderas produced an impressively cohesive sound overall. The balance was on the smooth, inviting, rich side of the tonal divide—unusual for Audio Physic—yet the resolution of detail was what you'd expect from a speaker company whose slogan is "No loss of fine detail."

Overall Conclusions
It's unfortunate that the Calderas 002/AB made it to my listening room, and worse that a pair measuring so poorly could have slipped through Audio Physic's quality-control regime. Whether the speakers were early-production units or were assembled in haste for a reviewing deadline, didn't anyone measure or listen to them before they were boxed and shipped? Apparently not—so that, along with my sonic observations and JA's measurements of 002A/B, is, of necessity, part of the story.

But subtract from my review of 002A/B all of my negative comments about prominent treble balance and you have my review of 057A/B: a tonally well-balanced, almost-full-range pair of loudspeakers capable of producing a rich, coherent, generous soundfield—but not a soundfield likely to elicit much short-term excitement. The Calderas' soundstage was less than vivid tonally, or particularly well-defined spatially.

The Caldera's strongest suit was its smooth, natural, creamy, nonmechanical midrange. Its weakest was its softish, underdamped bass, which gave the speaker a somewhat sluggish rhythmic demeanor.

Combine the Caldera's bass character with its rich midrange and its now somewhat polite and smooth top end and you at least have a speaker cut from sonic whole cloth. Combine the Caldera's easy-to-like tonal character, generous soundstage, and high resolution, and you have a credible, luxury-priced, surprisingly compact, attractive-looking loudspeaker more likely to impress and satisfy over the long run than on first encounter.