Acapella High Violoncello II loudspeaker Follow-Up, October 2010
When I reviewed this $80,000/pair, three-way horn loudspeaker in the September 2010 issue (p.71), for the bulk of my auditioning I had the level controls of the ionic tweeters set to their minimum positions. However, the tweeters were still a little "hot" in my room (ca 26' by 15' by 7' 10") when the speakers were driven by the solid-state Classé CTM-600 monoblocks, and I mentioned in the review's "Measurements" sidebar that it might be worth reducing the tweeter sensitivity with a series resistor inserted in the line-level drive-unit feed. (The tweeter has its own class-A amplifier with a nominal 600 ohm input impedance, so a series resistor doesn't detune the crossover.) I didn't have time to do any serious listening with series resistors for the tweeters before the deadline for that review, but I did do so after the September issue had gone to press and before the Acapellas were shipped back to the distributor.
The blue trace in fig.1 shows the spatially averaged response of the High Violoncello IIs in my listening room, as set up by importer Brian Ackerman and driven by the Classé amplifiers. (You can find a photo of the speakers in my room here.) The region covered by the tweeter is around 4dB too high in level, which projected transients and sibilance a little forward in the soundstage, and emphasized analog tape hiss on older recordings. The green trace shows the effect of inserting a 600 ohm resistor in each tweeter feed. This reduces the level above 10kHz by 4dB, but leaves the mid-treble a little too high in level.
I ended up using 680 ohm resistors with the Classés, which really did properly integrate the speakers' top two octaves in my room. I listened to John Butt and the Dunedin Consort's new, one-singer-to-a-part recording of J.S. Bach's B Minor Mass in a download from Linn Records: 24-bit/88.2kHz FLAC, converted to Apple Lossless with Soundbooth's Max program, so that I could play it with Pure Music and iTunes via USB through the dCS Puccini SACD player. The choral sound was rich and natural, and I never felt that the top two octaves were being pushed forward in the soundstage. The duet of sopranos Susan Hamilton and Cecilia Osmond in the Christe Eleison was to die for through the Acapellas, with natural tonalities and precise stereo imaging.
So if you have a room about the size of mine or smaller, particularly if its acoustics are on the bright or lively side, and you wish to use solid-state amplification with the High Violoncello IIs, it would be worth experimenting with series resistors in the tweeter feeds. However, this would not be necessary in large rooms, where the horn-loaded tweeter's limited directivity will give a flatter high-frequency response.
I also commented in the review on how the Acapellas took to being driven by tubes. The speaker's sealed-box, isobaric woofer alignmentthe two 11" doped-paper-cone woofers you can see on the High Violoncello II's front baffle are internally loaded by a third 11" unitis tuned for clarity and definition rather than ultimate bass weight. The Audio Research VSi60 integrated amplifier (also reviewed in the September issue) I used toward the end of the review period fleshed out the low frequencies, better balancing the speaker's top octaves, which is why I didn't feel the need to experiment with resistors in the tweeter feed with this tubed amplifier.
The red trace below 200Hz in fig.1 shows the spatially averaged response of the Acapellas in my room as driven by the ARC. While the upper-bass region is identical with the two amplifiers, the solid-state Classé monoblocks (blue trace) did indeed produce less mid- and low bass than the tubed VSi60. This is due to the ARC having a much higher output impedance than the Classé, even from the ARC's 4 ohm tap: 1.6 ohms from 20Hz to 1kHz.
Fig.1 Acapella High Violoncello II, spatially averaged, 1/6-octave response in JA's listening room as set up by importer (blue), and with 600 ohm series resistors in tweeter feeds (green), both with speakers driven by Classé CTM-600s (blue); and spatially averaged, 1/6-octave response with speakers driven by Audio Research VSi60 (red).
The solid trace in fig.2 shows the High Violoncello's impedance. It features a minimum magnitude of 4.6 ohms at 134Hz, with a peak of 12 ohms at 38.5Hz and a rising value in the midrange reaching 10.35 ohms at 1kHz. Taking the amplifier's output at 134Hz as the baseline, the resistive-divider interaction between the amplifier's source impedance and the speaker impedance gives predicted boosts of 1.5dB at 38.5Hz and 1.34dB at 1kHz. Fig.3 shows the response of one of the Acapella's woofers, measured in the nearfield, with the speaker driven by the Audio Research VSi60 (red) and a Classé CTM-600 (blue). The differences in levels are indeed similar to those calculated.
Fig.2 Acapella High Violoncello II, electrical impedance (solid) and phase (dashed) with Midrange switches set to "0" (2 ohms/vertical div.).
Fig.3 Acapella High Violoncello II, nearfield woofer response driven by: Audio Research VSi60 (red), Classé CTM-600 (blue).
Why should a relatively small change in level of bass quantity produce a significant improvement in bass quality? First, there is the fact that the "contour lines" in the graph of human sensitivity lie relatively close together below 100Hz; I conjecture, therefore, that it takes a smaller difference in level to produce a noticeable change in this frequency region than it does in the low treble. Second, the tube amplifier's damping factor is very much lower than in the solid-state designs, so it will exert less control over the woofers. In theory, this might be thought a bad thing, but it results in a better perceived balance with the Acapella speaker and its overly sensitive tweeter.
I felt it worth taking the time to explore in more detail these two areas of the High Violoncello II's performance because this admittedly expensive German speaker really is one of the finest-sounding speakers I have had the pleasure of using. It doesn't have quite enough low-bass energy in-room to be included in "Class A (Full-Range)" in this issue's "Recommended Components," but it is definitely in the "Class A (Restricted Extreme LF)" categorywith a bullet!John Atkinson