Floor Loudspeaker Reviews

Editor: Thank you for the opportunity to comment on Mr. Harley's review. I am pleased that he noted the Type B's ability to play at high volume levels "without any sense of strain or congestion," and that "It was especially rewarding to hear full-scale orchestral climaxes reproduced with effortlessness and ease." Mr. Harley also noted that "The B's tonal balance was quite smooth through most of the band. The treble was well balanced to the rest of the spectrum..." I would have to agree, with his system and setup as described, that the Bs had excessive bass and a slight midrange anomaly. I specified to Mr. Harley that the Type Bs, as is the case with all Snell Acoustics loudspeakers, were designed to be used with low output impedance amplifiers.

I also specified that the loop resistance of the cable should be no more than 0.1 ohm. A dynamic woofer and passive filter (crossover) must be terminated with a known impedance. If they are terminated with a significantly higher impedance (the combination of the amplifier's output impedance, and that of the connecting cable) than anticipated, the frequency response at low frequencies will rise dramatically, the speaker will "ring" at low frequencies, and the filter network will not operate at its intended frequency and slope.

Fig.1 shows the error (difference) of the actual measured amplitude response of a Type B front woofer when used with a typical solid-state amplifier as compared with an amplifier with the same 2.5 ohm output impedance as Mr. Harley's reference amplifier, as specified by the manufacturer. Note that a 3dB response rise, centered around 30Hz, is introduced. I would certainly agree that this is too much bass!

Fig.2 shows the measured amplitude response error of the rear woofer. Utilizing an amplifier with a 2.5 ohm output impedance produces a greater than 2dB rise.

Fig.3 shows the cumulative spectral decay of the front woofer with a solid-state amplifier. The cursor indicates that the signal stops at about 20ms at 43Hz. Fig.4 shows the cumulative spectral decay, with the same scale as fig.3, of the front woofer when driven by an amplifier with the output impedance of Mr. Harley's amplifier. Note that the speaker's output at 43Hz continues to about 53ms, more than twice as long as the output continues when used with an amplifier with a low output impedance.

Fig.5 shows the response difference at midrange and high frequencies between a typical solid-state amplifier and an amplifier with the output impedance of Mr. Harley's reference amplifier. With a 2dB error, it is not surprising that Mr. Harley heard a midrange anomaly!

One could design a loudspeaker that is optimized for a high output impedance, but then it would not perform properly with a low output impedance source. Since the vast majority of modern amplifiers have very low output impedances, it makes sense to design speakers for use with that type of amplifier. (For an excellent explanation of the effects of high output impedance or high-resistance speaker wire, please refer to "Why Amplifiers Don't Always Sound Right: Output Impedance & All That," by Robert E. Greene in The Absolute Sound, Issue 71, May/June 1991.)

Mr. Harley states that his room has optimum dimensional ratios for room-mode distribution. I have confirmed that using Snell Acoustics' CARA room-analysis program. Unfortunately, that is not enough. The placement of the speakers and listener(s) within this room is critical with any loudspeaker. Looking at the recommended loudspeaker and listener placement from Snell Acoustics' LEO program, and Sitting Duck Software's The Listening Room, it is clear that the loudspeaker placement used by Mr. Harley would emphasize low frequencies, and this would occur in the same frequency range most emphasized because of the incompatible amplifier output impedance.

Additionally, if Mr. Harley has not changed his listening position since he described it in Vol.13 No.12 (December 1990), it is also incompatible with the goal of hearing the smoothest possible low-frequency response. This might be desirable if small, limited low-frequency capability loudspeakers were in use. While this kind of placement of speakers and listening position increases the quantity of bass, it certainly decreases the quality by making the low-frequency amplitude response much more irregular. When the speakers are set up as Mr. Harley has done, my calculations indicate a 10 decibel increase in amplitude response at 40Hz, compared to the response at the listening position when set up according to our program's recommendation. This is in addition to the large response rise caused by mismatched amplifiers. (For an excellent primer on listening rooms, see Tom Norton's "Enough Room?" in Stereophile, Vol.14 No.10, October 1991.)

I invite readers to listen to the Type Bs for themselves, with compatible electronics and proper room placement. Thank you.—Kevin Voecks, Chief Engineer, Snell Acoustics

John Atkinson responds: Readers should note that, as mentioned in the review, Robert Harley's impressions of the Snell B were gained using a solid-state amplifier, the Rowland Model One, as well as the tubed VTLs.

While respecting Kevin Voecks's more-than-competence as a design engineer, he has somewhat overstated the case regarding the interaction between the amplifier's output impedance and that of the loudspeaker. The VTL 225's output impedance actually measured significantly lower than that quoted by KV or VTL, at 1 ohm at 20Hz (see Vol.14 No.10, p.197), meaning that KV's conjectured response changes in his figs.1 and 2 are exaggerated. Plugging the correct value of 1 ohm into the equation for the response variation, and using my measured figures for the Snell's impedance of 22.8 ohms at 28Hz and 4.7 ohms at 42Hz, gives an apparent bass boost of 1.3dB rather than 3dB. While this interaction exists, therefore, I don't feel it to be "large," to use KV's words, even though, as Martin Colloms mentions in his article on bass reproduction in this issue, the human ear is "more sensitive to small variations in bass level than [it is] in the midrange."

Substituting the correct measured figure of 0.9 ohms in the midrange for the VTL's output impedance means that the 2.5dB error at 300Hz shown in KV's fig.5 is also exaggerated. The correct difference in the Type B's output between 300Hz and 1200Hz when driven by the VTL amplifier is 1.1dB, which, while audible, will not be major, in my opinion.

I therefore don't feel that either the impedance interaction or the specifics of RH's room were the primary reasons for his value judgment that the Snell B has too much bass. (I also noted the B's overgenerous bass during auditioning in the Stereophile listening room, again using solid-state amplification.) As I note in this issue's review of the Spendor S100, a loudspeaker that has its bass tuned for maximal flat low-frequency extension under anechoic conditions will tend to sound bass-heavy in a real room unless that room is extremely large. I suspect that this is the case here.

Because of this possible mismatch between the Type B's bass alignment and its balance in moderate-sized rooms, I have asked Larry Greenhill, whose room is extremely large, to do a "Follow-Up" review on the Type B. In the meantime, as KV says above, readers should listen for themselves.—John Atkinson