Floor Loudspeaker Reviews

Sidebar 3: Measurements

Fig.1 shows the CS2 2's electrical impedance amplitude and phase, measured using the magazine's Audio Precision System One. Like all Thiel speakers, the value varies very little across the audio band. Note how low in amplitude the bass humps are. Apart from the bass, where the radiator tuning is apparent at a low 28Hz, the minimum value is 3.5 ohms at 120Hz, and the maximum 5.2 ohms at 4.3kHz. Although the CS2 2 requires an amplifier capable of driving 4 ohms, its sound will not change significantly in the midrange and treble as the amplifier output impedance changes. A very slight glitch in the amplitude trace can be seen at 300Hz, which indicates some kind of resonant problem at this frequency.

Fig.1 Thiel CS2 2, electrical impedance (solid) and phase (dashed). (2 ohms/vertical div.)

I inadvertently zapped our MLSSA board with static during the preparation of this review, meaning that I wasn't able to look at the spectrum of the cabinet's resonant behavior. Though it appeared to be very well-behaved overall, there is a strong mode present around 315Hz. The sidewalls and cabinet rear seemed reasonably inert to a stethoscope, but this mode could be strongly detected on the front baffle between the midrange unit and woofer. Interrupting the 315Hz warble tone from the Stereophile's Test CD 2 while listening to the baffle with a stethoscope revealed it to add a resonant decay in the ensuing silence. This might contribute to the occasional sense of lower-midrange congestion I noted on some music, though it lies too high in frequency to smear the sound of the tom-toms on Robert Harley's drum track on the second Stereophile Test CD, which sounded very clean.

Fig.2 shows the Thiel's impulse response on the midrange axis with the grille in place, calculated by DRA Labs' MLSSA system using a B&K 4006 microphone calibrated to be flat when used on-axis up to 30kHz (footnote 1). The time-coherent nature of the speaker on this axis is indicated by the relatively slow risetime (this due to the midrange and tweeter outputs arriving at the microphone simultaneously), with then only a minimal degree of overshoot on the other side of the time axis. The ultrasonic ringing is due to the tweeter but is quite well-suppressed.

Fig.2 Thiel CS2 2, impulse response on midrange axis at 45" with grille on (5ms time window, 30kHz bandwidth).

The impulse response can also be displayed as a step response, the output of the speaker when presented with a DC voltage. The ideal shape should resemble a right triangle, a perpendicular step away from the time axis followed by a sloped line back to it, due to the speaker (which is basically a high-pass filter) not being able to reproduce DC.

The step response of the CS2 2 on the midrange axis is shown in fig.3. The shape is excellent, with the initial sharp spike being due to the tweeter very slightly leading the midrange unit at this microphone position and perhaps being slightly too high in absolute level. Moving the microphone up to the tweeter axis gave a double-humped step response (fig.4), with the tweeter's output now definitely arriving at the microphone before that of the midrange. Just below the midrange axis appears to be where the speaker will sound its best.

Fig.3 Thiel CS2 2, step response on midrange axis at 45" with grille on (5ms time window, 30kHz bandwidth).

Fig.4 Thiel CS2 2, step response on tweeter axis at 45" with grille on (5ms time window, 30kHz bandwidth).

Another way of looking at a speaker's time coherence is to examine, not its phase response as such, but the phase deviation left over when that due to the speaker's departure from a flat amplitude response is removed. In a minimum-phase system—one that has just the right amount of phase deviation for its frequency response, an electrical tone control circuit, for example—the phase and amplitude responses are related mathematically by the Hilbert Transform. Subtracting the Hilbert-transformed amplitude response from the measured phase response will leave what is called the "excess phase"; ie, the speaker's departure from a true minimum-phase system.

The Thiel's excess phase on the midrange axis at 45" can be seen in fig.5, meeting ±2° limits from 1kHz to 12kHz and ±10° limits from 600kHz to 18kHz. While superb, this is still less good than the Thiel specification, but it is hard to ensure that the microphone is exactly on the right axis. There will also be an inevitable error in my measurement due to the windowing of the impulse response to eliminate room reflections. (That I didn't get rid of a very small reflection is evidenced by the slight ripple in the excess phase trace.) Nevertheless, this is actually amazing performance for a loudspeaker, especially one designed within reasonably tight price constraints.

Fig.5 Thiel CS2 2, excess phase response on tweeter axis at 45" with grille on.

From the impulse response, the MLSSA system can calculate a speaker's cumulative spectral-decay or "waterfall" plot, which shows how its frequency response changes as the exciting impulse dies away. A perfect speaker would show a straight line (representing its frequency response) that immediately dropped into the floor of the measurement. Fig.6 shows that the first 12dB of the Thiel's die-away is very clean. After that, there are few resonant modes noticeable in the low treble—the cursor is positioned at one of the longer-lasting ones at 4.4kHz, which is probably a residual mode from the midrange cone—but the overall performance is again excellent, contributing to the ease of the CS2 2's sound.

Fig.6 Thiel CS2 2, cumulative spectral-decay plot at 45" (0.15ms risetime).

Fig.6 (which hasn't been compensated for the microphone's departure from a flat response) suggests a slight treble prominence in the 2 2's balance. This is also apparent in fig.7, which shows the midrange-axis response averaged across a 30° horizontal window and adjusted for the microphone's error. There is a slight uptilt to the sound on this axis from 1kHz to 10kHz, which might be expected to add "air" to the speaker's balance but also to veer near brightness, as was found during the auditioning. The slight depressions on this axis, centered on 800Hz and 3kHz, are presumably due to the individual drive-units not quite adding correctly on this axis. As suggested by the step response, fig.2, just below this midrange axis is probably optimal. To the left of fig.7 are shown the responses of the woofer and passive radiator, measured with the microphone almost touching the center of each diaphragm. The woofer can be seen to extend down to 50Hz or so, with the passive radiator covering the octave below that frequency.

Fig.7 Thiel CS2 2, anechoic response on midrange axis at 45", averaged across 30° horizontal window and corrected for microphone response, with the nearfield responses of woofer (blue) and passive radiator (red) plotted below 300Hz and 700Hz, respectively, and the complex sum of the nearfield woofer and radiator responses, taking into account acoustic phase and distance from the nominal farfield point, plotted below 300Hz (black).

Fig.8 shows the changes to be expected in the Thiel's response as the listener moves to the speaker's side. (Only the changes are shown, which is why the on-axis response appears as a straight line.) The 2 2's dispersion is reasonably even up to 30° to its side, with the high treble output rolling off smoothly off-axis. Notches in the speaker's output appear at the crossover frequencies for extreme off-axis angles, but these are irrelevant unless you site the speakers close to bare reflecting sidewalls, in which case the sound might become a little bright.

Fig.8 Thiel CS2 2, lateral response family at 45" with grille on, normalized to response on midrange axis, from back to front: differences in response 90–5° off axis, reference response, differences in response 5–90° off axis.

In the vertical plane (fig.9), the speaker's dispersion is more uneven, as might be expected from its use of first-order crossovers (again, just the changes in response are shown). A standing listener will receive a sound afflicted with severe suckouts at the crossover frequencies (top or rear curve). Sitting on the tweeter axis results in rather a bright balance, while sitting just above the midrange axis accentuates the response uptilt. Sitting on or just below the midrange axis (about 31" off the ground) gives the flattest, smoothest measured response, but sit too far below and the sound will become rather rolled-off in the highs, accentuating the brightness region below.

Fig.9 Thiel CS2 2, vertical response family at 45", normalized to response on tweeter axis, from back to front: differences in response 20–5° above axis, reference response, differences in response 5–10° below axis.

Finally, fig.10 shows how these dispersion measurements add up in the listening room. (To derive this curve, I use an Audio Control Industrial SA-3050A 1/3-octave spectrum analyzer with its own microphone to take 20 1/3-octave spectral responses of left and right speakers individually across a 72" by 20" window centered on my listening seat. I then average these spectra with a slight weighting toward the listening position. This has proven to give quite good correlation with the subjective balance in my room.) The Thiel's bass extends down to below 30Hz in-room but is rather lumpy, the effect of room modes not being entirely removed by the spatial averaging. The midrange is smooth above that region, as is the treble (though slightly downtilted). There does appear to be a slight degree of prominence in the midrange unit's passband which would make the speaker a little fussy in being partnered with amplifiers that are themselves rather bright.—John Atkinson

Fig.10 Thiel CS2 2, spatially averaged, 1/3-octave response in JA's listening room.

---

Footnote 1: See Stereophile, October 1991, Vol.14 No.10, pp.205–206.