Floor Loudspeaker Reviews

Before listening to each pair, I ran a series of measurements on all four speakers to see if there was any obvious reason for the conflict of opinion. Fig.1 shows the spatially averaged in-room response of Larry's pair, a curve that correlates quite well with the subjective tonal balance. Though a slightly overdamped alignment leads to the response starting to roll off below 80Hz, it still features useful bass output down to 42Hz, the frequency of the bottom E string of the double bass and bass guitar. While the '1.2 does not compete with the Waveform or Infinity IRS Beta for seismic shock capability, and has perhaps half an octave less extension than the similarly priced Vandersteen 2Ci, it still has reasonable low-frequency performance for what is basically a small loudspeaker, in my opinion.

Fig.1 Thiel CS1.2, spatially averaged, 1/3-octave response in JA's listening room.

Fig.2 shows the frequency response of the four samples measured out of doors at 1m on the tweeter axis, using pink noise and an Audio Control Industrial SA-3050A 1/3-octave spectrum analyzer. Each speaker sat on a 36" stand. From top to bottom, the traces represent serial numbers 2010 and 2009 ((black and red, respectively, ST's pair), and 0624 and 0623 (blue and green, respectively, LA's pair). Certainly small differences can be seen within the 1dB resolution of the analyzer, Sam's pair appearing to have a slightly smoother response through the treble. But in the bass, all four speakers are to all intents and purposes identical, though the slightly overdamped alignment is confirmed. (As shown in fig.1, this is ameliorated when the speaker is sitting on the floor in a room, due to boundary reinforcement.)

Fig.2 Thiel CS1.2, 1/3-octave power response on tweeter axis at 1m with grille on, from top to bottom: 2010 (black), 2009 (red), 0624 (blue), 0623 (green).

My next task was to look at the response of each speaker to a 25µs rectangular pulse, again on the tweeter axis, and calculate the equivalent anechoic response in the midrange and treble using FFT analysis. The impulse responses of the speakers—that for S/N 0623 is typical and is shown in fig.3—were very similar, basically showing a time-coherent reproduction of the pulse on this axis, with both drive-units connected with the same polarity and the impulse tail broken up by ultrasonic ringing from the aluminum-dome tweeter. (Compare this impulse response with that of the Meridian D600 loudspeaker, which uses a version of the same tweeter but which lacks a time-coherent crossover or construction.) Fig.4 shows the step response, calculated from the impulse-response data by DRA Labs' MLSSA program. It is superbly time-coherent, with an excellent right-triangle shape. (Again, compare with the Meridian's step response.)

Fig.3 Thiel CS1.2, impulse response on tweeter axis at 1m with grille on (5ms time window, 30kHz bandwidth).

Fig.4 Thiel CS1.2, step response on tweeter axis at 1m with grille on (5ms time window, 30kHz bandwidth).

The results of the FFT analysis, plotted from 400Hz to 30kHz, are shown in fig.5 with, again, the colored curves again corresponding to samples 2010 (black), 2009 (red), 0624 (blue), and 0623 (green). (Note that these graphs have not been corrected for the response of the measuring microphone, a DPA 4006.—Ed.) With the exception of the exact position and height of the tweeter resonance of each speaker and the fact that the notch centered on 7kHz is a little more pronounced on both of Larry's pair, and that Sam's pair appeared to have slightly less sensitive tweeters, the curves are as alike as peas in a pod. Certainly from this measurement, it looks as though you could take any two from four to make a pair, suggesting excellent QA on Thiel's part. (The slight tweeter peakiness centered on 16kHz in fig.1 correlates with the "knee" in this region, the in-room response.)

Fig.5 Thiel CS1.2, anechoic response on tweeter axis at 1m of samples 2010 (black), 2009 (red), 0624 (blue), 0623 (green).

Bearing in mind, however, that Sam's criticisms concerned a lack of bass, I then measured the nearfield LF extension of each, as well as the way the impedance changed below 200Hz for each, to see if this would throw up any clues. Fig.6 shows a typical impedance curve for the Thiel CS1.2, revealing it to be fundamentally a 4 ohm design. The bass region shows very little change with frequency, though the port tuning is revealed by the dip a shade above 40Hz and the twin peaks above and below. None of the four showed any significant difference in positioning of the peaks and dips, the upper peak occurring at 63, 65, 62, and 62Hz (2010, 2009, 0624, and 0623, respectively) and reaching values of 7.25, 7, 6.1, and 6.7 ohms. The port tuning could be found at 43, 42, 42, and 40Hz, respectively, with an impedance ranging between 4.2 and 4.7 ohms, while the lower peak occurred at 26Hz for all four speakers, reaching heights of 6.5, 6.6, 5.7, and 6.25 ohms, respectively. It could be argued that the impedance in the bass shown by Larry's pair didn't quite reach the values of Sam's pair, but I think this is academic.

Fig.6 Thiel CS1.2, electrical impedance. (2 ohms/vertical div.)

Regarding extension, all showed a similar characteristic in the bass with the measuring mic positioned close to the woofer dustcap. The response was ostensibly flat from 200Hz down to 90Hz or so, with then a gentle roll-out reaching –6dB between 49Hz and 52Hz, depending on the sample. There was a slight difference between the two pairs here, however, in that the woofers of Sam's pair were slightly more damped, starting to roll out a little earlier than those of Larry's pair, with a shallower slope. (The response was –1dB at 77Hz and 85Hz, 2010 and 2009, compared with –1dB at 66Hz and 73Hz, 0624 and 0623). In view of the fact that the reflex port is tuned to 40–43Hz and therefore contributes useful output in the octave between 36Hz and 70Hz, I still can't see that this slight measured difference would explain the degree of the disparity between Sam's and Larry's opinions.—John Atkinson