Celestion SL700 loudspeaker 1991 Measurements

Sidebar 3: Unpublished 1991 Measurements

A year after my review of the Celestion SL700 was published, Stereophile acquired a MLSSA measuring system from DRA Labs and a B&K (now DPA) 4006 omnidirectional microphone, both of which the magazine still uses for loudspeaker measurements. As I was still using the Celestion SL700 as my reference at this time, I naturally performed a number of measurements on it in order to get the hang of how MLSSA could be best used. The speaker's on-axis response appeared in the October 1991 issue (p.209), but none of the other measurements have been published before now.

Fig.1 shows the SL700's quasi-anechoic response averaged across a 30-degree horizontal window on its tweeter axis at 48" and corrected for the microphone's own departure from a flat on-axis response at this distance. The dominant feature is the 15dB-high peak at 24.1kHz and the sharp notch between 14 and 20kHz, both due to the aluminum-dome tweeter's "oil-can" resonance. While the peak will be inaudible, the lack of top-octave energy will have an effect on the Celestion's sound quality. This is perhaps why the tweeter itself has been balanced a couple of dB too high in level in absolute terms, resulting in a positive response shelf between 3kHz and 10kHz. Compare this graph with the in-room response in the 1988 measurements section (fig.3), and you'll see why I noted in my 1988 review that "the sound was too forward in the upper midband." However, the upside of this response shape is that it will allow the listener to more easily perceive recorded detail.

Fig.1 Celestion SL700, anechoic response on-axis at 48", averaged across 30 degrees horizontal window and corrected for microphone response, with the nearfield woofer response plotted below 300Hz.

At the other end of the audio spectrum, part of the upper-bass response rise will be due to the nearfield measurement technique. But the combination of a slight energy excess in this region and the relatively slow rolloff typical of a sealed-box alignment results in a good balance between bass definition and extension, as I found in my auditioning.

The shape of the Celestion's impulse response (fig.2) is obscured by ringing from the ultrasonic tweeter resonance, but the step response (fig.3) shows that the tweeter is connected in inverted acoustic polarity, the woofer in positive polarity, which will give the optimal transition between the drivers in the crossover region. The output of the tweeter also leads that of the woofer by about 250 milliseconds. Other than the tweeter's oil-can mode at 24.1kHz, the Celestion's cumulative spectral-decay or waterfall plot (fig.4) is basically free from resonances. However, a slight amount of delayed energy can be seen at the top of the woofer's passband, which might add a slight presence-region "bite" to the speaker's sound.

Fig.2 Celestion SL700, impulse response on tweeter axis at 48" (5ms time window, 30kHz bandwidth).

Fig.3 Celestion SL700, step response on tweeter axis at 48" (5ms time window, 30kHz bandwidth).

Fig.4 Celestion SL700, cumulative spectral-decay plot on tweeter axis at 48" (0.15ms risetime).

I don't routinely measure speaker distortion. There have only been a few models in which noticeable levels of harmonic distortion have been associated with negative review findings. However, as I mentioned in one of my articles on loudspeaker measurements, I conjecture that listeners use overall distortion to set a comfortable playback level. If a loudspeaker has high intrinsic distortion, hence a limited dynamic range, it won't be played as loud. Once the level of harmonic distortion rises above a threshold (probably one that is different for each listener), the listener reaches for the volume-control knob.

Nevertheless, I did perform an analysis of the SL700's non-linear behavior back in 1991. Fig.5 plots the woofer's THD+noise percentage against frequency for three different sound pressure levels at 1m. At the lowest spl (76dB), the distortion drops below 1% above 100Hz, plateauing at 0.2% through the midband. below 100Hz, the THD rises with frequency, reaching 8% at 30Hz. However, this frequency is well below the speaker's cutoff point. The bass distortion has more than doubled when the speaker is played 10dB higher in level, but still drops to 0.2% in the midband. Only at 96dB (which is very loud on pure tones) does the Celestion's distortion rise to what are probably objectionable levels in the midrange, while the poor woofer is definitely in overload below 100Hz!

Fig.5 Celestion SL700 woofer, THD+N (%) plotted against frequency at (from bottom to top): 76dB spl at 1m, 86dB, 96dB.

To look at the SL700's distortion in the treble, I drove the speaker with an equal mix of 2kHz and 3kHz tones, again at 76dB and 86dB spls. (I am not sure 11 years later why I chose these frequencies, as 2kHz is handled by the woofer and 3kHz by the tweeter, minimizing the possibility for intermodulation distortion.) Fig.6 shows a spectral analysis of the Celestion's output at 76dB spl at 1m. While the second harmonic of 3kHz is low in level, at -56dB (0.15%), the third lies at -34dB (2%), which is disturbing. What is interesting is that increasing the playback level to 86dB (fig.7) doesn't increase the level of the tweeter's harmonic products but it does introduce the second and fourth harmonics of the 2kHz tone, as well as some low-level intermodulation.

Fig.6 Celestion SL700, spectrum of 2+3kHz DC-10kHz, at 76dB spl at 1m (linear frequency scale).

Fig.7 Celestion SL700, spectrum of 2+3kHz DC-10kHz, at 86dB spl at 1m (linear frequency scale).

With hindsight, the Celestion's distortion-limited dynamic range was possibly its biggest drawback. Nevertheless, working on these measurements almost a decade after I replaced the Celestion SL700 as my reference with the B&W Silver Signature made me realize how much I had enjoyed using this little speaker.—John Atkinson

89 Doug Brown Way
Holliston, MA 01746
(732) 683-2356