Meridian DSP8000 digital active loudspeaker Measurements part 2
However, the use of a short horn surrounding the tweeter dome does increase that drive-unit's directivity above 11kHz or so. In all but very small or very live rooms, this will tend to make the speaker sound rather mellow. As I pointed out earlier, the HF tilt control can be effectively used to adjust the speaker's output in the top three octaves to better match the room, but you still run the risk of emphasizing the presence region. This can be seen in fig.6, which shows the '8000's spatially averaged response in my Brooklyn listening room (footnote 1). The HF control was set to "+1" for this measurement, to bring the top octaves into optimal balance according to my ears, yet this results in an excess of energy in the mid-treble.
Fig.6 Meridian DSP8000, spatially averaged, 1/3-octave response in JA's room.
Lower in frequency, a lack of energy in the 200Hz region is probably due to a combination of room effects and the crossover between the midrange unit and the woofers. The lack of energy in the 63Hz and 80Hz bands is definitely due to my room. The Meridian's bass control was set to "-3" when I took this graph, which sounded right. Nevertheless, the woofer region can be seen to be still boosted by a couple of dB, and there was still strong output at 20Hz.
Despite its use of crossover filters that are realized in the digital domain, which allows optimal behavior in both time and frequency domains, the '8000 is not time-coherent. Fig.7 shows the step response on the tweeter axis. (Although the microphone was 50" away from the tweeter, equivalent to a time of flight of around 4 milliseconds, you can see that the sound doesn't reach the microphone until the 7.5ms mark. The extra 3.5ms are due to the latency of both the A/D converter used to digitize the MLSSA signal and the Meridian's integral DSP.)
Fig.7 Meridian DSP8000, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).
As best as I can tell, examining both this graph and the step responses of the individual drivers (not shown), the negative-going midrange output arrives first, followed by the positive-going tweeter output, then the negative-going output of the woofers. Just before this review went to press, I asked Meridian's Bob Stuart why they had not made the '8000 time-coherent if, in effect, it could be "done for free" in DSP. His response was that this particular crossover topology was the one they had found to sound best, despite its lack of time coherence.
Finally, fig.8 shows the Meridian's cumulative spectral-decay plot on the tweeter axis. As expected, the graph is dominated by the ultrasonic tweeter resonance, but is otherwise extremely clean, something that always correlates with excellent clarity and transparency. Those familiar with my loudspeaker measurements will note that one thing I have not shown is a similar waterfall plot calculated from the output of an accelerometer fastened to the Meridian's enclosure walls. The reason is that there was nothing to show. I could find no resonant modes, either in the woofer bin walls or in the head unit!—John Atkinson
Fig.8 Meridian DSP8000, cumulative spectral-decay plot at 50" (0.15ms risetime).
Footnote 1: Individual 1/3-octave response curves are taken (with an AudioControl Industrial SA-3050A spectrum analyzer) for both left and right loudspeakers at 20 positions covering an approximately 6' by 3' spatial "window" centered on the listening position. The curve shown is the average of these responses, weighted toward the actual listening position.