Loudspeakers

It's natural to think of the LSX as a junior, cut-price LS50W [HFN Oct '17] so how does it square up to KEF's original compact, two-way, DSP active loudspeaker? All our measurements were made via the LSX's analogue input at the default internal sampling rate of 48kHz. Attempts to repeat the measurements using the LSX's optical S/PDIF input failed, possibly because of the LSX's asynchronous sample rate converter. Conventional sensitivity measurement is not meaningful with the LSX, of course, so the on-axis frequency responses [Graph 1, below] have been normalised to 0dB at 1kHz on the left channel. The traces are rather different to those we obtained from the LS50W. Up to 3.5kHz the LSX's response is actually smoother as it avoids the LS50W's dip at 1.7kHz, but thereafter it becomes significantly less even with notable dips at 5.1kHz and 8.8kHz. The rapid roll-off above 22kHz is, of course, due to the 48kHz internal sampling frequency.

Difficulty with KEF's control app prevented 96kHz operation being selected but, on past experience with KEF's small Uni-Q drivers, the response should reach out to above 40kHz, albeit with a high-Q resonance peak at around 38kHz. Because of the treble unevenness, the LSX's response errors are somewhat higher than the LS50W's at ±3.9dB and ±3.0dB, respectively, (we obtained ±2.2dB from both samples of the LS50W) but not excessive by any means. What is disappointing is the high pair matching error of ±2.8dB over the same 200Hz-20kHz range, but this is largely accounted for by the disparity in final octave responses between 10kHz and 20kHz. Below 9.2kHz, pair matching error reduces to an excellent ±0.6dB. In the CSD waterfall [see Graph 2] both of the aforementioned treble dips are seen to be associated with resonances. KH