Amir has measured a decent amount of both, I don’t recall seeing such.
Ah, I see what you mean -- I thought you meant the lines didn't cross, but i see now that they're off by a fraction of a decibel. Wouldn't worry too much about it though. FWIW, the on-axis, port-open bass matches my nearfield-summed quasi-anechoic measurement pretty perfectly. Mine in blue:
The rest of the measurements match very nicely too if we ignore the bass <200Hz since their spin is a closed port measurement and I measured with the port open.
(Note the LW squiggle at 15khz is an artifact from trying to digitize warkwyn's measurement.)
This also reaffirms my initial belief that the spinorama in KEFs whitepaper is not quite representative of the final speaker. KEF's shows a downward tilt, while the speaker Warkwyn and I measured was flat or even ever so slightly bright. Here was how my measurements had compared against KEFs (mine are the solid lines in this one).
I prefer the flatter profile anyhow, but curious what caused the discrepancy.
Looks like a BBC dip in there.
It is rather odd that KEF has that dip in the 1-4k range on so many of their speakers. It isn't widely advertised but I've heard they test many designs with blind listening panels similar to Harman so maybe they're onto something but I can't help but wonder why they don't just shoot for the typical neutral design. Maybe @jackocleebrown has some insight into this seemingly designed dip?
I really think that hardly qualifies as a BBC dip, and it's possible that very shallow dip looks worse than it is synce warkwyn is using a significantly taller scaling than amir.
In any case do keep in mind that, being a coaxial, this dip is unlikely to be more audible than the ER/PIR/SP dip caused on most non-coaxial speakers around the crossover. It is no worse than the crossover dip in the Genelec 8030Cs, for example.
It also very well possible the 4K peak will be more audible than the dip to some people. My own overwhelming impression of the LS50 metas was neutrality, not any kind of tuning.
Last edited: