The change in crossover point Mads explained is notable and explains some of the differences between my measurements and Amir's. I understand the reasoning for not advertising it, but going forward it would be good to have a note on the site for such changes. But at least now I know I didn't screw up my measurements too much
.
I'm pretty sure I reviewed one of the older units, since all my dips are around 2kHz not 2.6. I'd personally take the higher crossover as it won't exacerbate the 2kish interaural crosstalk dip as much with typical speaker positioning, but obviously that depends on how your speakers are setup.
But I'll recuse myself from posting my measurements for comparison now as it is in effect a slightly different speaker. At least when it comes to the nitpicking we do around here.
That said, with regard to directivity plots, I do think the lack of normalization to the on-axis is affecting perceptions a bit too. If you don't normalize to the on-axis, the plot will simply exacerbate the dip that is already present on the on-axis data.
Using Amir's own data, normalized to the on-axis and scaled to be the same size, here's the VituixCAD-generated polar plot:
And without the contour lines:
Dip is still there and a bit less than ideal in Amirs data regardless, but IMO doesn't scream 'DIRECTIVITY MISMATCH!' the same as when not normalized.
This is understood. And I noted it in the review. Using your recommended axis did lower fidelity of the sound field extrapolation but that was limited to higher frequencies. In lower band where we saw the directivity issues, it was just as accurate as the tweeter center. It had the benefit then of showing the reference axis being your recommended one so there would be no objection that I did not use the correct manufacturer reference axis.
I believe Mads is differentiating between the expansion point for the klippel math, which as we all agree should be the tweeter, and the acoustical listening axis, I.E. where the speaker is designed to sound best and the point from which the other graphs are calculated relative to. I.E. The mic should be set to the tweeter as you do, but there's a change to way the listening axis in the klippel software?
Not a huge thing and there's the head in a vice argument, but just pointing it out as it relates to the appearance and interpretation of the data.