Don't know why. When comparing the measured distances transfered to ms to the measured impulse response in ms it's not far off, but i still have to flip the polarity to get a good summation at the xover.
And it wasn't just a small dip, it was a deep null when i measured with periodic pink noise of about 30db right on the crossover point with 1/48th octave smoothing before i flipped the polarity.
Maybe the drivers are out of polarity (although i checked this)? Maybe i need to check polarity again with everything defeated, crossovers, equalizers, time delay, everything.
The installer used BW 12db slopes on the initial tuning and didn't flip the polarity of any speakers and did only a minimum of equalization, so maybe he switched the physical connections on purpose? I don't know.
After all, the alignment tool is too far off in the xover region between midbass/tweeter to be correct imho. 2ms of delay on the tweeter where measured distances to the microphone tip in listening position are 71cm for the left tweeter and 104cm for the left midbass cannot be correct. Jazzis spreadsheet gives a delay relative to the right midbass of about 1.25ms for the left midbass 2.21ms of the left tweeter, 0ms on the right midbass and 0.47ms for the right tweeter. Put an additional 2ms delay on the tweeters on top of that... nope... too much delay. This would mean the left tweeter is 1 cm away from your head.
Good think is that now i'm only "missing" 2db at the midbass/tweeter xover which is the best i've ever achieved i think.
And it sounds oh so good now. Listened to "Chocolate Chip Trip" from Tool after i was done for today. Oh my, that drum really kicked when Danny Carey went crazy on the drums.