OK, I have finished the first stage of the "whole shebang" project and am at a decision point, seeking input. The measurements posted below were taken with Dirac in bypass and with bass management on.
First, a comparison of the impulse responses:
Compare front and rear sub pairs normal polarity, Dirac on
(from Matt's post yesterday)
Compare front and rear sub pairs normal polarity, Dirac in bypass
(some difference, but still clearly displaying a polarity issue, so Dirac is not the cause):
Now, Compare front and rear sub pairs with front pair reversed polarity
(not sure why the screen shot is smaller) Notice that the impulse response matches now (as expected).
And finally, compare Center+all four subs, normal polarity vs. two front subs with reversed polarity. Clearly not improved!
However, on the good news front, the non-minimum phase shown at 125Hz when the front subs have normal polarity has completely disappeared
when the polarity is reversed.
I need to decide now whether to proceed with a fresh Dirac calibration to see how Dirac handles a 15dB dip at 41Hz. On a side note, the combined sub level as measured in DLCT is now 3dB higher (measuring at -9 on the test tone level scale, as opposed to the target level of -12). I prefer not to re-adjust the subs' gain matching, because that would make reverting to the previous calibration much more difficult.
So, the opinion I am asking for is:
A. Proceed with the Dirac calibration, leaving sub levels 3dB high, or
B. Abandon the exercise, because things look worse with the polarity reversed.