Can you be more specific as to what info you want?
There is a wide range of "correct" phasing.
Which option is chosen/accepted is very dependent on the what the system constraints allow and the designers opinion as to what tradeoffs are preferred.
At the base level, I would say that any XO that provides significant SPL support in the XO range is a "correct" one. Many like to reverse the phase of one driver and measure that way to confirm there is a deep null at the XO point or pretty close to it. This is the same looking for the for the most SPL support.
The mic needs to be on the listening axis for the measurements. Hopefully the room interaction will not influence the results significantly. that is very much dependent on the measurement setup and XO frequency of interest.
If distances/delays and/or XO slopes are adjustable then the XO may be optimized further to achieve close phase tracking of the drivers through the XO range. It is not clear how much optimizing the direct sound phase tracking actually adds to the sound quality over the base level however.
That is correct. You should see a significant dip at the crossover point when the drivers are wired incorrectly.
The size of the dip will be determined by the order of the crossover, ie 1st order will typically be 3 dB. second order 6 dB etc.
To be more specific:
The acoustical XO is the point at which the SPL curves of the 2 drivers cross each other. The null that occurs when one driver's polarity is reversed is the point at which the 2 drivers are 180° out of phase. The "distance" or of or the "delay timing" of one of the drivers can be changed to bring the null to the same frequency as the acoustic SPL crossover frequency. I would consider that to be a correct (that is one correct) XO phase alignment.
The null can result anyplace in the XO range initially. It becomes "correct" if it is moved to the SPL XO frequency.
Yes, I will post some charts. It will be later today. I need to make a reverse polarity measurement first as I don't normally use that particular method. Are you more interested in and example using my 70Hz XO or my 1.8kHz XO?
Okay, Interesting. I learned something here. I would be wary of using the method of reversing the polarity on one driver and looking for the null when the XO freq is in the room mode range and the mic is at the LP.
I knew the room modes and mic position for this 70Hz XO could make this method harder to interpret, but it is much worse than I expected.
I use 2 SWs (front/back). I must measure phase relationships at my 14ft LP or the timing will be wrong. The numerous room modes in the XO region make this method unusable in this case. With a different room and test setup it may work out okay. My higher XO with a close mic location worked out just fine.
Based on these measurements I would suggest adjusting the timing such that there is the maximum SPL support through the XO range. This is easiest to evaluate using 1/3 or 1/6 octave smoothing.
The chart below shows my current timing setup. Only my front left channel was used for these measurements. The legend identifies the SW, MW driver and the SW+MW combination. The XO range shown extends from about 31-120Hz for 35dB SPL suppression. The current SPL support with both drivers working together extends across the entire range.
The chart below shows the combined drivers vs the same, but with the MW polarity reversed. I had expected a clear null at about 80Hz where the direct phase of the 2 drivers cross. I assume room modes and reflections are the reason it did not show up as expected?
The chart below shows the impact of the changing the delay on the MW +2ms and also -2ms (still with the reversed polarity). We can see that there is starting to get some better SPL support above or below the XO freq as we move away from the worst case timing condition. The position of a clear null freq is still not apparent however.
Below is the impact on my 1.8kHz XO when measured near field to reduce the impact of room reflections. The null with the reversed polarity is now clearly at the 1.8kHz XO freq. This method works fine in this situation. The null moved up and down as expected as I change the timing (not shown).