Four ways active speakers alignment

The measurements say you have selected L+R as the output, so both channels driven with the measurement signal, and the timing reference output selected is the R channel, which is not what you described.

You have the needed data here for running the 2 drivers together with no XO. Running in that mode the best setting is SB -4 ms to -5 ms.
Using the alignment tool (AT), I found SB -4.5 ms as the best compromise setting and that agrees closely with your AT results #5 (-4 ms) and #6 (-5 ms). Anything in this range is good.
If you intend instead to use a XO it is best to activate it and measure each driver again. It's very possible that the most favorable delay timing will then change.
What I don't understand is that measurements #14 and #15 were apparently actual measurements using these same settings and they do not agree with the AT results. There is something wrong as they should confirm the AT analysis. Is that what your question really is?
The REW settings appear correct so I don't know why there is disagreement. I would confirm the acoustic timing is indeed working properly by repeating several measurement (maybe 5) with no changes on each of the 2 drivers to confirm that each set of 5 impulses overlay each other. The other possibility is that the delay setting in the DSP was not correct for some reason. That is also easy to check by making successive measurements on the sub at 0, 2, 4, 6, 8, 10 ms delays and confirm that the impulse locations have that relative 2 ms spacing between them. If so maybe you delayed the wrong driver.
Maybe I am just missing something?

The relative phase is very chaotic 80-120 Hz due to room effects so so I cannot predict what frequency is likely to be more favorable. It is likely to be a tradeoff no matter what frequency is chosen. The problem areas may just move to different frequencies. The strongest room effect are in the bass driver so the higher XO (120 Hz) is a good choice.

Picture the speaker with all the drivers in place with the front baffle facing to the right and the mic placed an arbitrary measurement distance further to the right of the front baffle. Because the largest driver, in this case your sub driver, will have the greatest depth from the front baffle—and from the measurement mic—use the front pole piece of the sub driver magnet as your zero time alignment point. Assuming that all the other drivers have a shallower depth from the front baffle than the sub driver, the plane of the front pole piece, therefore, will be closer in the time domain to the mic. If you measure the depth of each driver from the front of the baffle to the front pole piece of the magnet on each driver, and do a little bit of arithmetic, you can arrive at an approximate negative delay for each driver to start with and then work to fine tune from there.

There is no difference in the results. The apparent difference in the phase results is the due to the where each impulse is located relative t=0 ms.

If we use AT to move the impulse of SB relative to the sub impulse vs moving the sub impulse relative to the SB then the REW calculation is ends up with a relative shift in the location of the calculated impulse relative to t=0 ms.

Try this:
Go to the impulse chart and place the cursor on the largest impulse peak of AT measurement #4. Activate 'Set t=0 at cursor' in controls. Repeat this operation for AT measurement #5. The 2 impulses are now overlaying each other at t=0 ms. See charts below.

Now look at the comparison of the phase results. They should be identical.

As currently positioned relative to t=0


Overlaying impulses after shifting them.

Use the Overlays window.

kmajjo said:

Thank you for this useful information.
Could you please read the pdf joined ?
I would like to know if my reading is correct.
if you have any remarks, advice, they are welcome.
Br
Jacky

Click to expand...

The aligned sum of 0 ms is indeed correct for the conventional timing alignment. I found 2 other favorable alignments that give the same SPL support in the XO range, but they both create unneeded additional Phase/GD/Spectrogram irregularity at the XO. I posted my mdat analysis FYI.

For example, if we measured an electrical LR-24 XO or acoustic filter responses provided that same acoustic phase result then the conventional alignment would be is found when the start of the rise of the 2 impulses are aligned together. The first peaks will be separated by several ms. The initial rise of the impulse is the best indication of the actual delay. We normally don't have ideal acoustic XO handoff so we pick the closest favorable delay. We do that by finding close phase tracking through the XO range.

In your case the acoustic XO is reasonably symmetrical and not too different to the LR-24 shape so that the best delay was with the initial rise starting at the same time. I posted some teaching exercises Here. You may find them helpful to clarify this. Of course this is a simplification to a complex subject.

SPL 3 Options:


Phase 3 Options:


Group Delay 3 Options:

General steps I recommend to use: (not required - just good practice)

1. Set 1/48 smoothing for both measurements
2. Select the higher frequency driver as first (top) in AT
3. Select the lower frequency driver as second (bottom) in AT [Make adjustments only to this lower frequency driver]
4. Adjust the gain as needed
5. Set FDW 3-6 cycles as needed to see the direct sound phase
6. Place the cursor near the center of XO range and select align phase at cursor
7. Adjust the delay slider manually to confirm the closest overlap of the phase tracking in the XO range
8. Remove the FDW to see if SPL support is still favorable. If not, adjust delay as little as possible to improve it
9. Save the aligned copy of the driver and the aligned sum.

Then invert the lower frequency driver in the AT and repeat the process.

Sometime it is difficult to decide what setting best, so it just takes experience to chose the most favorable setting. I hinted at that in providing the other 2 optional settings that provide similar SPL support. A close review of the other charts helps to identify the most favorable setting.

This is not a fool proof process as room modes can make this very problematic.

Frequency-Dependent Window, the IR window gets narrower as frequency increases.

Frequency Dependent Window
Good Discussion linked within this thread.

The values follow the cursor position - currently positioned at 16.38