quick REW question

im getting confused with REW time alignment process, whether to use loopback, or to follow this (easy looking)

Conversely I came across advice to sweep the box as-is, full range with no filters or EQ, since what may happen out of passband will influence how you process the signal (ie, addressing nasty spikes, or factoring in natural rolloff)

Norseman wrote: I don't understand that, if we already know where we want to cross over our cabs...surely you want to time align with regard to the chosen crossover frequencies or each cab

If you're talking about actual time alignment, then this should be done without any filters in place. Use the IR rise time (not peak) or the group delay values for each driver (see Charlie Hughes' method for the latter, although you need good clean measured data).

This way, the sources are aligned, regardless of what else is put in place.

Then, your filters should be chosen to provide complementary phase response at your target acoustic crossover frequency. This may require asymmetrical filters - frequency and order.

If you go the other way round, you can be in phase but a lap or more behind. Or align to a reflection, or struggle to get an alignment at all because the phase responses of the drivers and filters are simply not complementary in slope angle.

Don’t try to align the phase response at a single frequency. Time alignment is about ensuring the sources play at the same time.

Once that is in place, you then pick electronic filters which combine with the acoustic phase of each driver to produce a phase-matched response through the desired crossover region. Ideally for at least one octave either side of the target frequency.

Go back over the paper I linked from Charlie Hughes. Use either the group delay outside of the passband, or zoom right in on each windowed IR plot to find the time where it first begins to deviate from zero. That's your offset.

You might need to zoom the y-axis by 10x or more to really see it.

I’m not sure if REW can model various types of filters and orders to do the next part. If not, you might be better off taking the measured response of each source out of REW and throwing them into VituixCAD, which is also free. That will also show you the summed response of the delay and filters, for that given micnpositij.