I tried the time-alignment method using just the "impulse" in REW's overlays window for my 'pseudo' 5.1mch desk setup... and it's pretty straightforward
once you get the hang of it.
As suggested, I removed all filters -- *other than for the sub and three 'tone filters' above 1kHz for the surrounds (helps match LSR305s to KH120 fronts).
I had to use a 12dB/oct slope for the speakers since the roll-off looked too steep (target mismatch) if I used the 'normal' 24dB/oct slope. Could be caused by placement and room or something else altogether -- I'm too tired to investigate any further.
As you can see above, the actual crossover is way over 120Hz... not to mention the surrounds miserably fail to even xo properly.
As expected, after switching the sub's phase/delay back from max setting to the default zero or 'no delay', I needed to makes changes to all my delays: the sub had to be inverted to match fronts; also, extra delay had to be added to both the mains and sub to accomodate my late surrounds (much further away).
Impulse for the surrounds looks somewhat more uncontrolled (
LSR305 situated right against sidewalls [above head level] with ports plugged). It looks better with the fronts (
KH120), but I had to create a vector average since there's a small time difference between left and right and I wanted a clearer picture of just the average.
The step between fronts and surrounds fits nicely, except for the sub which is a little late -- well, kind of expected really.
The phase is confusing, though:
L&R are well aligned to each other, but there's a large gap (~300 degrees) between them and the surrounds & sub. Ironically, while the phase of the sub and surrounds are closer to each other here, the phase of the SL & SR past 1kHz comes apart and cancels quite a bit when playing a mono sweep. With actual music or mch audio tracks, this should not be an issue. And, regardless of the gap, the subwoofer and front mains sum with no noticeable induced cancellation in the FR around the xo region.
Sub's response looks a bit weird right now -- I know, I'll fix it fixe later -- and that huge null at ~70Hz... Yeah, don't worry, it disappears completely when using my desk's 'standing mode'.
Overall delay for all channels is not too bad!
Relevant mdat files are attached if interested.
*BTW, it looks like 512k length is perfectly fine (2M or longer not necessary) if you're just trying to align the impulse.