TOPIC: Active 2 way speaker with shd studio

Hi everyone.
I'm starting my new project of 2 way active speaker with shd studio.
Should i just decide the crossover point (no notch filters, transforms..) and run DIRAC Live to achieve flat reponse?
Thanks

Yes

I'm using SHD for crossover between main speakers (which are full-range so they don't have a crossover) and subwoofers (2.2 configuration).

I tried 2 approaches:
(a) Just set crossover in SHD and then used Dirac.
(b) Used REW to do some correction so that measured response (rather than theoretical response) fitted closely to required crossover. This mainly involved correcting 2 peaks in response at 30Hz and 60Hz caused by room modes.

I found that with (a) Dirac over-corrected because of the room modes. Approach (b) achieved a better overall measured response.

With 4 Dirac slots on the SHD you can try different combinations and compare them.

you should not just set an x-over and hope for magic, the individual responses should add-up correctly, for that to happen the individual phases should align, Dirac cannot do that for you.

a general approach:

• measure each driver individually, same mic position. at least at listening angle, but preferably also off-axis. do a gated measurement to eliminate room influences; this should be the (quasi) anechoic response of the units
• using the DSP make the responses "flat" for at least 2 octaves (more if you intend a low order filter) outside the x-over point. if you see disturbances that disappear under a different angle, than usually these need no correction
• set the correct delay to time-align the units
• apply x-over
• listen
• try different x-over (slope, freq)
• listen
• repeat
• optimise overall response with DSP of Dirac

miket wrote: just like I have. in this case the measurements I describe in the previous post should be made close range (like 1cm in front of the dustcap) of the sub and of the bass unit of the main speaker (in case this is a ported box you'll see the dip from the port, that should be ignored of course), make that flat and proceed ...