Back to Square One

I couldn't let it go.  I got back to sweeping the oscillator, which now makes it clear that the 120 Hz depression is still there (despite my having eliminated the nearly useless 140 Hz notch in the subwoofer drive).  I may have made it better but it's still almost as noticeable when sweeping.

Curiously, the 120 Hz depression appears in both the subwoofer response AND the panel response, and it exists only or primarily at the center of the room where the listening position is.  So it's a modal depression that's stimulated by a tall dipole as well as a floor hugging monopole.  Dipoles manage to avoid or at least mitigate most room modes, but curiously not this one, it seems just as bad in either response.  I think, however, it might be coincidental that the dipole-backwall cancellation happens at the same frequency as a room mode.  So it's possible there are different causes for the 120 Hz depression in the dipole response from the subwoofer response.  But the fact that a depression happens in both means I can't boost either one to correct the other.  This problem is NOT correctible by EQ.  Fortunately it doesn't actually seem that large.  It's very localized around 120 Hz and unless I'm sweeping slowly I might not notice it at all (which is why, on the first night, I didn't notice it at times in either panel or sub responses, but actually it's in both).

The "1 kHz suckout" that appears in the right channel  RTA graph seems non-existant in oscillator sweeping.  Response seems audible flat from 250-2.5kHz, and on up through 15kHz where my hearing cuts off.  My system response (as measured by phone) seems to cut off about 22kHz.  That possibly shows the phone is sampling at 48kHz.

I also cancelled the first 100 Hz notch filter in the right subwoofer drive.  I appear to have had broad and narrow notches at 100 Hz, with the broad 1/2 octave notch intended to smooth the broad area around 100 Hz, but that makes the 120 Hz depression worse also.  I do in fact need a deep narrow notch just at 100 Hz but not much higher.

System response may now be about as decent as I can get it with sweeping and a handful of parametric EQ's.  It's probably time to move on to time alignment tests.

One exception is I might see what happens if I eliminate the 32 Hz bulge.  I didn't have that "problem" with the 1+1 speakers, even though in both cases the panels are crossed over at 100 Hz.  It may partly be because I've gotten more agressive with resonances in the 70-100 Hz region.  I used to excuse those as "room curve" but then it seemed like the bass got much cleaner if I eliminated the resonances underneath that curve.  In effect, I've whittled my way back to the bottom of the frequency response curve making it flat, where superficially the added boost remaining may seem helpful but may not be helpful with all music.

It sounded just fine playing the complete organ works of JS Bach played by Marie-Claire Allain today.