Polishing up the Bass

The two most recent posts revealed notable deficiencies in the bass response below 200 Hz.  The left channel has a minor depression around 90 Hz and the right channel has a minor depression around 50 Hz.  The former is more noticeable in the graphs.  However neither of these depressions coincides with a PEQ (programmable EQ) cut, so I can't fix either one simply by dialing back EQ.  And furthermore, nearby PEQ cuts were deliberately made quite narrow so as not to exacerbate them.  They are going to be tough to fix.

Probably the more audible issue anyway are the depressions in the range 120-180 Hz in both channels.  Those are entirely due to the panels, because the subs are nearly entirely cut off in those ranges (except just around 120) with very steep crossovers (phase linear 8th order pseudo Linkwitz Riley) that are already 6dB down at 120 Hz.  In the past I've sometimes boosted the sub response to fix such things, but I've decided it's best to keep the sub down below.  I'm already running the "subs" higher than most people at 120 Hz, but I can do that because my subs have output up to 300 Hz, very rigid cones, and I'm using 8th order filters on them (don't try that unless you also use phase linear Finite Element Response (FIR) filters like I do).  I run the subs to 120 Hz because that way I avoid both the Acoustat diaphram fundamental resonances 40-80 Hz, which fills in the bass at the expense of distortion for those without subs, and the Acoustat wall bounce cancellation in my setup from 100-120 Hz.*

Though the 130-180 Hz depressions are not exactly the same, I think they result from the same issue: dipole bass cancellation.  Dipole speakers have output that is out-of-phase in the back compared with the front.  That is always going to be a problem except for room width dipoles.  Linkwitz always solved it with very judiciously applied EQ which boosts the required frequency range.  I have done the same thing before and plan to repeat it now.

But it doesn't look like my old settings are optimal any more.  I used a boost around 182 Hz in the left channel, and 133 Hz in the right channel.  And yet, in the measured response, the right channel has the depression which reaches from 140-180 Hz and the left channel has a very narrow depression right around 160 Hz.  It looks like I previously got these backwards, but also the work was done from the center room which may be different from the back, which the Movie EQ is intended to optimize, or there may have been other salient changes since I came up with the earlier boosts.

Even with these existing bass dips, the current Movie EQ settings are wonderful I think.  I watched the movie Contact (97) on Sunday, which is full of deep bass, and the sound was full and beautiful, better than I recall hearing in theaters, and with no added "boom" as often happens in theaters.

I dislike adding boosts in my digital system because they can be headroom reducing.  Imagine you have a sine wave that sweeps from low to high very quickly (so as not to damage anything) at maximum level (0dB).  If you have a boosted range, it will digitally clip in the boosted range.  Now you would probably never want to run such a test, but if you happen to be listening at 0dB or nearby, as I often do for movies which have very low average level usually (to allow for big peaks) 0dB transients can sometimes occur in these ranges.

However, IIRC, I deliberately cut all panel inputs to the panel crossover device (a miniDSP-DRC with all digital I/O) by 6dB to prevent clipping due to asynchronous digital sampling.  I truly hate ASRC, it "resamples" a digital signal rather than just passing the data along unchanged, and therefore creates the possibility of digital clipping where there was none before.  I wish all digital were done fully synchronous as the Tact does it using 1999 digital chips, but hardly anyone does it that way anymore because they've all been scared by the "jitter" parade, meanwhile the importance of jitter is vastly overstated, and of no consequence at all for a device which is digital in and digital out.  I think there are still a few chips that support fully synchronous operation over SPDIF without a separate timing line--I use a separate timing line with my Lavry AD converter for truly perfect transfer--but it's simply not used much anymore otherwise--the miniDSP's don't let me select it because they don't have a separate timing line.  The Behringer 2496 DEQ's are also fully synchronous just like the Tact.  But sadly miniDSP decided to do things the "new" way, which introduces the possibility of digital clipping unless you are using their analog outputs.  To be absolutely safe from digital clipping, therefore, you have to reduce the level to the internal digital resampler by 6dB as I am doing.  This is no serious loss in the 24 bit or higher domain that the resample is working in.

Then I re-boost the signal 3dB in the DEQ's, which are fully synchronous.  That leaves me about 3dB headroom in the worst case.  I did things this way to allow for the 3dB boosts I was using before I started developing the Movie EQ, when I zeroed them all out for a fresh start.

So 3dB boost is all I can permit with my current setup.  In previous experience, that can be surprisingly useful, and you'd probably not want to use more than 3dB boost anyway even if you had the digital headroom for it.

I started with the Left channel and started boosting +3dB around 160 Hz with 1/6 octave.  But this created the condition where 160 Hz was higher than the 180 Hz band.  I pushed the PEQ frequency higher and higher, until at 170 Hz the two bands were equal, which also allowed me to broaden the PEQ to 1/3 octave without raising the already elevated band at 140 Hz, a pretty good compromise, and the "suggested EQ boost" fell exactly as much as with 1/6 octave centered at 160 Hz.   There's still a big visible dip, but the lowest point on the dip is not much more than other dips in the response above 200 Hz.  Also notice that the suggested EQ boost at 160 Hz has dropped from +7.6 dB (very bothersome) down to +5.2dB (meh). Don't aim for perfection, that usually ends up horrible, do what you can with limited EQ boosts like 3dB.   I've previously found such small improvements uniformly beneficial when larger or more complicated ones ones can get tiresome.  "Ship it."  

Also note that +/- response numbers vary enormously with filter bandwidth.  I could make this look nearly flat with octave bands, but's that nothing like music.   I use the minimum bandwidth and maximum fft size (which has huge effect on resolution) available to me to see the finest detail I can.  A sine wave generator or MLS generator would be better but requires much more complicated setup.

Left Channel before adjustment

Left Channel with boost at 170 Hz 1/3 octave

Duplicating this filter in the right channel wasn't any good.  It raised the respose at 180 Hz compared to 160 and 140.  140 happens to be already depressed, rather than peaking as it does in the other channel.  So I lowered the PEQ frequency to 160 Hz and broadened it 3/4 octave, which gives a fairly smooth U shaped response in the region centered at 160 Hz, probably the best that can be done with only one 3.0 dB boost, and the suggested EQ boost has dropped over 3dB (probably lucky sampling and less than that, it never falls as much as you have boosted and you're lucky if it's 0.5dB less...on a second measurement I got 4.8dB suggested which would be a 2.7dB drop from 7.5 suggested before adjustment but similar looking spectrum).

Right Channel before adjustment

Right Channel with boost at 160 Hz 3/4 octave

Right Channel with same boost later

Final PEQ settings for today 9-7-26 are

Left EQ 9-7-26

Right EQ 9-7-26

I played the entire album Bass Ecstasy by Bass Erotica at -1.4dB total (-1.3 being applied by Roon's album leveling) in the center back at it was captivating.  3D image, wall to wall, with impressively deep powerful tuneful and never unpleasant bass notes.  I remember when I could hardly listen to this at -10dB because of all the room boom.  I measured 93.1 dBC "fast" peaks on my SPL Pro app at the listening position.  dBC de-weights the deepest bass which will have peaks much above that measured flat.

The Movie EQ sounds great all around the house because of suppressing the primary house modes, and the way the Acoustats couple to entire space.  The Hotseat EQ exaggerates some house modes to get flat response at a close seating position.

(*It's possible the dips I avoid in the panels below 120 Hz by crossing that high are dipole cancellation too, but they are easily fixed crossing over just above that and this also avoids panel resonances and excess excursion, while the 160 Hz dip is not fixable by raising the crossover even higher...it's already as high as it can easily be...or by moving the speakers...at 4 feet from wall they are already out as far as they can be without sacrificing the multipurpose nature of my living room as a place where I talk with people and watch movies with them.  If I had a "dedicated" listening room I might put the Acoustats midway in the room and sit in the back, that's probably the most optimal location for dipoles and I knew a guy who set up his Magnepans like that.  Acoustat suggested minimum 3 feet from wall and I found 4 feet to be considerably better and 4 1/2 feet not any better than that.)