Getting a subwoofer to blend with the speakers is mostly a matter of order: place it, hand it the right frequencies, set its level, then line up its timing. Room correction changes some of these settings, so check them again after it runs, and let a measurement confirm each step.
Placement first
Where the subwoofer stands decides which room modes it excites. SVS suggests starting in a corner, because a corner excites every mode the room has and makes a deep null less likely, though it can also exaggerate peaks. Try the subwoofer crawl: put the subwoofer at your seat, play a steady bass line, and listen at each place it could go with your head near knee height. Where the bass sounds most even is where it belongs.
For even bass across several seats, RP22 says a single subwoofer is not the best way; with only one, it suggests the midpoint of the front wall or a quarter point on a side wall as places to start. In Todd Welti's research at Harman, in a rectangular room with the seats in the centre or towards the back, two subwoofers at the midpoints of opposite walls performed very nearly as well as four at the wall midpoints, with more low-frequency output for each subwoofer. RP22 adds that the subwoofers should be identical, mounted alike, at the same polarity and the same level.
Small speakers, a sensible crossover, a matched level
Set the speakers to Small in the receiver, even large ones, so their deepest bass goes to the subwoofer. 80 Hz is the usual starting crossover: Denon's manual calls its 80 Hz default the setting that works best with the widest variety of speakers, and MartinLogan calls it a good starting point for most systems. Raise it for small speakers, and set the subwoofer's own crossover to maximum or bypass so the receiver does the crossing.
Then match the level with the receiver's automatic setup, or its test tones and a sound level meter. Many people like the subwoofer a little hotter: SVS suggests trying 2 to 4 dB more and warns that beyond that the bass will likely overwhelm the rest, and listening tests by Sean Olive and colleagues, reviewed by Floyd Toole, found a general preference for a room curve that rises toward the low frequencies, with untrained listeners choosing the most bass. That is taste, not a rule.
Near the crossover, the two together should be louder than either alone.
Timing: distance and phase
The receiver's distance settings delay the nearer speakers so every sound arrives together. A subwoofer's own filters and processing add delay, which is why MartinLogan notes its distance may need setting further than the tape measure says, and Denon that automatic setup may report it further away than it is. The subwoofer's phase control or polarity switch is the other half. When the timing is wrong, subwoofer and speakers partly cancel around the crossover and the bass there turns thin.
To see it, measure at the main seat with a timing reference and unchanged settings: the subwoofer alone, one main speaker alone, then both. Near the crossover the pair should be louder than either alone; if it is lower, they are cancelling. Move the subwoofer distance in small steps, then try the polarity switch or phase control. Sound travels about 0.34 m in a millisecond, so a 0.3 m step shifts the timing by just under a millisecond.
Correct, then check
Now run room correction, such as Audyssey, Dirac Live, YPAO or Anthem ARC. It can pull down the peaks that room modes leave at the seat, but it cannot fill a deep null: RP22 lists correcting a cancelling room mode with amplitude filters among the jobs equalisation is not suitable for, and prefers cutting to boosting. That is why placement came first.
Automatic setup measures distances itself, so check the crossover region again afterwards with the same three measurements. The REW Analyzer on this site has a Subwoofer integration view: pick the subwoofer alone, the speaker alone and, if you measured it, both together. It marks the crossover region, and the combined measurement shows whether the two add or cancel there. When both files carry phase, it also predicts the sum with the polarity flipped and with delays of up to 10 ms either way, and holds the prediction back when it misses the combined measurement by more than 2 dB (rms) or a file says it was measured without a timing reference, so you know what to try in the receiver before you change it.
