Originally Posted by chashint
In addition to the recommendations already stated.
I recommend turning the volume on the subs down until audessey sets the sub channel trim closer to 0dB vs -8dB.
If the AVR only has one sub trim set each sub volume separately so the AVR sets each sub to the same trim level.
0dB (-3dB) is the target I would go for.
Keep the mic in the same position.
Andoskyy, it might also be helpful to try walking around the room with lively bass playing before you change anything, to see if you can tell if the bass is substantially better in some parts of the room than in others with your current setup. That might help you get a feel for what you need to do with the placement. If you hear a substantial or dramatic frequency nonlinearity at the sofa compared to other places in the room, you know immediately that your sofa is sitting in the middle of a big fat null or two. In my room I have such situation and had to move my sofa a little forward from my living room wall to fix the problem because I have virtually zero placement flexibility for my subs.
Using two identically configured subs will get you between 3dB and 6dB extra bass gain over one sub, 3dB from the doubling of amplifier power, plus up to another 3dB more for the doubling of driver surface area but only for any frequency where the wavelength is 4 or more times the distance between the subs.
With two subwoofers tightly co-located as shown in your diagram, all bass frequencies gain 6dB, but with the subs widely separated, only the deeper bass gets the full 6dB of gain and the shorter wavelength bass only gets 3dB of gain. This separation of gain factors for differing frequencies actually works to your advantage because room mode nulls tend to interfere more with the mid and deep bass anyway.
Probably you already have 6dB max theoretical gain boost over all bass frequencies with the subs co-located like your diagram shows, explaining why the output level seemed to increase with that placement. The boost in output did not improve the situation for you because the room modes that are stimulated by co-located subs are identical. Whatever frequency response issues exist with one sub in that location do not change at all with two subs in that location.
Usually with multiple subs, they would only be co-located in a large venue such as a concert hall where raw sonic power is more important than frequency response due to room modes (the room modes are very low frequency or nonexistent in a large/outdoor venue). In a home theater it is probably better to split the subs up because room modes in a smaller room tend to dominate the subwoofer response. Placing the subs in locations where they stimulate different room modes tends to even out the response.
I have no idea what your level of education is on this topic. Just in case this explanation is not quite clear to you, I believe what is intended by chashint is to level-match the subwoofers to each other and also to level-match the combined sub output to reference, once the subs are separated from each other.
The -3dB final receiver output trim requested is per subwoofer, with each sub used separately, so that the final receiver output trim with both subs used together comes out close to the suggested target of 0dB receiver output trim setting.
As long as the receiver trim does not peg at +/-12dB (or whatever the max adjustment range is) on the receiver's sub channel, you should be OK, but having the trim at 0dB allows you maximum flexibility in your ability to gain up or down with your temporary sub output level adjust in the receiver controls with your remote, as required for any given program/circumstance. The adjustment range of the temporary trim is limited to the same range as the level cal trim because it uses the same electrical attenuator. That is why your sub (and center, for that matter) temporary trims come out non-zero and change with every cal (as well as resetting to initial cal trim level with every power cycle of the receiver), so you want to set the input attenuators on the subs such that the receiver will initially cal the sub out trim close to zero if possible.
Once you have the subs properly positioned for loud and smooth response, the gain at the MLP might reflect the elimination of a big null that you are finding objectionable, and (especially with the suggested nearfield placement) the resulting new gain be enough to max out the trim in the receiver, so turning down the gain on each sub a little is probably a good idea.
If you try doing two simple room calibrations (do two '3 position calibrations'), with one subwoofer at a time turned on at e.g. AC input for each cal (but both subs connected to receiver sub out and both sub input gains set to 50%), you can record the final sub level cal for each sub and tell how well the output level of each one matches reference with its individual input attenuator set to mid-gain. Leave the microphone in the sweet spot for all 6 total 'positions'.
With that info then you can approximately adjust the subs individually on their input trims so that they level-match to each other and to reference -3dB, then the final post-cal with both on should be nearly 0dB at the final receiver output trim.
Level-matching subs that are in radically different placements is probably a good idea, to smooth out the overall response, even if one sub ends up working harder than the other. You have power to burn with those subs in that room and should not need to gain-match (as opposed to level-match) them, but in the extreme case where one gain control is turned up much more than the other, it is conceivable one sub could begin distorting early during a heavy-duty listening session, so use reason when level-matching. Probably not an issue for you.
Another tip is to try to keep the distance between sub and sweet spot matched so that the delay of each sub output is the same at the MLP. Then you avoid phase cancellation due to placement and do not need to adjust the delay independently at the subs (not really possible with the analog subs you are using anyway since they lack digital delay line and cannot add a straight delay term, only a frequency-dependent one).
Hope this info helps without confusing you. I tend to give too much information in the expectation that more education is usually better than less, and assuming that I actually understand what I am discussing as well as communicating clearly (not necessarily a given!).