From your description it sounds to me like your mix position is in a bass node [where bass waves add in a negative manner resulting in a lack of bass in that position].

Bass will tend to couple to the hard surfaces of the control room shell [also known as the walls, ceilingl, and floor]... so you will often find more bass along the walls than in the center of the room.

While the only way you can tune the bass response of a room is by moving walls... you can add bass traps which will absorb some of the bass energy before it can come back to the mix position as a reflection.

Understand that sound basically travels at 1 ft. per millisecond and that the Haas effect states, reflections that arrive at a listening position in under 19ms are perceived as part of the original signal... so, what seems to be happening is that you have a reflected bass wave arriving at the listening position in under 19ms that is out of phase with the original wave and cancelling.

Either move the mix position to where they are additive [which will give you too much bass at the listening position], or add some bass traps to minimize the amplitude of the reflected signal.

As for the postioning of the subwoofer... invariably they will have some kind of timing issue with the main monitors... try to keep them in the same plane as the main monitors and you'll be a bit better off. It's usually best to use two subs [one under each main in the same plane] but if you have to use one try to: A) get it off the floor; B) put it in the center of the other monitors.

If you put the sub in the center you might want to move it forward a bit so the signal from the sub will arrive at about the same time as the signal from your main monitors... if you're using nearfields this may not be possible as the sub may end up like a foot over your work surface.

Best of luck with it.

arriving at the listening position in under 19ms

Under 40 msec is the accepted value.

Accepted by whom?

Haas states 19ms... if you have some other principle to which you refer please fill us in.

Hass's study is quite old. Every modern source i've ever read (at least 5) says 40 is an upper limit. 30 is where the effect starts to break down, but 19 is way too low. This is easy to test with any delay processor. Try it.