Okay, I measured your speakers and here is the deal.
The up firing woofer is too far back of the front woofer. In other words the acoustic center of the front woofer is about an inch back (center of the voice coil) while the acoustic center of the up facing woofer is about 4.25" back. So there is quite a bit more delay involved than if the upper woofer were at the front of the cabinet verses at the back edge of the cabinet.
Here is the effect.
The green line is the front facing woofer. The Red line is the up facing woofer. And the blue line is the two of them together. As you can see there is cancellation going on in two places and a big peak at 3kHz.
So I started dropping parts on it. The parts used can correct for response errors and also shift phase.
This one got a notch filter on the front facing woofer. It corrected it's peak, but the cancellation effects are still there and in the same places. The peak is still there too.
Here is one with a little more of a notch filter on the front woofer. It is the light Blue line. It looks really good on its own. The green line is the up facing woofer with a 12mH inductor in line with it. This makes it do low duty only. Unfortunately it adds even more delay. The dark blue line is the two of them together. The upper woofer is actually cancelling out the lower one some. I could reverse the polarity of the upper woofer to cause it to add a little gain, but then they would be fighting against each other in that air space.
The solution looks like requires the upper woofer to be moved to the front edge instead of the back edge. This will leave only about 1.25" of offset delay. Hopefully the area of cancellation will then be high enough in the frequency range that the limited off axis response will limit interaction or cancellation.