A dip in response can have a couple of causes and it's best to understand what is going on before you move to a solution. One cause is due the phase cancellation. You can eliminate that one by measuring the sub on its own with the crossover running it up high enough to see if the dip disappears.
Sometimes a dip is simply related to the modal issues of your room, and this is the most likely. In that case, the solution is to measure the transfer function of your room. You place your sub with the mouth around the listening position, then move the mic and measure every spot in which you could use a sub. Even consider unusual locations, because with DIY you can put a sub in the ceiling potentially, if that location works. Now once you've done that you will see a stack of plots, and you can look over them to see if you can find positions that when used together, will fill in your dip.
One way to integrate a big sub like the F20, is to place it for maximum LFE gain, and ideally avoiding dips below 40 Hz. Then you place other smaller subs around the room to cover above 40 Hz, as many as you need for a smooth response.
I've written about this at length in a 3 part series on bass integration:
As Bill says, bass traps won't fix the problem. Even with very large ones that show an improvement down to 25 Hz, you still need to EQ because they tend to do a better job at fixing the time domain, than the frequency response. Why? My best theory on this so far is that bass traps tend to have a billiard table kind of effect. Bass traps are like the padded sides of the billiard table. EQ is a bit like the speed at which you fire off the ball. The bass traps get many chances to damp the standing waves, which would seem to explain why they can actually work better than expected. If you try to treat speaker boundary interference with traps, what do you find? It does very little, most likely because the absorbers only get one shot at it. However, let's consider that you have a 44 Hz peak from two walls 4m apart. It's a primary room mode. In 150 ms it will have done 13 laps, so you might say that the trap is now 13 times more effective at damping the mode, compared to SBIR. Over that period of time with decent bass traps, you can expect to see quite a bit of improvement.
Someone may argue that the reduction in ringing is a reflection of the EQ effect of the bass trap. I suspect this is not the case, but it's an idea worth testing.