That dip is cancellation from the front wall reflection. The back wave from a dipole is 180 degrees out of phase with the front wave. When it reflects off the front wall and back to the speaker, it cancels the front wave at the frequency at which the distance from the speaker to the wall plus the distance from the wall back to the speaker is equal to a wavelength, and you get that dip. In other words, twice the distance to the wall. At a 5' distance from the speaker to the wall, the round trip to the speaker is roughly 10', so cancellation occurs in the 100 Hz region. Monpoles also suffer from cancellation, but since their backwave is in phase with the front wave, it occurs when the distance is equal to half a wavelength. It also is more likely to occur from multiple reflections -- side walls, front wall, floor and ceiling. That's why it's generally recommended that monopoles be positioned at different distances from the side and front walls. A dipole parallel to the front wall will be affected mostly by the front wall reflection, but the null will be very deep. If you angle the speaker in, it will start to illuminate the side wall and you'll get two reflections with shallower nulls. Single panel dipoles are generally use with toe in so the same rule applies as to monopoles, you want to split the difference between front and side walls to reduce the null. (Things actually get a bit complicated at those frequencies since the wavelength is large compared to the room and treating them as specular reflections doesn't really work -- it becomes a matter of wave mechanics. So rooms and positions will be somewhat idiosyncratic in this regard. I'm using my woofer in a somewhat unconventional arrangement against the wall.) Bass trapping can reduce the null by reducing the magnitude of the reflection. Another possible solution is to use multiple woofers at different positions. I just got some Owens-Corning 703 FRK that I'll be using to make corner bass traps. Hopefully, that will reduce the null to the point at which it can be equalized out -- otherwise, I'll have to look at multiple driver solutions.
|