Tom, Thanks... just trying to keep up with you :mrgreen: I agree about raising the xover freq. I'm guessing that the JBL drivers just won't play cleanly up to 3.5K so they opted for a lower cutoff, which in turn yielded a thin sound in the front seats, hence the addition of the center speaker for freq band support from 2.5K to 5K. My totally non-scientific by-ear tests (sitting in the back seat, faded to full rear) tend to confirm this. I'm sure EQ would make a difference, and I suppose time alignment could really improve the sound from the driver's seat, but I wonder what the other seats might be like. I guess that's the nice thing about head-units with built-in time alignment, you can flip between presets quickly/easily. As always, it's bang Ã· buck and I think you're getting pretty close to the point of diminishing returns already. I had to drive for about 3 hours yesterday, so I had an unusually large number of discs with me, and I noticed that with some of them I was getting a pretty decent center image, while others were L/R. It seemed to vary with mix and frequency content. I'll see what it's like when I change the speakers (fingers crossed). NuShrike, I don't really get what you're asking here. I know what pink and white noise are, and there is an easier way to describe them: pink noise is just filtered white noise. OK, maybe that's a little too flip (even though it's true). How about white noise has equal energy per cycle and pink noise has equal energy per octave or fraction thereof. Anyway, I know dipping can be heard. What I was trying to say is that if a car system dips in the area of 2.5K it isn't necessarily that bad a thing, or IOW there are worse areas for a dip. Yeah, I wouldn't say identical, but similar - using different mics and analyzers introduces too many variables for a serious comparison, but just for fun I put Tom's results on your chart, centered at 500Hz. Unless I'm reading your chart wrong, I see the 'par' points being a little different - not significantly though. It's a good strategy. It looks like it's working. I think it's cool that you've approached the JBL problem in such a creative way.