How would one determine phase correction in a 2-way crossover? (1 Viewer)

[SteveEdwards]

I'm looking into crossover design and the one thing I don't understand is phase correction. I understand what it does, but I don't understand how to locate phase problems between drivers and correct it. It seems to me that crossover design is at least as important as the actual speaker drivers, yet I cannot find a tutorial that touches on phase correction. Still, almost any respectable loudspeaker (even DIY kits) boast premium phase-corrected drivers). For a crossover between a midwoofer and a tweeter, is this something I should be worrying about? Where can I find more information on this? I've seen suggestions of tilting the front baffle 15 degrees, but this just doesn't seem like the right way to do things. If I ever take the plunge and try DIY with expensive ScanSpeak components, I want to do it right!

Thanks as always to all who respond.

 

[Patrick Sun]

Is it a coincidence that "phase" rhymes with "haze"?
You'd probably get better answers by posting your questions over on the 2 main DIY speaker boards:
Parts Express Discussion Board
Madisound Discussion Board
Usually the polarity of the tweeter will correct the phase of the output at the crossover (xo) point, but it's very dependent on the acoustic slope of the xo point. What you want is in-phase response at the xo point, and for most 4th order Linkwitz-Riley xo's, you'll either get good SPL output summation at the xo point between tweeter and midwoofer, or a deep null at the summation depending on the polarity of the tweeter (hooking it up the leads in the same way as the midwoofer is considered "in-phase" and vice-versa is "out of phase").
Each "order" of the filter will give you degrees of "phase" or lead or lag of the angle due to the xo components (inductors and capacitors), I think it's 45 degrees of lag for inductors and 45 degrees of lead for capacitors (I may have that backwards though, curse my sieve-like memory). As you use normal serial/parallel xo topology, the degrees of phase add up and you want the phase to be going in the same direction at the xo point.
Say you have a 2nd order filter on the drivers, with one filter you have +90 degrees of phase, but the other filter you have -90 degree of phase, this would lead to a null at the xo point, but if you reverse the leads on one of the drivers (getting 180 degrees of phase correction), then you'd have the phase of both drivers going in either +90 degrees of -90 degrees, but as long as they are going in the same angle at the xo point, they should sum normally and not cause a null at the xo point.
For a 4th order filter, you get 180 degrees for one filter, and -180 degrees for the other filter, but this is still going in the same direction, so you would most likely not reverse the leads of either driver to have the response sum normally at the xo point. By reversing the leads of one driver, you cause one driver to be 180 degrees out of phase from the other driver, thus causing the null at the xo point.
Just wait until you read up on the minimum phase designs, your head will explode (you can only use 1st order filters for this type of design, and its very tricky to get it right).

 

[Patrick Sun]

BTW, the tilting of the baffle is to align the acoustic centers of the voice coils for both drivers so that they are radiating from the same distance away to your ears. Otherwise, if you use a standard baffle, the sound from the tweeters will arrive at your ears a small fraction faster than the midwoofer because it has travelled a slightly less amount of distance to get to you ears. How much this makes in the overall quality of speakers can be debated, but many who "know what they are doing" do tilt the tweeter back to line up their acoustic centers. I did not for my recent projects (it's a pain to construct the box for the right amount of acoustic center offset).
But if you want to do it, it is simple geometry:
Find the distance the voice coil for each driver from its "face". Say the offset is 0.5" for the tweeter and offset is 1.5" for the midwoofer. That means you need to angle the baffle so that the tweeter winds up 1" "behind" the midwoofer. Depending on the distance between the drivers from each other, that will allow you to find out the angle of tilt using simple geometry (rise/run = slope). Whatever angle that turns out to be is the angle of the tilt.
From the Madisound board on acoustic centers.
One more thing, it's been shown that flush-mounting the tweeter is beneficial in yielding predictable smoother sonic quality due to diffraction effects from the baffle, but it's not as critical to flush-mount the midwoofer, so you may have to keep this in mind when calculating the relative acoustical offset for the tilt calculation. Also, narrow baffles are more desirable than wide baffles (the width of the speaker) - also due to diffraction effects on the sound.