3 way total crossover phase at ~850Hz?

[Brewski]

I'm playing with sims for a 3 way W/TM/W center channel in PCD and keep seeing the best results crossover wise when I have all the drivers wrap in phase close together around 850Hz. My thinking on three ways was it was preferable to have a pattern like \|\|\ where the mid wraps with the woofer and then separately the mid and tweeter then wrap together. Is having all the phases wrap together a problem in a three way crossover?

Take it easy
Jay

[johnnyrichards]

I am no expert, but I think what that means is all three drivers are actually in-phase, at all frequencies. The other example you describe would have the woofer out of phase with the tweeter - I think.

Also, this looks like another example of the vertical line causing some confusion. If it helps, imagine each phase line as a ribbon - which has been cut right in half and dropped down a ways. It has no real significance whatsoever, since if you move it back and forth, it would change. When the vertical lines all line up, it just means phase is identical at that frequency.

Whether it is important to have the woofer in phase with the tweeter is probably arguable to no end, as the frequency overlap is very, very far down in amplitude.

[dcibel]

Where the phase "wraps" around from -180 to +180 is not important in the least. That is to say, no more important than the point where phase changes from 55 to 56 degrees. The reason it looks the way it does is because of the type of plot. Imagine a circle with a line drawn horizontally through the middle, like an equator. At the very left is zero, at the very right is 180 degrees. the values on the top half are positive, the values on the bottom are negative. Where the phase wraps around is the point where it transitions from the top to the bottom.

What is important is the phase relationship between drivers. Look at the phase of your mid and tweeter around the crossover frequency. At whatever point you've chosen as "on-axis", the phase between the two drivers should be as closely matched as possible. Also look about an octave above and below that point and try to get the two lines to "track" as good as possible. You can do the same for the woofer and mid, but keep in mind that the lower you go in frequency the less important this becomes because the wavelengths are longer.

Now, play around with PCD. Use the vertical and horizontal off-axis controls to simulate your design off-axis. Look at it from all angles. Now purposely change something and do it again to see what effect it has - at all angles or whatever angles are important to you. What you see as you increase and decrease the angle is a visual representation of your polar response (provided you entered all the required data correctly).

[Edit: I should add, looking at your images, your phase is pretty well matched, but I don't know what is going on in that transfer function. It is reflected in your frequency response as well.]

[Brewski]

[Edit: I should add, looking at your images, your phase is pretty well matched, but I don't know what is going on in that transfer function. It is reflected in your frequency response as well.]

Dcible,

Hmmm I'll then take another shot at the low pass I was "bending" the woofer harder than it wanted to go to get it in phase going to a smaller inductor will make it much happier but I've got baby duty so this has taken a little longer than normal to post

[Brewski]

Think I'll go ahead and snag some of the 12 ohm buyouts changing the inductor to the woofer brought the woofer's hump from 93 to 92 and I was able to keep imp above 4 ohms which was something I was worried about.

Take it easy
Jay

[Pete Schumacher ®]

Think I'll go ahead and snag some of the 12 ohm buyouts changing the inductor to the woofer brought the woofer's hump from 93 to 92 and I was able to keep imp above 4 ohms which was something I was worried about.

Take it easy
Jay

Just a thought. Relax the woofer slope just a tad, and pad the mids a bit more. Also, relaxing the slope of the mid low pass should help fill in the dip at the tweeter XO point.

[Brewski]

Just getting ready to order the crossover parts but i figured I'd post the current iteration of the center channel. I'd have liked to get the power response to be tighter but was driving the impedance lower than I wanted to go when trying different tweaks.

What I'm not sure about is if the resistors I'm using to tame different knees in the crossover will be subject to strains that would possibly get them excessively hot.

Low Pass
L2 3.0 mH DCR 1.08
R2 2.0 Ohm ???
C2 107uF

Mid Pass
C16 90.0 uF
L16 0.65 mH
R21 10 Ohms ???

L18 0.80 mH DCR .62
C18 8.2 uF
R25 8.0 Ohms ???
L19 1.0 mH DCR .71


High Pass
C9 8.2uF
L9 .15 mH
C10 18 UF

Take it easy
Jay

[DjDisturbed]

If your worried about heating up resistors buy two double the value that you need for each and parallel them . This will double the power handling and probably never get more than warm no matter how long you play them. At the cost of resistors is very low it would only cost a few dollars more for a big piece of mind. :)

I would even buy different values of resistors so you can play around some more once you get drivers mounted. Another good thing to buy is a few low value caps to voice your speakers as well. You may not need them but it's always better to have to many than not enough.

[Brewski]

If your worried about heating up resistors buy two double the value that you need for each and parallel them . This will double the power handling and probably never get more than warm no matter how long you play them. At the cost of resistors is very low it would only cost a few dollars more for a big piece of mind. :)

I would even buy different values of resistors so you can play around some more once you get drivers mounted. Another good thing to buy is a few low value caps to voice your speakers as well. You may not need them but it's always better to have to many than not enough.

Yeah I was thinking of spreading the load on the woofers across as many as 4 or 5 as overkill and then the mids possibly across 3 or 4. I just figured I'd double check since it works in PCD but flies in the face of the calculated values where the slopes are passing.

Take it easy
Jay