Blimey im asking a bit today 
So its pretty late and i might not be thinking too straight but here we go...
I've decided it is now time to time align, or at least get the measurements for time aligning my Mid/bass and tweeters, im doing this by whacking both of them in their box, back panel taken off, and the tweeter wired into the left channel of an amp and the mid/bass wired into the right side of an amp
pointing my measurement mic somewhere in the middle of the drivers about 0.5 m back. I then run 2 phase shifting sine waves (3.5k (cross over frq) )one for each channel and freeze the playback of the waves when measurement mic - coupled to PC oscilliscope - tells me output is highest, and then i measure the pahse difference and work out how much i need to offset the drivers.
As far as i can tell this part of my method isnt flawed however...
When i whack the back on the box, add a little stuffing and seal off the tweeter will i then alter the phase response of the 2 cones anyway, making my previous efforts a bit of a waste, or is the phase response of a cone (at 3.5Khz anyway) un-affected by the box its in?
Thanks, Beggar
So its pretty late and i might not be thinking too straight but here we go...
I've decided it is now time to time align, or at least get the measurements for time aligning my Mid/bass and tweeters, im doing this by whacking both of them in their box, back panel taken off, and the tweeter wired into the left channel of an amp and the mid/bass wired into the right side of an amp
pointing my measurement mic somewhere in the middle of the drivers about 0.5 m back. I then run 2 phase shifting sine waves (3.5k (cross over frq) )one for each channel and freeze the playback of the waves when measurement mic - coupled to PC oscilliscope - tells me output is highest, and then i measure the pahse difference and work out how much i need to offset the drivers.
As far as i can tell this part of my method isnt flawed however...
When i whack the back on the box, add a little stuffing
and seal off the tweeter will i then alter the phase response of the 2 cones anyway, making my previous efforts a bit of a waste, or is the phase response of a cone (at 3.5Khz anyway) un-affected by the box its in?Thanks, Beggar
Well, "phase" encompasses a lot, including shifts from driver rolloffs. What you really want to measure to determine driver offsets is the offset in the acoustic center. The best way to do this is to take individual impulse responses from the drivers at a fixed mike position, unwrap the phase curves to get to a minimum phase condition and note time-of-flights, then subtract time-of-flights and multiply that delta-time by speed of sound to get the offset.
This is covered in exquisite detail in d'Appolito's book on loudspeaker measurement.
This is covered in exquisite detail in d'Appolito's book on loudspeaker measurement.
The problem with that is that it conflates phase difference from delay with phase difference from rolloffs. Because one is minimum phase and the other isn't, optimizing at one frequency will not guarantee optimization at others.
It's such an easy measurement to do that there's no need to use shortcuts.
It's such an easy measurement to do that there's no need to use shortcuts.
hey, cool done my measurements
Well i tried to do all 3 methods and failed on one count as i dont actually know how to find the phase response of a cone, i guess i could have written a program to do it, but there's probably already one around somewhere, ah well.
Well my method yeilded a result that at 3500hz the cones were approx 212 degrees out of phase, and the move it fwd and back method (trusting my ear!) hinted at the same thing, so i fliped the tweeter 180 out of phase and hey presto this verified 32 degrees out of phase, means i recess the mid/bass about 8.5 mm into the baffle with a phase reversed tweeter, yey guess im lucky not knowing the phase responses when i bought the cones, wouldn't it suck if the drivers were around 90 degrees out of phase, thats about 5cm recessing at 3500hz 
As always thanks very much for everyone's help, Beggar
Well i tried to do all 3 methods and failed on one count as i dont actually know how to find the phase response of a cone, i guess i could have written a program to do it, but there's probably already one around somewhere, ah well.
Well my method yeilded a result that at 3500hz the cones were approx 212 degrees out of phase, and the move it fwd and back method (trusting my ear!) hinted at the same thing, so i fliped the tweeter 180 out of phase and hey presto this verified 32 degrees out of phase, means i recess the mid/bass about 8.5 mm into the baffle with a phase reversed tweeter, yey
guess im lucky not knowing the phase responses when i bought the cones, wouldn't it suck if the drivers were around 90 degrees out of phase, thats about 5cm recessing at 3500hz As always thanks very much for everyone's help, Beggar
One of the most misunderstood concepts in speaker design is "time alignment." Time alignment doesn't actually involve time aligning the drivers, it involves intentionally misaligning the drivers to offset delay in the crossover. A couple of points are worth clarifying here:
1) most standard (symmetric) filters are designed to sum flat electrically. To get these crossovers to sum flat acoustically, the acoustical centers of the drivers have to be aligned in the crossover region. However, the crossover introduces a delay on the woofer, andso a system (drivers plus crossover) will not be time aligned if the drivers are time aligned (unless you areusing a firstorder crossover).
2) the idea of time alignment (as introduced by Ed Long) is to deliberately move the woofer forward of the tweeter so that the system is time aligned. However, if you do this with a crossover that sums flat electrically, then the system won't sum flat acoustically.
3) in order to get time alignment and a flat summed magnitude response, you have to use an asymmetric filter--higher slope on the woofer than on the tweeter and this crossover will not sum flat electrically. You can look at the following papers in the AES Loudspeaker Anthology for an idea how to do this:
"Loudspeaker Driver Phase Response: The Neglected Factor in Crossover Design" by Marshall Leach Jr.
"Acoustic Alignment of Loudspeaker Drivers by Nonsymmetric Crossovers of Different Orders" by B. Hillerich
"Time Offset and Crossover Design" by Dennis Fink
If you want a truly time aligned system, then you'll have to design a crossover meant for the purpose. If, on the other hand, you're just trying to optimize the setup of your current crossover/driver combination, I wouldn't worry about time alignment because you're not going to get it anyway. It is better to play around moving the relative positions of the drivers back and forth until you get the flatest frequency response possible.
I hope that helps, John
1) most standard (symmetric) filters are designed to sum flat electrically. To get these crossovers to sum flat acoustically, the acoustical centers of the drivers have to be aligned in the crossover region. However, the crossover introduces a delay on the woofer, andso a system (drivers plus crossover) will not be time aligned if the drivers are time aligned (unless you areusing a firstorder crossover).
2) the idea of time alignment (as introduced by Ed Long) is to deliberately move the woofer forward of the tweeter so that the system is time aligned. However, if you do this with a crossover that sums flat electrically, then the system won't sum flat acoustically.
3) in order to get time alignment and a flat summed magnitude response, you have to use an asymmetric filter--higher slope on the woofer than on the tweeter and this crossover will not sum flat electrically. You can look at the following papers in the AES Loudspeaker Anthology for an idea how to do this:
"Loudspeaker Driver Phase Response: The Neglected Factor in Crossover Design" by Marshall Leach Jr.
"Acoustic Alignment of Loudspeaker Drivers by Nonsymmetric Crossovers of Different Orders" by B. Hillerich
"Time Offset and Crossover Design" by Dennis Fink
If you want a truly time aligned system, then you'll have to design a crossover meant for the purpose. If, on the other hand, you're just trying to optimize the setup of your current crossover/driver combination, I wouldn't worry about time alignment because you're not going to get it anyway. It is better to play around moving the relative positions of the drivers back and forth until you get the flatest frequency response possible.
I hope that helps, John
hey john thanks for the advice
Thing is im building up a set of speakers (2 way satellittes and 1 sub) from scratch, which only use the 4th order Linkwitz-Riley Xovers and identical midrange and high amplification lines, so no relative phase difference will be introduced at any point apart from the speakers. For these im after a smooth phase transition at the crossover point (3.5khz) so i dont loose intelligibility (think thats the right spelling/word) you caught me just as i was about to cut some 4cm baffle raising things to mount my midrange cones on. At this moment in time the biggest thing on my mind is baffle diffraction (a slight insight into my somewhat uncomplicated life )
edit: oh yeah forgot to mention last night i realised how dumb i'd been, i though (for some reason) that flipping the tweeters polarity would equate to a 180 degree phase shift, wrong!
Thanks Beggar
Thing is im building up a set of speakers (2 way satellittes and 1 sub) from scratch, which only use the 4th order Linkwitz-Riley Xovers and identical midrange and high amplification lines, so no relative phase difference will be introduced at any point apart from the speakers. For these im after a smooth phase transition at the crossover point (3.5khz) so i dont loose intelligibility (think thats the right spelling/word) you caught me just as i was about to cut some 4cm baffle raising things to mount my midrange cones on. At this moment in time the biggest thing on my mind is baffle diffraction (a slight insight into my somewhat uncomplicated life
)edit: oh yeah forgot to mention last night i realised how dumb i'd been, i though (for some reason) that flipping the tweeters polarity would equate to a 180 degree phase shift, wrong!
Thanks Beggar
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.