For starters, there's plenty of books you may want to have read before attempting this. I think many of them will improve your understanding.
Begin by designing the appropriate baffle. What you have right now may not allow for an optimal crossover. Try to keep the drivers extremely close if possible. Round off edges, and offset the drivers (especially the tweeter) so they're not dead center in the baffle. Try to keep the baffle thinner if possible for best imaging.
Next you need something like this to get real T/S parameters, in YOUR baffle.
Next you need to measure the frequency response of the drivers in the baffle. Make sure to measure not just the ON-AXIS response, but the OFF-AXIS response, as far out as 60 degrees IMO. Ideally you want to make it so that the woofer and the tweeter have a similar response off axis near the crossover point as well as on-axis, else you get a wierd shift in tonality near that point, because what we hear is defined by on and off axis sound, both direct and reflected.
I don't know if it's just me, but I prefer higher order crossovers. I know a lot of people say to design so as to use the least amount of crossover components to get smooth response. I understand this idea, especially on a budget. But the best speakers i've heard do use higher order (LR4) slopes, and ideally would use FIR digital slopes. Many programs exist to help you calculate component values
Do try to take advantage of the natural rolloffs of the drivers. If you use a 2nd order electrical slope, it doesn't mean you've got a 2nd order crossover. You may have a fourth order crossover once you factor in the natural acoustic rolloff. So when you say "4th order crossover" it might be a 3rd order electrical filter one way(low pass), and a 2nd order electrical filter the other way (high pass), centered around different frequencies, but giving a symetrical response. Sometimes you may need to wire one driver out of electrical phase with the other, else you will have a huge null near the crossover point. Acoustical phase is what we're concerned with.
Also keep optimal power handling in mind. You don't normally cross the tweeter lower than an octave above its measured fs, and its distortion may be reduced as you go higher in crossover point. Unfortunately since you've already chosen drivers, it's somewhat "too late".
Also don't forget notch filters. The kevlar woofer you are using appears to break up above 4.5khz or so unless I'm mistaken, so it will need notching.
Also remember zobel filters if it's passive. You may be better off trying this out going digital/active, with something like a miniDSP. I think this would be the best approach.
After that, you still may need to build a new box appropriate for the woofer
The final goal should be a flat response, with +/- 2 from 200hz-5khz or so, and +/-3db from the lower limit of the woofer to somewhere in the top octave (IE 13-20khz). Because of this, you will need to implement baffle step compensation as well. If possible, try to obtain a second pair of that woofer to roll in for a 2.5 way to improve sensitivity and power handling, else another approach will be necessary. This is another reason you may be better off going with something like a miniDSP. Additionally, I believe you can import response information from programs like REW into the miniDSP, which will be relevant as it allows you to optimize the BSC.
After that you will need to listen to the speaker and see if it sounds good. Chances are it won't, but c'est la vie
Begin by designing the appropriate baffle. What you have right now may not allow for an optimal crossover. Try to keep the drivers extremely close if possible. Round off edges, and offset the drivers (especially the tweeter) so they're not dead center in the baffle. Try to keep the baffle thinner if possible for best imaging.
Next you need something like this to get real T/S parameters, in YOUR baffle.
Next you need to measure the frequency response of the drivers in the baffle. Make sure to measure not just the ON-AXIS response, but the OFF-AXIS response, as far out as 60 degrees IMO. Ideally you want to make it so that the woofer and the tweeter have a similar response off axis near the crossover point as well as on-axis, else you get a wierd shift in tonality near that point, because what we hear is defined by on and off axis sound, both direct and reflected.
I don't know if it's just me, but I prefer higher order crossovers. I know a lot of people say to design so as to use the least amount of crossover components to get smooth response. I understand this idea, especially on a budget. But the best speakers i've heard do use higher order (LR4) slopes, and ideally would use FIR digital slopes. Many programs exist to help you calculate component values
Do try to take advantage of the natural rolloffs of the drivers. If you use a 2nd order electrical slope, it doesn't mean you've got a 2nd order crossover. You may have a fourth order crossover once you factor in the natural acoustic rolloff. So when you say "4th order crossover" it might be a 3rd order electrical filter one way(low pass), and a 2nd order electrical filter the other way (high pass), centered around different frequencies, but giving a symetrical response. Sometimes you may need to wire one driver out of electrical phase with the other, else you will have a huge null near the crossover point. Acoustical phase is what we're concerned with.
Also keep optimal power handling in mind. You don't normally cross the tweeter lower than an octave above its measured fs, and its distortion may be reduced as you go higher in crossover point. Unfortunately since you've already chosen drivers, it's somewhat "too late".
Also don't forget notch filters. The kevlar woofer you are using appears to break up above 4.5khz or so unless I'm mistaken, so it will need notching.
Also remember zobel filters if it's passive. You may be better off trying this out going digital/active, with something like a miniDSP. I think this would be the best approach.
After that, you still may need to build a new box appropriate for the woofer
The final goal should be a flat response, with +/- 2 from 200hz-5khz or so, and +/-3db from the lower limit of the woofer to somewhere in the top octave (IE 13-20khz). Because of this, you will need to implement baffle step compensation as well. If possible, try to obtain a second pair of that woofer to roll in for a 2.5 way to improve sensitivity and power handling, else another approach will be necessary. This is another reason you may be better off going with something like a miniDSP. Additionally, I believe you can import response information from programs like REW into the miniDSP, which will be relevant as it allows you to optimize the BSC.
After that you will need to listen to the speaker and see if it sounds good. Chances are it won't, but c'est la vie
Last edited:
For starters, there's plenty of books you may want to have read before attempting this. I think many of them will improve your understanding.
Begin by designing the appropriate baffle. What you have right now may not allow for an optimal crossover. Try to keep the drivers extremely close if possible. Round off edges, and offset the drivers (especially the tweeter) so they're not dead center in the baffle.
Next you need something like this to get real T/S parameters, in YOUR baffle.
Next you need to measure the frequency response of the drivers in the baffle. Make sure to measure not just the ON-AXIS response, but the OFF-AXIS response. Ideally you want to make it so that the woofer and the tweeter have a similar response off axis near the crossover point as well as on-axis, else you get a wierd shift in tonality near that point.
I don't know if it's just me, but I prefer higher order crossovers. I know a lot of people say to design so as to use the least amount of crossover components to get smooth response. I understand this idea, especially on a budget. But the best speakers i've heard do use higher order (LR4) slopes, and ideally would use FIR digital slopes. Many programs exist to help you calculate component values
Do try to take advantage of the natural rolloffs of the drivers as well as keep optimal power handling in mind. Don't cross the tweeter lower than an octave above its measured fs, and try to similate its excursion so that you're not bottoming the tweeter.
Don't forget notch filters. The kevlar woofer you are using appears to break up above 4.5khz or so unless I'm mistaken, so it will need notching.
Also remember zobel filters if it's passive. You may be better off trying this out going digital/active, with something like a miniDSP. I think this would be the best approach.
Thanks for your answer.
However, i don't have enough of experience to perform the measurments that you suggest.
I was thinking if someone who has had some experience with these drivers, could give me some advice, what would be the easiest thing to do.
Experience can be obtained. What you need is instruments. This means you either purchase them or find someone locally who owns a pair, or even mail them in a baffle to someone to do them for you. The chances of a person who has already measured the same two drivers in a desirable baffle is unlikely, although good luck. There's too many factors at play there.
Passive Crossover Network Design
Measurements are easy: Download ARTA: ARTA http://www.fesb.hr/~mateljan/arta/download.htm, Make this simple jig: ARTA Jig - http://zobsky.blogspot.com/2008/01/s...t-jig-for.html
Measurements are easy: Download ARTA: ARTA http://www.fesb.hr/~mateljan/arta/download.htm, Make this simple jig: ARTA Jig - http://zobsky.blogspot.com/2008/01/s...t-jig-for.html
Last edited:
(Fine)If you want to go active.
If you want a passive xover (like you first suggested) model the frd/zma curves with the frd consortium software tools (not difficult) and we can make a simulation fy.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.