I am attempting to tune a crossover by ear. Yes, no software! For an 8 ohm woofer, and 6 ohm tweeter, I have built (thus far) a textbook 3rd order on the woofer, 2nd order on the tweeter at 2.2k.
I was told that I would have to at leat add 66 milliseconds of phase correction due to the choice of slopes. Just how do I start to address that?
I was told that I would have to at leat add 66 milliseconds of phase correction due to the choice of slopes. Just how do I start to address that?
I see you have gotten no answer. I don't have an answer either, frankly, but might I suggest you rephrase your thread title?
"Call me stupid but... " might not elicit an answer, but something like "Phase Compensation For Passive Crossover Networks?" might.
I can only say this. If you need 66 milliseconds, it won't be achievable by driver offsets or slanted slopes on the front. At 13,500 inches per second, it would require a driver offset of 891 inches to achieve 66 milliseconds of Time Delay.
It seems to me that many loudspeakers use slopes like yours without any phase compensation, but I am not so expert in the matter to say you don't need it.
"Call me stupid but... " might not elicit an answer, but something like "Phase Compensation For Passive Crossover Networks?" might.
I can only say this. If you need 66 milliseconds, it won't be achievable by driver offsets or slanted slopes on the front. At 13,500 inches per second, it would require a driver offset of 891 inches to achieve 66 milliseconds of Time Delay.
It seems to me that many loudspeakers use slopes like yours without any phase compensation, but I am not so expert in the matter to say you don't need it.
Okay, it's a bit early in the morning for my brain to be doing anything like this, but I'll have a go.......
The reason that you have been told to create a phase correction is due to the difference in group deley between the two filters at the crossover frequency. In good digital crossovers you can automatically insert group delay compensation, but in passive networks your only way to realign the drivers is to physically offset them (probably by recessing the tweeter in your case).
But, 66ms is a darn big delay - that'll be an offset of about 66 feet between the drivers, which I'm guessing is probably bigger than the room the speakers are going into. My guess (I haven't done the sums though), is that the delay you need is actually 66 microseconds. That would equate to an offset of about 20mm (that's a very rough guess - you might want to work that out properly). I hope that helps a little. I also hope I've got that about right.
A few other things to just mention :
With different order crossovers it might make it a lot harder to get a flat response at the crossover frequency. You might also want to pay careful attention to the combined phase response 'cos that might make things sound a bit strange too. Tuning by ear is fine for same order filters, but using different orders and not modelling them first you're making life very difficult for yourself unless you've done this many many times before.
My personal opinion (which is frequently wrong) on time alignment is that there are more important things to worry about. Unless you are exactly on-axis all the time, there's always going to be a vertical path difference between your ears and the acoustic centres of the drivers. So, unless the group delay at the crossover frequency is huge (unlikely) or you're using horn-loaded speakers, then I'm not sure it's worth worrying about too much.
The reason that you have been told to create a phase correction is due to the difference in group deley between the two filters at the crossover frequency. In good digital crossovers you can automatically insert group delay compensation, but in passive networks your only way to realign the drivers is to physically offset them (probably by recessing the tweeter in your case).
But, 66ms is a darn big delay - that'll be an offset of about 66 feet between the drivers, which I'm guessing is probably bigger than the room the speakers are going into. My guess (I haven't done the sums though), is that the delay you need is actually 66 microseconds. That would equate to an offset of about 20mm (that's a very rough guess - you might want to work that out properly). I hope that helps a little. I also hope I've got that about right.
A few other things to just mention :
With different order crossovers it might make it a lot harder to get a flat response at the crossover frequency. You might also want to pay careful attention to the combined phase response 'cos that might make things sound a bit strange too. Tuning by ear is fine for same order filters, but using different orders and not modelling them first you're making life very difficult for yourself unless you've done this many many times before.
My personal opinion (which is frequently wrong) on time alignment is that there are more important things to worry about. Unless you are exactly on-axis all the time, there's always going to be a vertical path difference between your ears and the acoustic centres of the drivers. So, unless the group delay at the crossover frequency is huge (unlikely) or you're using horn-loaded speakers, then I'm not sure it's worth worrying about too much.
You are right, microseconds! Typo on my part. My question is this - can I add resistance in parallel to the tweeter to compensate for 66 microseconds?
I have been cautioned against using a 3rd order on the woofer and second order on the tweeter, but the woofer requires a third order rolloff to avoid some nastiness at 2600 hz and above, AND because the particular tweeter I am using ends up with a 3rd order acoustic rollof when using a second order filter. So wouldn't that make for a good match?
I have been cautioned against using a 3rd order on the woofer and second order on the tweeter, but the woofer requires a third order rolloff to avoid some nastiness at 2600 hz and above, AND because the particular tweeter I am using ends up with a 3rd order acoustic rollof when using a second order filter. So wouldn't that make for a good match?
The only thing I would add right now is that for a 66 microsecond delay between drivers, one driver only has to be 0.891 inch behind the other.
Normally, you think of the tweeter as being offset. But I am not sure why this is the way to go. I might also point out that because of the depth of the woofer cone, I wonder if you don't already have that offset.
Anyway, it is 0.891 inch for an offset that would equal 66 microseconds.
Good luck!
Normally, you think of the tweeter as being offset. But I am not sure why this is the way to go. I might also point out that because of the depth of the woofer cone, I wonder if you don't already have that offset.
Anyway, it is 0.891 inch for an offset that would equal 66 microseconds.
Good luck!
No, you are not wrong if you want to go with the electronic approach. I was just giving you an option.
This kind of thing is beyond my experience, but I do know that it is done. Reading reviews of high end equipment in Audio magazine, many of their high end speakers had delay networks and phase compensations in their crossovers.
This kind of thing is beyond my experience, but I do know that it is done. Reading reviews of high end equipment in Audio magazine, many of their high end speakers had delay networks and phase compensations in their crossovers.
Linkwitz is NOT stupid.
and active filter designs,
that tries to adress the phase correction issue.
Linkwitz Lab - Loudspeaker Design
You will find a lot of articles, papers and circuit & crossover
information at his site. He is a Legend in his own lifetime
when it comes to these matters.
---------------------------------
17 - Active Crossover Networks for Noncoincident Drivers.
The spatial separation between drivers in a loudspeaker system affects the radiation pattern over the frequency range where more than one driver contributes to the total acoustic output. An analysis of conventional crossover networks shows that the main lobe of the radiation pattern shifts in direction and increases in amplitude. A new network transfer function, which can easily be realized with operational amplifiers, eliminates this problem. - e.g. xo12-24.gif , xo12-24b.gif
Additional active delay networks are used to compensate for offsets in the acoustical planes from which the individual drivers radiate. - allpass.gif
The audibility of phase distortion is investigated with the conclusion that it is undetectable for the proposed types of networks.
(This paper is the origin of the Linkwitz-Riley crossover.)
-------------------------------------------
Delay correcting filter - allpass.gif
Just some samples of his material.
/halo
S.Linkwitz has done a lot of thinking, writingLNeilB2 said:
and active filter designs,
that tries to adress the phase correction issue.
Linkwitz Lab - Loudspeaker Design
You will find a lot of articles, papers and circuit & crossover
information at his site. He is a Legend in his own lifetime
when it comes to these matters.
---------------------------------
17 - Active Crossover Networks for Noncoincident Drivers.
The spatial separation between drivers in a loudspeaker system affects the radiation pattern over the frequency range where more than one driver contributes to the total acoustic output. An analysis of conventional crossover networks shows that the main lobe of the radiation pattern shifts in direction and increases in amplitude. A new network transfer function, which can easily be realized with operational amplifiers, eliminates this problem. - e.g. xo12-24.gif , xo12-24b.gif
Additional active delay networks are used to compensate for offsets in the acoustical planes from which the individual drivers radiate. - allpass.gif
The audibility of phase distortion is investigated with the conclusion that it is undetectable for the proposed types of networks.
(This paper is the origin of the Linkwitz-Riley crossover.)
-------------------------------------------
Delay correcting filter - allpass.gif
Just some samples of his material.
/halo
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.