My views on tweeters and crossovers
Posted: Fri Mar 07, 2008 10:00 amThe following is a discussion I had with some technical folks on another forum. I thought I'd post it here for the enduring value:
-----------------------------------------------------
This is a fantastic discussion; one that I am glad to be a part of.
Before I continue, I have virtually no home audio experience. Studio, yes (I am a studio musician), and 22 years worth in the car audio environment, but not so much with home audio, other than purchasing a pair of satellites and connecting my 12” powered subwoofer to my 1990 Phillips 4X oversampling CD changer, and 1994 125 watts X 4 amplifier. So what I am about to say has virtually no application to home audio. I’ll explain why as I proceed.
User 1 says:
The first part of this pair of sentences may be right for the majority of 7" midbasses, but the last part is not, in my opinion. First, the part that you’re probably right about: most 7” midbass don’t do so well out to 4,000 Hz. The Hybrid Audio Technologies Legatia L6 does, as does a select few other midbass, but most do not. Why? Because of cone distortion, edge modes, and cone break-up. This is usually related to cone material, and whether or not the cone allows for a smooth low pass filter, or a jagged one which results in considerable cone break-up (no midbass is pleasant to listen to in its cone break-up mode). Exacerbating the situation is that the midbass is 60+ degrees off axis, and the beaming effect at high frequencies. But there’s a trade-of. The second part of the above sentence is wrong, IMHO, because of the following:
This I will unfortunately disagree with 100%. The brain cannot localize sounds in the vertical plane nearly as well as in the horizontal plane due to head shadowing and HRTF. So this is a non-argument. Why do most properly set-up kick panel cars, with speakers mounted at your ankles, stage at dashboard level? It’s because of head shadowing and precedence effect. Thank God our ears are at the sides of our heads, instead of top to bottom, because if they were top to bottom, stacked vertically, the vertical boundary would no longer be bound by head shadowing (but the horizontal boundary would). So this is not a good argument for crossing tweeters lower, and then placing them higher in the door or dashboard to raise stage height.
By putting tweeters high in the door, dash, or a-pillars and playing them into the midrange frequencies does do one thing though (besides the obvious phase shift at the crosspoint and phase distortion), it splits the vocal spectrum between four speakers, instead of two. With tweeters playing to 2,000 Hz, you will now split the imaging cues between four drivers…about ¾ of the tones will emanate from the midrange down in the door, and ¼ from the tweeter up high. In this instance, there will be no defined center image in the vehicle. Why? Because imaging cues are critical to 6,000 Hz or so, as defined by the size and shape of your head and torso…once again, the head shadowing phenomenon. No, what you will have is your male vocals coming from the center of the car (or wherever your center image happens to go, based upon time and amplitude considerations), and your female vocals will be to the far sides of the doors, dash, or a-pillar, wherever you have mounted the tweeters. This is because ITD helps to solidify the lower and middle midrange frequencies, up to around 2,000 where ITD is most notable. Assuming these frequencies are in phase, ITD will "fool" your brain into thinking they are dash height and are centered. But above 2,000 Hz, ITD is a non-issue, and IID and HRTF takes over, so time is not on your side with respect to tweeters mounted up high. Since IID and HRTF is dominant at your proposed crossover frequency of 2,000 Hz, you will have three or more center-images. This doesn’t include instruments as well. Guitars will be centered, bell trees, triangles, and high-hats will be on the doors. The fundamental tones of drums, particularly a snare drum, will be centered, and its ambience effect will be at the doors.
I have tuned hundreds of car with tweeters mounted up high, and midbass or midrange down low. I can usually tell within 1/3 octave of where the tweeter’s crossover point is because if the tweeter crossover is improperly set, and is too low, I will hear three or more center images.
In summary, there’s a trade-off here. My belief is that you find a 7” midbass that can play to 4,000 Hz without cone break-up, and make this entire discussion a non-issue. But if you want to use a midbass that cannot play to 4,000 Hz, then the tweeter *must* be mounted right next to the midbass. No question about it. There’s not a computer program that tells me this, Klippel is of little use…it’s is pure, unadulterated experience tuning hundreds of cars, both for competition and daily listening. Back to User 1's comments:
I agree. I didn’t realize this was going to be a major factor in this discussion. But it is at least as important as other factors, such as phase response and absolute polarity. They are all inter-related in a very big way, and they are worth discussing.
I won’t argue with this…why? Because you’re no longer in a near-field environment, you are no longer at 60 or 75 degrees off axis (but rather 0 degrees on axis), beaming is no longer a consideration (because you’re on axis with the speakers), power compression with respect to tweeters playing low frequencies in a large room is MUCH different than it is in a car, and etc. I could write three pages how User 1's point of view on home audio is right, because he is right, but only because of the room, the listening environment, the phase, the axis of listening, and etc. The above has no application in a car audio environment, if nothing more from the perspective that power compression of a tweeter in the far-field sense is not a concern, and you’re listening to the speakers at a theoretical 0 degrees of phase.
Once again, not to be contrary, but you are describing a far-field system with lower power compression and lower air impedance in the home environment. And I never said that tweeters are not “able” to handle the low crossover point. They can in fact “handle it”, otherwise people would be blowing stuff up. But just because the tweeters can handle it doesn’t necessarily mean it sounds good in a car environment. Most subwoofers can play a 500 Hz tone, but that doesn’t mean that it sounds good at 500 Hz.
In a car? I doubt it. IMD for a midrange at 4,000 Hz is a non-issue. It is an issue at 2,000 Hz with a tweeter though, especially onhe placed up high in the car with no acoustic damping, because of lack of suspension to provide restorative force, and the lack of damping at those frequencies to reduce IMD. In the Linkwitz information you presented, generation of intermodulation products & harmonics is in the list, as is phase response and absolute polarity.
And in your argument, you mention impulse response. Unfortunately, impulse response works in my favour, not yours. By having all of the vocal information being emitted from one device, the midrange, the impulse response is unity. The moment you cross the tweeters into the vocal spectrum, the impulse response of the tweeter no longer has the vocal information in unity; the tweeter will emit those tones to your ears 2-5 ms before the midrange does, even further exacerbating the situation. No to mention the phenomenon know as zero delay plane, which will undoubtedly occur with the tweeter being mounted so much closer to you. So not only is there going to be a phase shift at the low crossover point, you’re going to have phase distortion, a misalignment of the zero delay plane, tweeter power compression, tweeter intermodulation distortion, and a split sound stage with three or more centre images.
Perhaps a bit strong. I am not usually one for blanket statements. Ok, I will revise: “95% of the time, the tweeter should not be crossed over lower than 4,000 Hz in a car environment. When it is crossed over lower than 4,000 Hz, the tweeter must be located within a distance of 1/4 wavelength at the crossover point from the midrange, and even still this is not an ideal situation considering phase shift at the low crossover point, phase distortion, tweeter power compression and air impedance shift, poor modal response, poor polar response, tweeter intermodulation distortion, and making the tweeter considerably more susceptible to mechanical deformation.”
User 2 wrote:
I couldn’t agree more. I wrote one and a half pages above on the serious negative effects of crossing a tweeter over too low in a car, but at the end of the day, the decision is based upon your own two ears, listening in a car. Not on a sound board, not in a living room, not on the computer screen evaluating parameters and test data, but in the car with the sensitive testing instruments and database of sounds known as your ears and your brain.
User 3 wrote:
The first part of your statement is your own personal philosophy, and seeking advice of those that you trust. But sadly, the second part is dead wrong. Since the vocal spectrum between 2,000 and 6,000 Hz is dominated by IID, HRTF, and precedence effect, the ear readily hears these differences, phase shift, IMD, and etc. Even if the ears couldn’t evaluate these things, you’d still hear multiple center images if the tweeters were separated from the midranges at those frequencies.
I am not always right, and I learn something each day. But I am pretty confident about this subject, because I have tested my theorems in hundreds of cars (yes, physically sitting in cars, tuning, in the near-field environment). Everything I have said above has no application to home audio, for the reasons discussed. As for the remainder of the arguments, I am all ears. But I would like to suggest something. I have a team of guys combing the country hitting shows. If you want to hear first-hand what I am talking about, we can do a bench-scale experiment with high crossover points and low crossover points, in a car. The way to learn is to hear for yourself. I would be honored to be given the chance to do a “before and after” evaluation with whomever might be interested, to show the real-world effects of a high tweeter cross-point and a low one. I have 20 cars at my disposal, all around the country. We can sit, listen, and evaluate without conjecture, philosophizing, or theorizing. You let your ears decide.
Oh, and before I forget, none of the above was meant to bolster Hybrid Audio Technologies products or attributes. There are a few brands out there that make good mids for use in wide-bandwidth. No, I wrote the above because I find this to be one of maybe four of five main issues with car audio today. Many people don't seem to really grasp zero delay plane, tweeter power compression and IMD, vocal cues as they interelate with IID and HRTF and split sound-stages, tweeter verses midbass impulse response, application of on-axis far field information to car audio, and etc. It has been the subject of literally hundreds of hours of research for me. How many of you sit for hours at a time just listening to tweeters in a car, playing with crossovers and evaluating? Yep, that’s me, I’m a car audio dork, and of all the things I have researched, I have researched this subject the most.
-----------------------------------------------------
This is a fantastic discussion; one that I am glad to be a part of.
Before I continue, I have virtually no home audio experience. Studio, yes (I am a studio musician), and 22 years worth in the car audio environment, but not so much with home audio, other than purchasing a pair of satellites and connecting my 12” powered subwoofer to my 1990 Phillips 4X oversampling CD changer, and 1994 125 watts X 4 amplifier. So what I am about to say has virtually no application to home audio. I’ll explain why as I proceed.
User 1 says:
Well, in that case the guy has a 2 way setup with mids in stock locations in the doors, and a tweeter in the upper portion of his door. Imho, the proper thing to do is to cross as LOW as possible to avoid the nasty upper end response coming out of the door on his mids, and to raise the stage.
The first part of this pair of sentences may be right for the majority of 7" midbasses, but the last part is not, in my opinion. First, the part that you’re probably right about: most 7” midbass don’t do so well out to 4,000 Hz. The Hybrid Audio Technologies Legatia L6 does, as does a select few other midbass, but most do not. Why? Because of cone distortion, edge modes, and cone break-up. This is usually related to cone material, and whether or not the cone allows for a smooth low pass filter, or a jagged one which results in considerable cone break-up (no midbass is pleasant to listen to in its cone break-up mode). Exacerbating the situation is that the midbass is 60+ degrees off axis, and the beaming effect at high frequencies. But there’s a trade-of. The second part of the above sentence is wrong, IMHO, because of the following:
Crossing the tweets at 4khz in that instance (which is more typical than running a 3-way setup or kicks) would probably not be the best thing as your stage will drop big time and the power response at 4khz would be god awful for a 7" driver in the door.
This I will unfortunately disagree with 100%. The brain cannot localize sounds in the vertical plane nearly as well as in the horizontal plane due to head shadowing and HRTF. So this is a non-argument. Why do most properly set-up kick panel cars, with speakers mounted at your ankles, stage at dashboard level? It’s because of head shadowing and precedence effect. Thank God our ears are at the sides of our heads, instead of top to bottom, because if they were top to bottom, stacked vertically, the vertical boundary would no longer be bound by head shadowing (but the horizontal boundary would). So this is not a good argument for crossing tweeters lower, and then placing them higher in the door or dashboard to raise stage height.
By putting tweeters high in the door, dash, or a-pillars and playing them into the midrange frequencies does do one thing though (besides the obvious phase shift at the crosspoint and phase distortion), it splits the vocal spectrum between four speakers, instead of two. With tweeters playing to 2,000 Hz, you will now split the imaging cues between four drivers…about ¾ of the tones will emanate from the midrange down in the door, and ¼ from the tweeter up high. In this instance, there will be no defined center image in the vehicle. Why? Because imaging cues are critical to 6,000 Hz or so, as defined by the size and shape of your head and torso…once again, the head shadowing phenomenon. No, what you will have is your male vocals coming from the center of the car (or wherever your center image happens to go, based upon time and amplitude considerations), and your female vocals will be to the far sides of the doors, dash, or a-pillar, wherever you have mounted the tweeters. This is because ITD helps to solidify the lower and middle midrange frequencies, up to around 2,000 where ITD is most notable. Assuming these frequencies are in phase, ITD will "fool" your brain into thinking they are dash height and are centered. But above 2,000 Hz, ITD is a non-issue, and IID and HRTF takes over, so time is not on your side with respect to tweeters mounted up high. Since IID and HRTF is dominant at your proposed crossover frequency of 2,000 Hz, you will have three or more center-images. This doesn’t include instruments as well. Guitars will be centered, bell trees, triangles, and high-hats will be on the doors. The fundamental tones of drums, particularly a snare drum, will be centered, and its ambience effect will be at the doors.
I have tuned hundreds of car with tweeters mounted up high, and midbass or midrange down low. I can usually tell within 1/3 octave of where the tweeter’s crossover point is because if the tweeter crossover is improperly set, and is too low, I will hear three or more center images.
In summary, there’s a trade-off here. My belief is that you find a 7” midbass that can play to 4,000 Hz without cone break-up, and make this entire discussion a non-issue. But if you want to use a midbass that cannot play to 4,000 Hz, then the tweeter *must* be mounted right next to the midbass. No question about it. There’s not a computer program that tells me this, Klippel is of little use…it’s is pure, unadulterated experience tuning hundreds of cars, both for competition and daily listening. Back to User 1's comments:
I typically consider phase distortion the least important factor in system design, although I do find phase linear filters to be very useful.
I agree. I didn’t realize this was going to be a major factor in this discussion. But it is at least as important as other factors, such as phase response and absolute polarity. They are all inter-related in a very big way, and they are worth discussing.
Take for example a home speaker with 4th order filters at 2khz, and compare the imaging to that of a 3-way setup with no crossover in the midrange and see which will image better... and then you kind of understand my point. So why sacrifice non-linear distortion performance and tonality in order to reduce crossover phase distortion?
I won’t argue with this…why? Because you’re no longer in a near-field environment, you are no longer at 60 or 75 degrees off axis (but rather 0 degrees on axis), beaming is no longer a consideration (because you’re on axis with the speakers), power compression with respect to tweeters playing low frequencies in a large room is MUCH different than it is in a car, and etc. I could write three pages how User 1's point of view on home audio is right, because he is right, but only because of the room, the listening environment, the phase, the axis of listening, and etc. The above has no application in a car audio environment, if nothing more from the perspective that power compression of a tweeter in the far-field sense is not a concern, and you’re listening to the speakers at a theoretical 0 degrees of phase.
I'd also disagree about tweeters not being able to handle low xover points. There are many successful examples, such as Linkwitz's Orions utilizing Seas Milleniums ... and many measurements that support tweeters being able to handle xover points below 2khz. It's all about priorities.
Once again, not to be contrary, but you are describing a far-field system with lower power compression and lower air impedance in the home environment. And I never said that tweeters are not “able” to handle the low crossover point. They can in fact “handle it”, otherwise people would be blowing stuff up. But just because the tweeters can handle it doesn’t necessarily mean it sounds good in a car environment. Most subwoofers can play a 500 Hz tone, but that doesn’t mean that it sounds good at 500 Hz.
A single driver playing 250hz-5khz is arguably going to have higher IMD, while a tweeter playing 2khz-5khz will typically have a faster time response and significantly wider power response.
In a car? I doubt it. IMD for a midrange at 4,000 Hz is a non-issue. It is an issue at 2,000 Hz with a tweeter though, especially onhe placed up high in the car with no acoustic damping, because of lack of suspension to provide restorative force, and the lack of damping at those frequencies to reduce IMD. In the Linkwitz information you presented, generation of intermodulation products & harmonics is in the list, as is phase response and absolute polarity.
And in your argument, you mention impulse response. Unfortunately, impulse response works in my favour, not yours. By having all of the vocal information being emitted from one device, the midrange, the impulse response is unity. The moment you cross the tweeters into the vocal spectrum, the impulse response of the tweeter no longer has the vocal information in unity; the tweeter will emit those tones to your ears 2-5 ms before the midrange does, even further exacerbating the situation. No to mention the phenomenon know as zero delay plane, which will undoubtedly occur with the tweeter being mounted so much closer to you. So not only is there going to be a phase shift at the low crossover point, you’re going to have phase distortion, a misalignment of the zero delay plane, tweeter power compression, tweeter intermodulation distortion, and a split sound stage with three or more centre images.
A smaller midrange in the car can allow for much better positioning, ease of use, etc. that are critical don't get me wrong, but to say that no tweeter should be crossed below 4khz or that it can't be done successfully I feel is a bit strong.
Perhaps a bit strong. I am not usually one for blanket statements. Ok, I will revise: “95% of the time, the tweeter should not be crossed over lower than 4,000 Hz in a car environment. When it is crossed over lower than 4,000 Hz, the tweeter must be located within a distance of 1/4 wavelength at the crossover point from the midrange, and even still this is not an ideal situation considering phase shift at the low crossover point, phase distortion, tweeter power compression and air impedance shift, poor modal response, poor polar response, tweeter intermodulation distortion, and making the tweeter considerably more susceptible to mechanical deformation.”
User 2 wrote:
I don't think it's about converting the non-believers. Better yet, to just do your best to let others know where you are coming from. Leave the final decision up to other person. I know I am getting off topic now.
I couldn’t agree more. I wrote one and a half pages above on the serious negative effects of crossing a tweeter over too low in a car, but at the end of the day, the decision is based upon your own two ears, listening in a car. Not on a sound board, not in a living room, not on the computer screen evaluating parameters and test data, but in the car with the sensitive testing instruments and database of sounds known as your ears and your brain.
User 3 wrote:
If I remember correctly it is better to cross around 2-2.5khz, at least that's the advice I got from the master minds when I was playing with my setup. The ear has a harder time noticing a shift around those frequencies.
The first part of your statement is your own personal philosophy, and seeking advice of those that you trust. But sadly, the second part is dead wrong. Since the vocal spectrum between 2,000 and 6,000 Hz is dominated by IID, HRTF, and precedence effect, the ear readily hears these differences, phase shift, IMD, and etc. Even if the ears couldn’t evaluate these things, you’d still hear multiple center images if the tweeters were separated from the midranges at those frequencies.
I am not always right, and I learn something each day. But I am pretty confident about this subject, because I have tested my theorems in hundreds of cars (yes, physically sitting in cars, tuning, in the near-field environment). Everything I have said above has no application to home audio, for the reasons discussed. As for the remainder of the arguments, I am all ears. But I would like to suggest something. I have a team of guys combing the country hitting shows. If you want to hear first-hand what I am talking about, we can do a bench-scale experiment with high crossover points and low crossover points, in a car. The way to learn is to hear for yourself. I would be honored to be given the chance to do a “before and after” evaluation with whomever might be interested, to show the real-world effects of a high tweeter cross-point and a low one. I have 20 cars at my disposal, all around the country. We can sit, listen, and evaluate without conjecture, philosophizing, or theorizing. You let your ears decide.
Oh, and before I forget, none of the above was meant to bolster Hybrid Audio Technologies products or attributes. There are a few brands out there that make good mids for use in wide-bandwidth. No, I wrote the above because I find this to be one of maybe four of five main issues with car audio today. Many people don't seem to really grasp zero delay plane, tweeter power compression and IMD, vocal cues as they interelate with IID and HRTF and split sound-stages, tweeter verses midbass impulse response, application of on-axis far field information to car audio, and etc. It has been the subject of literally hundreds of hours of research for me. How many of you sit for hours at a time just listening to tweeters in a car, playing with crossovers and evaluating? Yep, that’s me, I’m a car audio dork, and of all the things I have researched, I have researched this subject the most.
