Page 1 of 6 12345 ... LastLast
Results 1 to 10 of 56

Thread: Unadultered Crossover Points

  1. #1
    Senior Member cvjoint's Avatar
    Join Date
    Mar 2011
    Location
    Northern California
    Posts
    801
    Real Name
    George

    Unadultered Crossover Points

    Let's abstract for a moment from rattles, cone breakups and output limitations. What makes a crossover desirable from the point of view of staging and related hearing mechanisms?




    In an older post on a different forum werewolf mentioned areas of confusion where the brain switches from one mechanism to another such as time differences to level differences. I can't quite find that thread anymore.



    On the otherhand manufacturers tend to recommend rather esoteric crossovers that benefit the design of their product, i.e. if the tweeter can cross low compared to the competition then the manufacturer is likely to argue for lower corssover points as an ideal choice.



    With my 3 way plus sub I seem to have lots of freedom of choice for xovers without sacrificing output. What should I pick?
    I have the BG Neo8 and Vifa Ne19 in the pillars and a 7" B&C in the doors. On the BG forums the guy recommends a crossover point of either 200hz or lower, 800hz or higher. 200-800hz is the "meat" of the midrange and a crossover in this region is a dominated strategy.




    Wiki says between 800hz and 1600hz the brain switches from time to level differences. Is this an area of confusion? Should I cross here?

    On HT forums I see 1500hz-3500hz as an ideal choice. Is this because most tweeters have difficulty going lower, merely an output restriction?

    Car audio guys seem to want to cross their pillar drivers as low as possible but is there any "truth" behind the benefits?

    Wiki also says that we use ITDs down to 80hz. Does that make 63hz an ideal crossover point for the sub?
    Last edited by cvjoint; 04-07-2011 at 04:39 AM.

  2. #2
    Founding Member Subwoofery's Avatar
    Join Date
    Sep 2010
    Posts
    353
    Real Name
    Kelvin
    Always interested in those discussions... Subs'

    Kelvin

  3. #3
    Founding Member
    Join Date
    Aug 2010
    Posts
    496
    I stopped paying attention to numbers. Once the drivers are pretty well in phase, even EQed a bit (or not), it's pretty clear how the crossover cutoff affects the sound.

    If you can get some pink noise, both mono ans stereo, and listen to it for awhile with a great reference, it's pretty clear there is a timbre difference when you listen to the same in the car.

    If you have a processor that is capable of sweeping the crossover frequency (ideally both individually and for both drivers) of pairs of drivers, it's pretty clear when the sound is getting closer to the reference or farther away.

    I wish I could get a control where I could have two knobs, one in each hand, where one hand controls both midbass lowpass filters and the other hand controls the midrange highpass filters. THAT would be a heck of alot of fun to play with with pink noise.

    But I'm sure there are many processors that let you fiddle with one, then the other, then link together and sweep up and down, maintaining the underlap (or overlap, or none, as the case may be)

    ------------

    Where it lands and why? you guys can discuss. But this technique is pretty good at further narrowing the selected crossover points for my own car. And is also why I am a big big fan of drivers that have as much overlap as I can get. Gives me more wiggle room to find that optimized value.

  4. #4
    Founding Member
    Join Date
    Aug 2010
    Posts
    180
    Quote Originally Posted by Whiterabbit View Post
    I wish I could get a control where I could have two knobs, one in each hand, where one hand controls both midbass lowpass filters and the other hand controls the midrange highpass filters. THAT would be a heck of alot of fun to play with with pink noise.
    You can...it's called pc processor using VST's + midi controller ;) Someday...
    Last edited by durwood; 04-08-2011 at 12:00 AM.

  5. #5
    Senior Member cvjoint's Avatar
    Join Date
    Mar 2011
    Location
    Northern California
    Posts
    801
    Real Name
    George
    Good experiment idea. My p99 falls way short in the processing area. I'm not sure it can even save two sets of xover points, I think it just saves two EQ. curves. That brings me to another point. To make this thing work you need to vary not only the bandpass filters back and forth but also EQ. the frequency flat at the same time. Surely enough you will still have cone breakups, and different distortion profiles that cannot be equalized. All in all it's still heavily dependent on other variables not changing.

    Let me start off with a quick proposition:
    Kick panel midbasses crossed 63hz-1600hz, subwoofer on the low end anywhere in the car, tweeters in the pillars ear level (supertweeter 6000hz up optional).

    Why I think this would work. The 1600hz crossover is the meat of the problem so I'll start here. Our binaural hearing uses ITD up to 1500hz, and IID from 800hz up. The trick here is that if crossed between 800hz and 1500hz both mechanisms function which may result in better localization, in other words two stages. Above 1500hz but bellow 3000hz we use only IID, and it becomes better the higher the frequency. I would think a crossover at 1600hz would not allow ITDs to be utilized while IID will function weakly. If there is an area of confusion 1500hz to 3000hz would be it imo. This gives credit to HT guys who have been doing this consistently over the years. It goes against what the Neo8 is recommended for, crossing 800-1500hz on the low end, which I think is most problematic. A side benefit is that the stage is high and planted.

    The kick range of 63hz-1600hz will give the longest pathlength and the shortest pathlength differences. The 63hz crossover will allow a subwoofer to take over in order to relieve the midbass of excruciating demands. Bellow 80hz sound cannot be localized with any of the hearing mechanisms nor is it stereo. A mono sub anywhere would work.

    1600hz up might be a problem, especially in a car where the output requirements are high from the tweeter. The duty can be split in two. An oversize dome 2 or 3 inches would do well 1600hz to 6000hz giving high sensitivity and low distortion, and so would the BG larger planars Neo 8 and Neo 10. From 6000hz up where the pinnae and head cues become very important a 3/4 dome can take over.
    Last edited by cvjoint; 04-08-2011 at 01:46 AM.

  6. #6
    Member Highly's Avatar
    Join Date
    Aug 2010
    Location
    Piedmont, OK
    Posts
    71
    Real Name
    Todd
    First of all, I don't think there is an 'always right' way of doing this. I am simply going to mention things the way I see them. I may very well be wrong, but I find it works for me. I very much welcome alternative views on this as well, so don't think I'm getting preachy just because I use a lot of words...


    In the nearfield I would submit that crossing over in the region where we are able to use both time and level differences to localize sound is a poor choice. In the farfield the sound of the various drivers in a speaker has the chance to fully develop into a complete or coherent wavefront. Not so much in a car. Crossing over in the critical midrange where we can hear phase differences of as little as 10 to 20 uS means that your crossover point, where there is a phase shift/inversion, will be clearly distinguishable. I think you have to be very careful about applying farfield concepts in a nearfield environment for this kind of reason.

    0 Hz to 80 Hz or so -> no directional cues
    80 to 200 Hz, phase difference is barely discernable
    200 to around 800Hz -> ITD rules, Intensity barely discernable, head is smaller than a half wavelength.
    800 Hz to 1500Hz or so -> ITD is our first discriminator with Intensity is a close second. We can be swayed either way very easily in this region.
    2000Hz+ -> head shadowing and IID rules, one wavelength fits between two ears.

    I would posit that, based on this information, you would be best served crossing as close to the 200Hz region as possible at the low end of the midrange and above 2000Hz as the top of the midrange. I would personally tend to aim higher than 2000Hz where possible (4-6K if I can) as it allows you to keep the tweeter out of areas where IMD is normally getting quite high. Over 4Khz allows for more placement options of the tweeter for kickpanel midrange locations with the tweets mounted up high. This keeps the vast majority of the vocal spectrum in one driver, reduces the crossovers in the most sensitive areas for localization to zero, and allows you to most effectively use both time and intensity controls to best align the image on your soundstage. Below 200 doesn't matter (or begins not to matter very quickly) and above 6000 is clearly into intensity-only localization. Cars set up this way often have a single stable, coherent soundstage that tends not to wander or drop with male/female vocalists.

    Again, just my take on the problem and the solution. There are others that work, possibly some better.
    Last edited by Highly; 04-08-2011 at 09:08 AM.
    Team MSE/JL Audio
    2010 MECA OK State Champion Modified SQ and Install
    2011 MECA OK State Champion Extreme SQ, SQ BOS
    2011 MECA World Champion Extreme SQ
    2012 MECA OK State Champion Extreme SQ, SQ BOS

  7. #7
    Tester Extraordinaire ErinH's Avatar
    Join Date
    Aug 2010
    Posts
    1,247
    pink noise is all fine and dandy until you play a track with a bass guitar and bzzzzzzzzzttttttttt.... because pink noise doesn't light up resonant modes like real music does.

    choose your weapons wisely.
    Your ears: The best tools you have... and they're free, too!

  8. #8
    Founding Member
    Join Date
    Aug 2010
    Posts
    496
    Agreed. I don't use pink noise to diagnose or fix rattles in my car. Just tune.

    Same reason I don't use music to tune in my car. Right tool for the right job.

  9. #9
    Senior Member cvjoint's Avatar
    Join Date
    Mar 2011
    Location
    Northern California
    Posts
    801
    Real Name
    George
    @Highly: I was reading this the other day: Buwalda Hybrids International Bulletin Board • View topic - My views on tweeters and crossovers, same theory in more detail.

    It is an interesting approach. Some of the points are well made, like choosing one driver for the entire vocal range. This is similar to what full range fanatics would say. Some of the supporting evidence is anecdotal however. Here is my take.

    Say we have driver X that can cover the entire vocal range. To be fair, let's cover the entire range from 85hz for some males to 6000hz. This will cover 99% of the sources out there for the voice range. Where would you mount it? In the kicks while the path lengths are maxed out and differences minimized the stage will be split in two height wise. Like you said, HRTF operates 2khz and up, your X driver will give a low stage height while your tweeter will give a high stage height. Now let's place driver X in pillar pods. The height problem is fixed as the tweeter will lie right next to it but you no longer have long paths to the drivers and small path length differences. The tradeoffs with driver X are height vs depth.

    Realistic drivers like the theoretical driver x have major shortcomings. For one there is not enough space for a large high sensitivity driver that can do 85hz to 6000hz. People may use 3 inch cones but they are all rather awful in a high background noise application like a car. Say you do have the space and you fit a high quality 6.5 either in the kicks or the pillars. The upper range is compromised by beaming and breakup. Manufacturers will tell you they use well damped paper, bamboo or what have you but unlike metal cones their breakup will be spread over a wider range. You will still gear it as a rough CSD, and high HD. On the low range their Xmax is far too low to wang with authority. Even great motors like XBL and the AE 6.5 would have a hard time pushing a suspension in the 10mm range imo. You may bring up waveguides but those are as big as 6.5 drivers or more. Even if you use a midbass in the doors to take over the beef of the load, you may still have ITD to notice the split and ultimately you will still have two sources for height cues (kicks and pillars). The 1600hz to 6000hz range in the kicks is also a prime target for leg diffraction, practically you will block the entire sound wave from time to time.



    Both because of my research and to make good a good debate I will argue for a 1600hz crossover point instead. The drawbacks as you mention them are phase distortion around the crossover point and IMD. I'm not a big buff on phase distortion. I would think that using a Linkwitz Riley 4th order filter and a smooth meshing of FR the phase will remain correct. A while back I was very into phase distortion and npdang said phase is a mess in the car, don't bother. Unable to find a single phase test in a car I gave up. Onto IMD, this is a non issue imo with a tweeter crossed in the 2000hz range. For one, there are many designed to do this and xmax is very low at those frequencies. Any driver well built will play in its designed range with little IMD. I give Augerpro's measurements as evidence. He finds that even for large drivers HD is the primary source of distortion by far. We are talking about 10 times more distortion from HD at xmax. IF IMD was an issue, and that's a big IF, driver X would suffer from more IMD. Think about delivering 85hz, or even 200hz at 100db while crossed over as high as 6000hz.

    Other supporting arguments, mostly a recap: ITD stop at about 1500hz. From 1500hz we can only rely on IID (same as ILD). The point I want to drive home is that 1500hz to 2000hz is an area of confusion, in some studies up to 3000hz. There seem to be hundreds of papers disagreeing on whether HRTF is good enough to do anything in this range. Furthermore, we are talking about lateralization. We can mount tweeters in the pillars and mids in kicks at the same lateral position. Height wise only the tweeter will give any cues, it should be solid.

    Here's a quick synopsis of the field, the simplest I could find:
    http://ocw.mit.edu/courses/health-sc...3_binaural.pdf

    The downsides to choosing the 1600hz xover from what I can tell are the following:
    *moving the head makes for better localization, not a big deal when you are driving but still.
    *even after time alignment the ratio of early to late reflections coming from the kick and pillar are different.
    *then there is this thing from the link I gave: "Ongoing ITDs in the stimulus envelope can be detected at high frequencies." Not exactly sure what that means in plain English, but I think for complex sounds we may be able to localize laterally drivers in the 4000hz area. Given that tweeters and midbasses are at equal lateral positions all should be ok still.

    Driver X would do better in these two scenarios.
    Last edited by cvjoint; 04-08-2011 at 05:00 PM.

  10. #10
    Tester Extraordinaire ErinH's Avatar
    Join Date
    Aug 2010
    Posts
    1,247
    I agree with Todd.
    IMO, crossing higher keeps you from having intermingling itd/ild issues and, in general it seems, most tweeters suffer distortion issues below 4khz (just start perusing Zaph's tweeter distortion measurements).
    Your ears: The best tools you have... and they're free, too!

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •