The best mid range x-over points

[Cazten]

Im wondering what you all consider to be the best points to cross over your midrange? It is widely accepted that midrange is the most important aspect of your system as its where most the action happens, so its no surprise either that from a purely sq standpoint many will advocate to build "around your chosen midrange".

Heres the part where perhaps my confusion can come in. Ive read many many posts across forums that generally agree that crossovers built in the meat and potatoes of the midrange can be heard. Badly done ones can be a catastrophe. And some critics even argue that well implemented ones can still be "spotted" or leave a trace and disruption in the critical frequency band.

So the question comes to this. What are the absolute best no compromise solutions? Its seems a given that the 3-way is the starting point if you want to actually reproduce the entire spectrum with any authority. But where do you choose your lowest and highest points (theoretically, assuming you could find the drivers to mate with it). I would assume that some sort of dome or small 3inch midrange would be the best solution as they can still get low enough to cross to a dedicated woofer yet will have the top end to zip right through 4k-7k+ and completely remove a crossover from the critical range that effects the voice, most instruments, and many of the higher "detailing" sounds. Then cross to a tweeter thatl handle all your whisps, sparkles, and air.

Yet it seems by far even in three ways the most popular route is usually 4"-6" mids crossed between 2k-3k, and sometimes even much lower? Even the seas excel regarded as one of the best midranges money can buy regardless of size isnt even really capable of being crossed above 2.5-3k is seems?
Why is this so?

Whats the best route to take?

[GranteedEV]

So the question comes to this. What are the absolute best no compromise solutions?

I'm only guessing here, but maybe something like

2X Thigpen Rotary in IBfrom 3hz - 20hz
5X AE AV15X Sealed from 20hz - 120hz
1X AE TD15M Sealed from 100hz - 500hz
1X TAD TD-2002 + exit angle customized 18" SEOS waveguide w/ super foam - 500hz - ∞

:eek:

Heck, make it a coaxial ultimate waveguide woofer made of beryllium :eek:

Seriously though, rather than worrying about what other people think, how do YOU feel about the audibility of crossovers?????

Start with a wide range driver (maybe the new CSS VWR126X???) and listen to it for a while. What does it do correctly to YOUR ears? What is it missing to YOUR ears?

Next listen to a crazy multi driver, multi crossover speaker at some hi fi shops. The Revel Salon 2s are a 4-way with six drivers, plus vents. Surely they're a mess, right? What is missing to YOUR ears?

Forget the concept of no compromise. Find the compromises that best suit your ears. Yes there's literally people out there who can't stand crossovers. But i'm convinced they don't listen to anything but that same female jazz artist who i don't bother with, and all they want is to make sure her S's sound good. I was actually auditioning a $60,000 YG Acoustics speaker a couple weeks ago (no reason). The dealer kept pointing out to me how much he hates crossovers and loved his old Quad electrostat. I mean great for him, but I was falling asleep listening to his stock demo material to even notice any crossovers. All I was hearing was "ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZzz". And of course, he kept talking about how amps sound. So I really question other people's opinions on what they're subjectively hearing. Cuz all I hear with most of the audiophile jazzy / rock for 60 year olds stuff is "zzzzzzz".

Learn from what your ears tell you to music you yourself are familiar with and interested in. And then apply the science as you see fit (that means you will need low compression of dynamic peaks, flat response, smooth polars, good enclosures and <<good>> crossovers regardless of where you place them). It's my opinion that speakers that "aim to get the crossover out of the speech band", end up sounding awful even if speech has really good ess's and h's or whatever these people are listening for. The speakers just don't sound right.

[Cazten]

lol yes i suppose listening to as much as possible and experience of my own listening impressions is helpful. But as it stands without investing years and years and $$ i dont have there are limits to what I can do so im trying to educate myself as much as possible for my own diy adventures.

And to clarify on the "no compromise solutions". Im not looking for an exact build but more what you all personally find, based on your own preferences, to be the best solution. EX: 2way, 3way, 4way, where you like your xover points? Do you like a midrange extending 4k+? Do you like a lower 2k point to a fabulous tweeter? Do you like a more seemless 1 driver approach to the lower midbass midrange 100-500hz + upper midrange region or would you run a midrange down and cross to a midbass before hitting subwoofer territory. ect

[GranteedEV]

build but more what you all personally find, based on your own preferences, to be the best solution.

I think that while there's subtle differences, good speakers sound more similar than not. They have the following characteristics

1) Flat frequency response +/- 2db from 200hz to 10khz especially, but also above and below that without aberrations.
2) Controlled, smoothly tapering off axis response out to at least 60 degrees.
3) Uncompressed output suitable for the room they will be used in
4) Clean impulse response free of ringing or mud.

Now #2 is a big one to me. #3 is the big limitation on maximizing #2.

Let's say I'm using this driver as my tweeter / mid:

http://reconingspeakers.com/products-page/pro-audio/kef-8-q900-coaxial-driver/

I'd need to measure it to be sure but to my knowledge, the directivity index most closely match somewhere between 1.5khz and 2khz. I can also see the massive metal cone breakup in the measurements pop up at about 5 to 10khz.

Quite frankly, the "one driver from 300hz to 4khz" crowd probably wouldn't like this driver. I like the ability of the 8" cone to behave as a waveguide to control the tweeter's dispersion. I like the power handing of a 1.5" tweeter and an 8" mid with good sensitivity.

So I'd have to throw away the advantages of "no crossover" to meet the goals that I consider much more fundamental. It's not that I prefer a crossover right at 1.5-3 khz where our ear is probably most sensitive to crossovers. But I know what I can hear and I prefer a smooth handoff between drivers, rather than to shove a rough handoff between drivers away from the immediately audible range. JMO.

The drivers just don't exist that can have good on AND off-axis response @ 10khz and good response below 1khz, except beryllium compression drivers like the TAD TD-2002 I mentioned earlier. And a pair of those will cost you about $3400. And you'll still need to find a world class waveguide worthy of them. And you'll want to stuff it with geddes' foam because he says it sounds better to listen to your speakers through 10 inches of foam. And that doesn't even mean it's the best sounding tweeter out there... maybe a RAAL still sounds better. I wish I had the money to tell ya.

So back on the topic of crossovers in the upper midrange - I consider them one option that is the lesser of some evils. They themself are an evil. But i don't have $3400 to spend on a tweeter.

FWIW, if you're intent on a classic 3-way with "one mid to rule them all" I would recommend this ATC:

http://www.solen.ca/pub/cms_nf_catalogue_fiche.php?id=5&recherche=&numRows=&manufacturiers=1&niveau1=&niveau2=&niveau3=

[billfitzmaurice]

Im wondering what you all consider to be the best points to cross over your midrange?No higher than where the response of the woofer is 6dB down at 30 degrees off-axis and well below any break-up mode on-axis. The same applies to where to cross from the midrange to the tweeter.

[Pete Schumacher ®]

For a floor standing 3-way, with the woofer near the floor, 300-400 Hz is a good XO point, handing off to the mid where baffle step loss is not an issue for the mid.

It's probably easier to tell when one driver hands off to another with steeper slopes in the crossover, but if the transition is smooth, even higher order XOs, with well matched drivers, sound quite good. In the Byzantium build, I chose 2nd order acoustic slopes, because the drivers have good overlap of their responses, and the crossover points kept the strain away from the mid and tweeter. But in that build, the upper XO point is at 2200Hz, right where it's supposed to be a problem for voices, etc. All I can say is . . . meh . . . it's a non-issue. I could have pushed it to 3KHz or higher, but why?

There's also the properties of the mid to consider. I've been listening to an 8" Vifa NE225W that is used as a mid-woofer to 1KHz, and the mid range reproduction is simply excellent. There's a certain weight and body to the delivery that is tough for a smaller driver to match. But it's size limits usage to 1KHz, which means a robust tweeter/waveguide combination to cross that low. But if WGs are not an option, I'd opt for a mid no larger than 5.25" so that you can cross to a tweeter above 2KHz.

[mzisserson]

No higher than where the response of the woofer is 6dB down at 30 degrees off-axis and well below any break-up mode on-axis. The same applies to where to cross from the midrange to the tweeter.

I like it. That is a great rule o' thumb.

I generally use the 6dB/30 rule for the bottom, and at least one octave below any on-axis breakup modes. I also like to use small mids and cross 4-6K. Why do a 3-way if you cannot utilize the advantage of a smaller dome/ribbon/planar?

[Deward Hastings]

It has a lot to do with how you "frame" the question, and the underlying presumptions. I'm curious first what *you* mean by "midrange", and what *you* think is "special" about it. And I question *starting* from the presumption of "hearing the crossover". So much depends on how you read that . . . I'd argue that you *don't* "hear the crossover", you hear failed driver integration around the crossover frequency (usually a change in directionality), and that leads to the possibility that the issue is not "the crossover" but mis-matched (or improperly used) drivers.

If you pick the crossover frequency based on the physics (acoustics) of the drivers then (if you've picked well) it doesn't much matter where it falls. If you're of the "300-3000" school you've got to have a midrange driver small enough to not be beaming at 3000 (and with its first "breakup" above 10k) and with enough linear volume displacement to give your design SPL at 300 (or a horn big enough to maintain directional control at 300 and not honking at 3000). That's a tall order either way (you might try a 3 inch "full range" crossed to a substantial woofer). On the other hand Linkwitz has built two speakers that cross right in the middle of that range (1000 and 1400 Hz., the latter even with significant tweeter "bloom") and I simply laugh at anyone who claims that they can "hear the crossover" in either of those speakers.

[cjd]

It's probably easier to tell when one driver hands off to another with steeper slopes in the crossover, but if the transition is smooth, even higher order XOs, with well matched drivers, sound quite good.

The second half of this is key, at least for me: well matched drivers with a good transition. I've heard 8th order that are incredibly tough to tell, and I've heard 2nd order where it was pretty obvious. How well the drivers match is probably the reason: less tonally matched = easy to tell, no matter the slope.

That's with music, which almost always has *some* high frequency output even on super low notes that provides just enough anchor to the tweeter that it holds center. With a sweep, it's a bit less guesswork on a higher order slope, but not much.

I also choose starting crossover points in a way similar to BF, except that I target a frequency for the crossover point such that it's not down at all at 30deg off-axis, and at the official -12dB crossover point it's no more than 6dB out at 30deg off-axis. So I'd cross an octave higher/lower than Bill on average, assuming I'm using a 4th order slope. These are still just starting points, and I let results (total integration, sound, etc.) alter this if needed.

C

[billfitzmaurice]

I also choose starting crossover points in a way similar to BF, except that I target a frequency for the crossover point such that it's not down at all at 30deg off-axis
That's nice when you can find drivers capable of that, as you won't have off-axis response that dips at the top end of one driver's pass band and then jumps back up again at the bottom end of the next. I'd call down 6dB at 30 degrees the worst acceptable case. But for sure what most newbies overlook is the basic underlying reason for using multiple drivers, which isn't axial response but off-axis response. It does no good to have an axial plot that looks like the skyline in Kansas when the off-axis looks like that in Nepal.

[rpb]

Im wondering what you all consider to be the best points to cross over your midrange? It is widely accepted that midrange is the most important aspect of your system as its where most the action happens, so its no surprise either that from a purely sq standpoint many will advocate to build "around your chosen midrange".

Heres the part where perhaps my confusion can come in. Ive read many many posts across forums that generally agree that crossovers built in the meat and potatoes of the midrange can be heard. Badly done ones can be a catastrophe. And some critics even argue that well implemented ones can still be "spotted" or leave a trace and disruption in the critical frequency band.

So the question comes to this. What are the absolute best no compromise solutions? Its seems a given that the 3-way is the starting point if you want to actually reproduce the entire spectrum with any authority. But where do you choose your lowest and highest points (theoretically, assuming you could find the drivers to mate with it). I would assume that some sort of dome or small 3inch midrange would be the best solution as they can still get low enough to cross to a dedicated woofer yet will have the top end to zip right through 4k-7k+ and completely remove a crossover from the critical range that effects the voice, most instruments, and many of the higher "detailing" sounds. Then cross to a tweeter thatl handle all your whisps, sparkles, and air.

Yet it seems by far even in three ways the most popular route is usually 4"-6" mids crossed between 2k-3k, and sometimes even much lower? Even the seas excel regarded as one of the best midranges money can buy regardless of size isnt even really capable of being crossed above 2.5-3k is seems?
Why is this so?

Whats the best route to take?

I would use a 5.5" mid crossed at about 350 hz to the woofer, and somewhere between 1.5k and 2.8k to a tweeter. Possibly a tweeter with a waveguide. I would look closely at the distortion of all the drivers in their intended range, and then make the best compromise between off axis response, cost, and low distortion, based on how relevant you think each is.

If you can find a very clean 4" mid that is clean beyond 3k, and within your price range, then crossing higher is an option. The one or two small drivers that I've played around with do not have unusually low distortion. As best I remember, the 5.5" drivers I have are better than the smaller ones below 1k, but they are not excellent at 2k. They may be good enough at 2k, but they tend to be better at 1.5k. I don't recall the fours being any better at 2k than the 5.5" drivers, but I no longer have my measurements on them, so they might be. Also, small drivers seem to have more problems with diffraction. So, in short, there's not one best approach, but lots of potentially good ones.

[Cazten]

Lots of excellent replies here i appreciate all the thoughts. It'l go a long way in broadening my driver matching process as i find or design a project for myself.

Here's a second question someone might clarify for me. The popular point for crossing mids to mid woofer/subs seems to be in the 250-400 range? Is there anything scientific to this or is it a result of where most smaller drivers require a crossover? Part of my reason for asking is you see many designs than try to keep the tweeter/mid range as close as physically possible near the top of a design, and the mid woofer might be near the bottom or a good distance away.

My understanding is the closely placed tweeter/mid is partly to make it sound like a single sound source instead of being able to pick them apart from difference locations. Doesn't a mid woofer in the 250ish+ range still have directivity? I always thought music didn't loose its directivity until you hit like sub 80hz? Thus wouldn't you want those semi close as well?

[Pete Schumacher ®]

Lots of excellent replies here i appreciate all the thoughts. It'l go a long way in broadening my driver matching process as i find or design a project for myself.

Here's a second question someone might clarify for me. The popular point for crossing mids to mid woofer/subs seems to be in the 250-400 range? Is there anything scientific to this or is it a result of where most smaller drivers require a crossover? Part of my reason for asking is you see many designs than try to keep the tweeter/mid range as close as physically possible near the top of a design, and the mid woofer might be near the bottom or a good distance away.

My understanding is the closely placed tweeter/mid is partly to make it sound like a single sound source instead of being able to pick them apart from difference locations. Doesn't a mid woofer in the 250ish+ range still have directivity? I always thought music didn't loose its directivity until you hit like sub 80hz? Thus wouldn't you want those semi close as well?

The 350Hz crossover point fits in with a typical floor standing 3-way, where the woofer is fairly close to the floor. Floor boundary adds to the apparent woofer baffle size. The mid, with shorter frequencies, will start to work with the baffle to cause the low frequencies in the forward axis to drop off. By crossing to the woofer at the frequency where the drop off begins to occur, the mid is eased of the responsibility for the upper bass/lower mids, where the woofer is more efficient. A typical 12" woofer will be good to over 500Hz, so 350Hz is a piece of cake for keeping things "omni."

[Deward Hastings]

Doesn't a mid woofer in the 250ish+ range still have directivity?It's all a matter of wavelength. As a "rule of thumb" you want a driver diameter no greater than a half wavelength (at its upper crossover) to avoid beaming, and center-to-center of less than a half wavelength to avoid lobing (the equivalent of beaming). At 250 Hz. a half wavelength is about two feet, so there's rarely any W/M issue with driver integration. At the typical M/T crossover, where wavelengths are much shorter, people commonly place the drivers too far apart . . .

[GranteedEV]

The popular point for crossing mids to mid woofer/subs seems to be in the 250-400 range? Is there anything scientific to this or is it a result of where most smaller drivers require a crossover?

other factors already addressed aside, i'd say two main reasons for this would be

1) Air core inductors don't like lower crossover points (high cost and DCR, and size of course), and cheap people don't like active crossovers that need multiple amps to get the crossover lower than roughly 300hz. Even at these frequencies most people will use steel laminate inductors.
2) In a lot of music (and movies BTW), this is roughly where the power demands are centered, so having a larger driver handle this high power frequency region maximizes dynamic range, which in turn gives a more realistic presentation. Not only do larger drivers generally have more volume displacement, but they are normally more free from thermal compression due to higher sensitivity and increased heat sinking.

One additional reason, is that having some driver overlap under the Shroeder frequency, can improve response a bit.


I always thought music didn't loose its directivity until you hit like sub 80hz? Thus wouldn't you want those semi close as well?

You're confusing "our ability to localize bass" with "directionality". In a box, most speakers lose directionality by 400-800hz. Some speakers like the Pluto, lose directionality as high as 3khz. Other speakers with gradient patterns, like the Nao Note, are "directional" as low as 50hz probably.

FWIW, in a good low distortion setup, I am unable to localize a subwoofer with its low pass filter centered at 120hz.

[Cazten]

You're confusing "our ability to localize bass" with "directionality". In a box, most speakers lose directionality by 400-800hz. Some speakers like the Pluto, lose directionality as high as 3khz. Other speakers with gradient patterns, like the Nao Note, are "directional" as low as 50hz probably.

FWIW, in a good low distortion setup, I am unable to localize a subwoofer with its low pass filter centered at 120hz.

Gotcha, so by directionality we are talking more about on/off axis response?

What do we do to make sure All driver integration sounds as much possible like a single source vs being able to "localize" the bass, or any other driver for that matter. Is that whats taken care of with the half wavelength spacing rules?

BTW- You guys are awesome. Seriously the folks on this forum really seem to know their stuff. Great community.

[Cazten]

It's all a matter of wavelength. As a "rule of thumb" you want a driver diameter no greater than a half wavelength (at its upper crossover) to avoid beaming, and center-to-center of less than a half wavelength to avoid lobing (the equivalent of beaming). At 250 Hz. a half wavelength is about two feet, so there's rarely any W/M issue with driver integration. At the typical M/T crossover, where wavelengths are much shorter, people commonly place the drivers too far apart . . .

In this case it seems nearly impossible to have a mid range cross over very high at all? Unless my math is way off?

Wave length (in feet) is 1130/frequency correct?

As such half wave length for ((1130/250hz)/2)=2.26ft so roughly in line with your 2 foot estimate.

At 3000hz ((1130/3000hz)/2)=0.188ft or 2.25inches
Center to center for that is nearly impossible unless using tweeters with small flanges with less than 3in drivers? And that's not even a very high x-over point.

As far as speaker diameter vs beaming i don't understand this? Unless my understanding is fundamentally incorrect i thought speaker beaming was a matter of on/off axis response. And many drivers can keep decent off axis response higher than indicated by half wavelength/diameter.

Edit: I found some material after posting this explaining the math behind beaming/cone size as it leads to on/off axis response. So since it still does hold true that there are drivers that keep a decent off axis response profile at higher frequencies would the equation for beaming just be where that point mathematically starts, even if moving past that point can still be deemed usable on a project by project basis.

[Chris Roemer]

Gotcha, so by directionality we are talking more about on/off axis response?

What do we do to make sure All driver integration sounds as much possible like a single source vs being able to "localize" the bass, or any other driver for that matter. Is that whats taken care of with the half wavelength spacing rules?

BTW- You guys are awesome. Seriously the folks on this forum really seem to know their stuff. Great community.

1120 fps = 13500 ips.

13.5k / 3k = 4.5" C-C.

That could be a 5"mid and maybe a 1" dome w/4"d faceplate.

[gowa]

http://www.lenardaudio.com/education/06_x-over.html

[GranteedEV]

What do we do to make sure All driver integration sounds as much possible like a single source vs being able to "localize" the bass, or any other driver for that matter. Is that whats taken care of with the half wavelength spacing rules?

well, a few things to start a To-Do list might be:

-drivers with the least sonic signature of their own
-matching directivity index at crossover point
-Tight driver spacing (dual concentrics may be optimal but are very rare)
-very good inter-driver phase tracking over as wide a range as possible.
-flat frequency response (that doesn't just mean the global response, but the individual drivers can't have mystery peaks "hidden" by the crossover.
- comparable SPL tracking. A tweeter that can't handle power will give itself away!
- lower order crossovers - this overlap can smooth transitions between drivers. Generally attempting 1st order is a mess, so I'm refering to 2nd/3rd order here. But i'm not convinced this is necessary either. In fact I'd rather do 110db/oct FIR slopes if it sounds good.

[Cazten]

1120 fps = 13500 ips.

13.5k / 3k = 4.5" C-C.

That could be a 5"mid and maybe a 1" dome w/4"d faceplate.

So 1-wave length c-t-c is acceptable then? No need for 1/2 wavelength? Cause that makes things MUCH easier and realistic

[GranteedEV]

So 1-wave length c-t-c is acceptable then? No need for 1/2 wavelength? Cause that makes things MUCH easier and realistic

It makes it more realistic/practical, but that doesn't mean it's not a compromise. If the drivers have destructive interference on the vertical axis, then it's already a compromise.

Whether it's "acceptable" is up the ear of the beholder. Just being a compromise doesn't make it an audible one, but me saying it's not necessarily audible doesn't make it not audible.

To prevent destructive interference, you would need driver centers 1/4 wavelength apart.

http://www.livesoundint.com/archives/2003/june/horiz/horiz_17.jpg

Realistically, for cone speakers, only coaxials can really pass these requirements.

[Deward Hastings]

No need for 1/2 wavelength?Consider two point sources 1/2 wavelength apart (or two points on the opposite sides of a speaker cone). On what we would call the "listening axis" radiation from those points sums just fine, but at 90 degrees (on the baffle plane) they are 180 degrees out of phase, and cancel. If you move the two points further apart (or raise the frequency of the signal) the cancellation null moves forward, producing a "lobe", or a "beam", until at one wavelength they sum again, producing a lobe on the plane of the baffle. How much of that behavior you regard as "acceptable compromise" depends a lot on nearby reflecting surfaces, and the overall effect on the speaker's power response.

Your calculations for M/T driver spacing are essentially correct (I did say it's rarely done right, didn't I?). The mid in a three-way should be no bigger than 4 inches, and 3 inches is better (this has the added advantage of pushing the mid driver's cone breakup high enough to be easily dealt with), and yep, small flange tweeter, too. The large spacing found in two-ways, and the large tweeters and mid-woofers, are a necessary compromise *for two-ways* that needn't, and shouldn't, be replicated in three-way speakers. There are, of course, other reasons for compromising on this . . . but as with all compromises one should weigh carefully the costs and benefits. There should be a good reason *why* you want a larger mid or greater M/T spacing or . . . well . . . you probably shouldn't.

Interestingly, when the two points get 3-4 wavelengths (or more) apart their radiation becomes effectively de-correlated . . . the lobes become so many, so small, and so rapidly shifting with frequency that they become unnoticable, and the radiation pattern devolves to that of each individual driver. That's why "super-tweeters" sitting on the top of the box are able to work . . .