Crossovers on individual drivers

[bobdehunt]

I was up late last night and doing so I came to an epiphany. Full range drivers aren't quite full range, but other loudspeaker systems have to have crossovers which lose information. Now, what would happen if I put a full range driver into a system and only used crossovers on the woofer and tweeter? This would lower the sensitivity by about six decibels I think, but this way, I would completely avoid losing information. Also, because a full range driver has the frequency response that it has, the crossover on the other drivers could be a very gradual slope.

 

[ski2xblack]

Such a driver won't perform well at it's extremes, particularly up high with cone break-up, so what you propose won't be without issue. You're definitely on the right track with appreciation for wide bandwidth in a driver, as it can be used to push those problematic passive crossovers to frequencies outside our most discriminating frequency band. See Dennis' BMR monitor for example (although the BMR mid is not a "full range" driver, it is a very wide band driver, so hopefully it's a good enough analogy to your own thinking).

The big lesson I took from messing with so-called "full range" drivers was "go fully active" (it makes crossovers and speaker tonal balancing far easier from a speaker design and room tuning perspective, without any insertion losses, better damping-if that really matters, and less power demands).

TLS Guy has a design up his sleeve that resembles your idea somewhat. I don't have the details so this is a bit of guesswork, but it involves a pair of wide band Mark Audio drivers crossed to a helper tweet for the top end. He mentioned the networks being minimal in design to accomplish proper bsc and crossover duties, so I suspect they he used a high crossover freq and shallow slope. Probably a TL bass alignment. If my guesses are correct, it is pretty close to what you were imagining, sans the woof.

 

[MrBoat]

Wouldn't a well made coaxial driver accomplish essentially the same thing, with perhaps better time alignment?

 

[TLS Guy]

Wouldn't a well made coaxial driver accomplish essentially the same thing, with perhaps better time alignment?

You are forgetting that phase and time shifts are exactly the same thing. They are literally different words for the same physical entity.

So you can place the tweeter at the acoustic center of the woofer, but that is NOT time alignment.
For one thing the tweeter ends up being physically small to fit in the woofer pole piece. So power considerations tend to requires a faster slope.

So take my center which uses a coaxial driver. The woofer crossover is second order at 2.9 Khz, so that delays the sound by half a wave length at 2.9 KHz. The tweeter crossover third order, so that advances the sound by 3/4 of a wavelength at 2.9 Khz. So that is about a 4.5 msec time shift.

 

[TLS Guy]

I was up late last night and doing so I came to an epiphany. Full range drivers aren't quite full range, but other loudspeaker systems have to have crossovers which lose information. Now, what would happen if I put a full range driver into a system and only used crossovers on the woofer and tweeter? This would lower the sensitivity by about six decibels I think, but this way, I would completely avoid losing information. Also, because a full range driver has the frequency response that it has, the crossover on the other drivers could be a very gradual slope.

You are on the right lines, but there are few good full range drivers. The old Jordan Watts with which I have a lot of experience is absolutely one of the best. Its successors are the Mark Auido drivers. However there is eventually cone break up. The other good full range drivers are the Lowthers.

Lets deal with the top end. At break up output tends to increase so, adding a tweeter would tend to compound the problem.

At the bottom end since these drivers are small bass output will fall, although the JW and Mark drivers have good output into the 40 Hz range, and so do the Lowthers horn loaded. The problem is power handling. So a crossover is generally required to prevent driver destruction.

I do have a design with no real need to build it other than curiosity, that uses two Mark Audio drivers in a TL, with simple first order crossover, where the break up mode is dealt with and BSC applied. It models very well and the time shifts small enough to be considered time/phase coherent.

The other thing is that recordings are virtually NEVER time coherent. They are just awash in time shifts. The only way to avoid this is to use a truly coincident single mic pair. This means either the Blumlein technique or M-S.

In times passed I have made a lot of Blumlein and M-S recordings. I can assure you time does matter. It takes carefully made recordings using the above techniques and time coherent speakers to reveal it.

I do try and incorporate these principles in my designs as far as possible. My large reference speakers use two mids robust enough that they do not need a high pass filter on the bottom end in their TL. An electronic crossover adds BSC first order and extends response to 20 Hz in a second line. So this crossover at 44 Hz is entirely acoustic. The crossover transitions to second above transition so as not to excite break up modes of the woofers.

Achieving time coherence from microphone to speakers is a difficult task and seldom attempted. My view is that it should be more often attempted.

 

[bobdehunt]

Wouldn't a well made coaxial driver accomplish essentially the same thing, with perhaps better time alignment?

Possibly, but the coaxial drivers intruduce distortion. I've always heard that full range drivers without coaxial are usually better if they're well made.

 

[ski2xblack]

Which specific full rangers are you considering?

My slacker standards are not as high as TLS', and find some of the other full range drivers to be ok. My most recent and deepest dive involved a TB W8-1772 in a mltl cab (the plans via Bob Brines site, and based on mlk's tl modeling and Bob's recommended bsc). They were ok, pretty good actually considering the inherent limitations. I liked the fairly deep extension, and they did the classic full ranger cohesiveness thing quite well. Kind of like you, I felt they could have used some tweets, and subs, and bandwidth limitation as well, and just weren't quite as fun and satisfying as other kit speakers I have. (Fusion 12 Tempests peg the 'Fun' meter IMO and all that.) But the TB drivers aren't bad at all. They're about mid tier, at about $250 a pop, so not exactly cheap entry level fostex, which I've also messed with briefly.

 

[bobdehunt]

You are on the right lines, but there are few good full range drivers. The old Jordan Watts with which I have a lot of experience is absolutely one of the best. Its successors are the Mark Auido drivers. However there is eventually cone break up. The other good full range drivers are the Loathers...

I do have a design with no real need to build it other than curiosity, that uses two Mark Audio drivers in a TL, with simple first order crossover, where the break up mode is dealt with and BSC applied. It models very well and the time shifts small enough to be considered time/phase coherent...

In times passed I have made a lot of Blumlein and M-S recordings...

I do try and incorporate these principles in my designs as far as possible. My large reference speakers use two mids robust enough that they do not need a high pass filter on the bottom end in their TL. An electronic crossover adds BSC first order and extends response to 20 Hz in a second line. So this crossover at 44 Hz is entirely acoustic. The crossover transitions to second above transition so as not to excite break up modes of the woofers.

I know that good full range drivers do exist, but I believe that the one's that would be best without adding an additional tweeter and sub are usually VERY expensive. I could be wrong though. I didn't know what BSC was, so I looked it up. Would a relatively thin speaker with a curved front eliminate the need for BSC or do I not fully understand what happens to the highs and lows? I know that the frequencies that would be able to reflect off of the front of the baffle would have good horizontal dispersion. I think that adding two tweeters at a slight angle to each other would allow the highs to do the same and create a relatively flat frequency response. It's just a guess though. I do not understand crossovers very well. I've only been into audio for a couple of years and I never found crossovers as interesting as speakers and baffles. I don't understand how there is an acoustic crossover (from the baffle?) at 44 Hz if the frequency response extends to 20 Hz? I'm lost as far as the second above transition is concerned. Thank you for leaving such a comprehensive reply. I can see that you know a lot about this stuff.

 

[bobdehunt]

Which specific full rangers are you considering?

My slacker standards are not as high as TLS', and find some of the other full range drivers to be ok. My most recent and deepest dive involved a TB W8-1772 in a mltl cab (the plans via Bob Brines site, and based on mlk's tl modeling and Bob's recommended bsc). They were ok, pretty good actually considering the inherent limitations. I liked the fairly deep extension, and they did the classic full ranger cohesiveness thing quite well. Kind of like you, I felt they could have used some tweets, and subs, and bandwidth limitation as well, and just weren't quite as fun and satisfying as other kit speakers I have. (Fusion 12 Tempests peg the 'Fun' meter IMO and all that.) But the TB drivers aren't bad at all. They're about mid tier, at about $250 a pop, so not exactly cheap entry level fostex, which I've also messed with briefly.

If I get the money to build a speaker system like this, I would really like to use Audio Nirvana's classic 15's. I've never heard them, but I hear that they sound amazing and last well.

 

[MrBoat]

Possibly, but the coaxial drivers intruduce distortion. I've always heard that full range drivers without coaxial are usually better if they're well made.

Since subwoofers are so well done these days, that helps solve some problems with FR for me.

One of the coaxials (again, I would use a subwoofer) that intrigues me is this one.

I do have a pair of the markaudio drivers tlsguy spoke of, and I also have two pair of the Peerless by Tymphany TC9FD18-08 3-1/2" that are currently going into a small project I am building.

 

[bobdehunt]

Since subwoofers are so well done these days, that helps solve some problems with FR for me.

One of the coaxials (again, I would use a subwoofer) that intrigues me is this one.

I do have a pair of the markaudio drivers tlsguy spoke of, and I also have two pair of the Peerless by Tymphany TC9FD18-08 3-1/2" that are currently going into a small project I am building.

I find that a lot of people prefer not to have an external subwoofer. I have a pretty good entry sub (Dayton Audios 15) and although, it is loud and way more than I could ever expect for the price, it has a barely noticeable delay that really bothers me. I don't know if all subwoofers have a delay as obvious. Either way, the crossover would be difficult to deal with and so will the volume. I kind of like the ability to be able to change the crossover and volume though, due to the imperfections in the room and my fondness of slightly more bass than what is needed. I'd still rather have a set of good full range speakers though. I listened to an old pair of cirwin vegas concert speakers at a friends house on a pretty pitiful amplifier, and I felt like the music soared in front of me compared to the speakers that I have now with the subwoofer.

 

[ski2xblack]

The sub shouldn't be offset or delayed, unless you're using a sub amp that involves an extra ad/da conversion maybe?? Distance setting for speakers on your AVR (if you were using an AVR as a front end) provides enough adjustment to compensate for that sort of millisec delay.

Back on topic, I think that those Audio Nirvana 15" drivers would make for, well, a highly idiosyncratic speaker. Probably as sensitive as it would be beamy at even mid frequencies. The AN site is a trip, too. Do they provide any measurements or full t/s specs for those drivers? Here's the description of that driver on their site:

The Audio Nirvana 'Super 15 ALNICO' ($1099/pr[]) has a test bench frequency response from 28 hz to almost 15,000 hz (varies by temperature). You can expect at least 95 db efficiency in any of our cabinets. It handles 50 watts continuous RMS (normal listening level will be about 1/10 watt). Impedance is 8 ohms. Voice coil is 2.0 inches (50 mm). Magnet weight is 2.3 lbs. Total weight is 19 lbs. The cone is made of paper. The surround is accordion style and made of treated cloth. The phase plug is copper anodized, machined solid aluminum. The frame is cast aluminum. A separate tweeter is not necessary. Few people can hear above 14,000 cycles, so the upper limits of this speaker are all most people will ever need. You might be interested to note, for example, that FM radio is limited to 15,000 hz. Our alnico magnet models are the smoothest Audio Nirvana models and make the most bass. In one of our larger cabinets, this is the best sounding speaker we have for sale and maybe the best sounding speaker of all time.

If you're going to spend that kind of coin, why not pony up for the Lowther? Of course the Lowther wouldn't be the "best sounding of all time" if that mantle has been claimed by the AN. It's a real pickle for the discriminating auidophile.

This little area of the audiophile rabbit warren can get pricey rather quickly. Maybe as a starter course get a copy of Vance Dickeson's book on speaker design to chew on, at least before going off half cocked and buying an $1100 15" full ranger.

 

[bobdehunt]

The sub shouldn't be offset or delayed, unless you're using a sub amp that involves an extra ad/da conversion maybe?? Distance setting for speakers on your AVR (if you were using an AVR as a front end) provides enough adjustment to compensate for that sort of millisec delay.

Back on topic, I think that those Audio Nirvana 15" drivers would make for, well, a highly idiosyncratic speaker. Probably as sensitive as it would be beamy at even mid frequencies. The AN site is a trip, too. Do they provide any measurements or full t/s specs for those drivers?

I do have the outputs of the stereo going to the subwoofer, so it would have to convert the analog signal to digital, then back to analog I think. I know that all of the measurements for the drivers can found on the website. Personally, I haven't looked at them till now, but I know they are awful. Almost all of the reviews that I've found on their speakers have been very positive though. I think it's the same type of thing as a triode amp. It introduces much more distortion, but the sound is said to be more musical. Unfortunately, I have yet to have the chance to listen to either a triode amp or Audio Nirvana's 15's. I'll agree that the website is VERY sketchy though. I think that the speakers either sound like the voice of angels or two tin cans rattling in the wind. I'm eager to find out which it is.

http://www.commonsenseaudio.com/an15classicalnicospecs.jpg

That's the link to the classic 15's specs.

 

[ski2xblack]

Hmm. Just a quick glance at those specs tells me that if you were to pair it with the undamped output of a SET you would have a real boomer, as in extremely peaky, wooly, ringy bass. It would probably be a boomer on a more well sorted amp too. Of course that's just a semi-educated wild eyed guess. I actually have a soft spot for single ended tube amps, but such amps, and the whole approach that is preached at the AN website is quite a way off the reservation.

 

[bobdehunt]

Hmm. Just a quick glance at those specs tells me that if you were to pair it with the undamped output of a SET you would have a real boomer, as in extremely peaky, wooly, ringy bass. It would probably be a boomer on a more well sorted amp too. Of course that's just a semi-educated wild eyed guess. I actually have a soft spot for single ended tube amps, but such amps, and the whole approach that is preached at the AN website is quite a way off the reservation.

It is strange how bad the speaker seems to be, yet how good the reviews are.

http://audiokarma.org/forums/index.php?threads/my-audio-nirvana-super-15-4-month-review.253978/

http://forums.audioreview.com/speakers/audio-nirvana-full-range-speakers-38135.html

http://techtalk.parts-express.com/forum/tech-talk-forum/48577-my-audio-nirvana-15-neo-s

I don't hold much trust in the last one, considering the writer said that he used a certain type of speaker wire and it made an audible difference. I don't have top of the line speakers or a top of the line amp, but I don't think I could ever be able to hear the difference between two types of speaker wire and I like to think that I have a pretty sensitive ear. Unless the previous wire he was using was extremely thin, I don't think that he could hear a difference either. But maybe I'm wrong.

 

[TLS Guy]

If I get the money to build a speaker system like this, I would really like to use Audio Nirvana's classic 15's. I've never heard them, but I hear that they sound amazing and last well.

I'm aware of that "Common Sense" site. It should be labelled the "Lost all Sense" Site.

Whoever claims those drivers can sound amazing must have a serious hearing defect.

Lets start with the classic 15. The cone is starting to break up at 500 Hz, and well one at 1K. The frequency response is +/- 15db!

The Qts is almost 1 at over 0.9! That is the sort of driver you might have found in an old pre WW II or 1950s radiogram. That high Q will give it a horrid sloppy boomy bass, which would do a crude masking of the shrill HF response.

All those driver are high Q and have break up modes quite low down. Even the 6.5 inch unit has totally lost it by 3k and has a response +/- 12 db!

None of those units can be considered even remotely as high fidelity transducers.

If you are going to make a full range unit it needs to be small. The Lowther units are very sensitive, and low Q requiring horn loading to bring up the bass. If you want to experiment with SETS, and I don't advise it, then Lowther is really your only decent option and you have to pay the piper.

BSC by the way is quite simple. It is the point at which a monopole speaker transitions from a half space radiator to a full space omni directional radiator. That is determined by the width of the front baffle no matter what shape it is. To compensate a 6 db boost is required below the transition point.

I think you need to do more study or you will spend a lot of money and have something worse than a cheap table radio.

 

[bobdehunt]

BSC by the way is quite simple. It is the point at which a monopole speaker transitions from a half space radiator to a full space omni directional radiator. That is determined by the width of the front baffle no matter what shape it is. To compensate a 6 db boost is required below the transition point.

I think you need to do more study or you will spend a lot of money and have something worse than a cheap table radio.

I'm still curious about the sound of the drivers. I know they should sound bad, but I want to listen to it. I know that if you go to the actual shop, you can listen to the speakers and bring a set of your own to compare. I'd like to take a trip down there sometime in the future, but according to google maps, the place doesn't exist.

I was thinking that if the front baffle was relatively thin and the speaker was placed against a wall, the wall would essentially act like the front baffle. I was thinking that the curve of the front would hopefully smooth out the dip in the frequency range. Thank you for the insight. It'll be a few years before I will be able to make a decent pair of speakers, but I still like to learn what I can in the meantime.

 

[Swerd]

I was thinking that if the front baffle was relatively thin and the speaker was placed against a wall, the wall would essentially act like the front baffle. I was thinking that the curve of the front would hopefully smooth out the dip in the frequency range. Thank you for the insight. It'll be a few years before I will be able to make a decent pair of speakers, but I still like to learn what I can in the meantime.

Baffle step has everything to do with front baffle width, but it has little to do with whether its edges are curved. Some simple and not-so-simple explanations of baffle step and how to compensate for it:

• http://www.salksound.com/blogtopic.php?id=2
• https://www.trueaudio.com/st_diff1.htm
• https://www.harbeth.co.uk/blog/posts/What-the-'crossover-network'-does...

When thinking about the "baffle step" response or loss in speakers, its important to remember that it varies greatly with the wavelength and speaker cabinet width. When speaker cabinets are narrow enough, they exhibit baffle step in the mid range, right where our hearing is sensitive. It tends to make mid range, if it occurs within the elevated baffle step region, sound as if its too forward. It's sometimes spoken of as sounding "nasal" or "shouty". Baffle step also makes bass, at frequencies lower than the baffle step, sound weak and thin.

See the blue line (trace 2) in the figure below. It shows the response of the SEAS L17 (17 cm aluminum cone) without any crossover or BSC. Pay attention to the hump that appears between roughly 400 Hz and just above 1000 Hz. There is a prominent (~3 dB) step up in response in that region. (I got that figure from http://speakerdesignworks.com/speakerbuilder.html) That's what baffle step looks like in a 9" wide speaker cabinet. Below ~400 Hz the response drops off steadily, making bass sound weak.

A well-designed crossover network will compensate for the baffle step to make it level with regions below and above the baffle step. In a passive crossover, the only way to do that is to flatten the elevated range. See the green line (L17 woofer with crossover and BSC, trace 3) or pink line (summed response of woofer + tweeter with crossover and BSC, trace 5) in the figure below.

 

[MrBoat]

Some of the optimal cabinet designs for full range drivers are lot of work and careful measurements like with the back loaded horn types. Either way, you're gonna pay. I have a friend in England (he works for the BBC) who jumped well down this rabbit hole and he showed me a design he attempted on some huge, and rather elaborate corner speakers and then discouraged me from it.

To quote him: "This way madness lies - trust me."

 

[TLS Guy]

I'm still curious about the sound of the drivers. I know they should sound bad, but I want to listen to it. I know that if you go to the actual shop, you can listen to the speakers and bring a set of your own to compare. I'd like to take a trip down there sometime in the future, but according to google maps, the place doesn't exist.

I was thinking that if the front baffle was relatively thin and the speaker was placed against a wall, the wall would essentially act like the front baffle. I was thinking that the curve of the front would hopefully smooth out the dip in the frequency range. Thank you for the insight. It'll be a few years before I will be able to make a decent pair of speakers, but I still like to learn what I can in the meantime.

Pipe Dreams! Do you want an accurate speaker or some highly colored retro outfit?

Without hearing those speakers I know I would absolutely hate them. I have done this long enough to know to a high degree of accuracy how those speakers would sound.

I can also reverse it, as I have demonstrated to friends that I can listen to a speaker I know nothing about and draw the FR curve to a high degree of accuracy.

Trust me those drivers will have a huge gap from reality. Some people do not want accurate speakers, but I do.