Weird design idea, I'm wondering if it is worth exploring. The concept is to use 3 mid-tweeters with 20 kHz (or, the desired) on-axis HF extension, angled at 15-45 degrees horizontally to each other, at the same vertical level.
The basic idea is that the only the center MT would play from mid-woofer XO (say, 1 kHz) to the system upper limit. The two angled MTs would (in the basic version of the idea) only play to maintain horizontal HF dispersion as the directivity of the 2" (or so) MT drivers causes them to beam. So there would be some inter-driver lobing/comb cancellation in the horizontal plane, but mitigated by the driver directivity/angled mounting/high pass filter. (In a more complex version of the idea, the mid-woofers are eliminated in favor of also playing all 3 MTs at these lower, still-looks-like-one-driver frequencies. So the side MTs in this case would have 2 separated pass bands, like 250-500 Hz and 7-20 kHz).
Example MT drivers would be Tymphany NE65W, or TC6FC00, or Dayton ND64.
Obviously the devil is in the details, any of which may not work out. Even to me this seems like a high-complexity approach to a minor problem with, supposedly, simpler known solutions. I've just never wrapped my head around how one is supposed to use a "real" tweeter, with a 4" face plate that needs to cross above 2 kHz, especially in an MTM, or with a first order XO, or if one balks at a $150 tweeter. (Admittedly: actual complaints may be smaller than shown.) But thus my floundering around to avoid them.
Now I find myself wondering about a cylindrical ring of 2" MTs. I've seen pictures of little balls of drivers used as omni test sources...why not IRL, ha? It's not that I just love complexity, or spending more money. (Don't listen to my wife's slander.)
But regardless, any insights from the more experienced (that's you) would be welcome. Has this angled MT thing been tried, or obviously unworkable for some reason?
Thanks in advance.
Patrick R.
The basic idea is that the only the center MT would play from mid-woofer XO (say, 1 kHz) to the system upper limit. The two angled MTs would (in the basic version of the idea) only play to maintain horizontal HF dispersion as the directivity of the 2" (or so) MT drivers causes them to beam. So there would be some inter-driver lobing/comb cancellation in the horizontal plane, but mitigated by the driver directivity/angled mounting/high pass filter. (In a more complex version of the idea, the mid-woofers are eliminated in favor of also playing all 3 MTs at these lower, still-looks-like-one-driver frequencies. So the side MTs in this case would have 2 separated pass bands, like 250-500 Hz and 7-20 kHz).
Example MT drivers would be Tymphany NE65W, or TC6FC00, or Dayton ND64.
Obviously the devil is in the details, any of which may not work out. Even to me this seems like a high-complexity approach to a minor problem with, supposedly, simpler known solutions. I've just never wrapped my head around how one is supposed to use a "real" tweeter, with a 4" face plate that needs to cross above 2 kHz, especially in an MTM, or with a first order XO, or if one balks at a $150 tweeter. (Admittedly: actual complaints may be smaller than shown.) But thus my floundering around to avoid them.
Now I find myself wondering about a cylindrical ring of 2" MTs. I've seen pictures of little balls of drivers used as omni test sources...why not IRL, ha? It's not that I just love complexity, or spending more money. (Don't listen to my wife's slander.)
But regardless, any insights from the more experienced (that's you) would be welcome. Has this angled MT thing been tried, or obviously unworkable for some reason?
Thanks in advance.
Patrick R.
So in the lower area, where the driver has to fight the most, you'd use the smallest cone surface of just one driver?
They won't play coherently. The filters for them already changes the phase. And yes, there will be tons of lobes. That's a pretty bad idea.
Well, the goal has to be the drivers being as close as possible to each other. With an angle of 15° to each other, the distance between the centers will be ~8cm minimum, otherwise the magnets and baskets would be in each others way and you need still material for mounting in the corners. At 10kHz the wavelength is 3,43cm, at 8k it's 4,3cm. Distance of the sound sources (center of the membrane) is 8cm between just two 16cm over all three. Don't you think you're missing something there?
If you move just a few cm, the sound will drastically change, it's horrible. Plus, your filters will add uneven phase errors.
Sure it can be done! With a little help from ds processors sound might be tolerable...
The SL-1 prototype is shaped like an hourglass or a dress form. Each waist is belted with 12 dome tweeters flanked, above and below, with two parallel belts, each consisting of 8 midrange drivers. The 4 woofers are paired vertically, with two each in front and two in back. The 22 power amps with over 1250W are controlled by internal DSP and a slick and friendly web app.
Read more at SoundSteer Technology from Lexicon | Stereophile.com
The SL-1 prototype is shaped like an hourglass or a dress form. Each waist is belted with 12 dome tweeters flanked, above and below, with two parallel belts, each consisting of 8 midrange drivers. The 4 woofers are paired vertically, with two each in front and two in back. The 22 power amps with over 1250W are controlled by internal DSP and a slick and friendly web app.
Read more at SoundSteer Technology from Lexicon | Stereophile.com
Sure it can be done! With a little help from ds processors sound might be tolerable...
That's a completely different principle. They are creating the sweet spot by delaying the drivers to let the soundwave reach the 'sweet spot' at the same time. They only use the drivers in a much smaller range and don't let the same drivers run to a different frequency band. The DSP has to be programmed to the exact location of the sweet spot with the room either per impulse delay measurement or simulation of the distances.
Just because you think it looks the same doesn't mean it acutally is the same.
That is also a different concept and does not use angled drivers, they are in a straight array. That will sound much better than the angled differently filtered drivers, it does not compensate for the stronger beaming in the higher frequencies though. Vertically, its dispersion will be quite narrow though it will also show comb filter effects vertically at very high frequencies (estimately >15k). That's not nearly as bad as horizontal lobes.
Beolab 90 is showing what can be done with dsp-controlled multiaxial speaker construction, and the Lexicon SL-1 uses pretty much same idea.
Like ICG says, it is not easy to make these work, the challenge is in wide variation of wavelength through audio band. Inter-driver distance means easily also delay off-axis, and these don't go hand-in-hand in a 3-dimensional construction.
Op's idea for sure doesn't work just like that, but as these examples show, it has been developed in many ways
Here we can read more of how B&O has done it, and we see some prototypes
B&O Tech: BeoLab 90 – Behind the scenes – earfluff and eyecandy
Like ICG says, it is not easy to make these work, the challenge is in wide variation of wavelength through audio band. Inter-driver distance means easily also delay off-axis, and these don't go hand-in-hand in a 3-dimensional construction.
Op's idea for sure doesn't work just like that, but as these examples show, it has been developed in many ways
Here we can read more of how B&O has done it, and we see some prototypes
B&O Tech: BeoLab 90 – Behind the scenes – earfluff and eyecandy
They are vertically in a straight line, that does not give you the horizontally interferences. That's not the same the TO proposed.
From my experience, multiple tweeters can be a good thing on the bottom line. In some cases multiple mid drivers too. I've heard many variations of that idea that I thought sounded great. Comb filter effects will happen, but since high treble (above 6kHZ) is very directional, if tweeters are angled away from each other by 20 degrees, I'd be surprised if you would hear much in the way of comb filter effects.
I'm hesitant to recommend doing this below about 7kHZ though, if you want the cleanest stereo imaging effects. The frequency range of about 800HZ to 6kHZ is where most stereo imaging effects happen without problems. Below about 1kHZ imaging gets blurred by the inter-aural crosstalk.
Having said that, Nola, Linkwitz and many others are doing open-baffle di-pole arrangements for midrange drivers/frequencies, and in the right room they sound fantastic. They effectively re-create a usable sense of imaging in the lower midrange, using listening room acoustics, rather than imaging cues embedded in the program material.
The thing to know about open baffle speakers is that in the right room (not small) they sound fantastic, but only in rooms that comfortably allow them to be at least 3 feet away from any walls. There's a psycho-acoustic effect that's such that they don't sound nearly as good when they are put closer to walls than that. So rear firing drivers are a little risky depending on the listening room.
I'm designing some tower speakers now, where I may have a rear firing 3 inch Peerless TG9 mid driver (480HZ - 7kHZ), that can be switched on or off, in it's own camber, separate from the main mid driver in its own chamber on the front of the cabinet. That way I can try it both ways and use it or not.
I'm hesitant to recommend doing this below about 7kHZ though, if you want the cleanest stereo imaging effects. The frequency range of about 800HZ to 6kHZ is where most stereo imaging effects happen without problems. Below about 1kHZ imaging gets blurred by the inter-aural crosstalk.
Having said that, Nola, Linkwitz and many others are doing open-baffle di-pole arrangements for midrange drivers/frequencies, and in the right room they sound fantastic. They effectively re-create a usable sense of imaging in the lower midrange, using listening room acoustics, rather than imaging cues embedded in the program material.
The thing to know about open baffle speakers is that in the right room (not small) they sound fantastic, but only in rooms that comfortably allow them to be at least 3 feet away from any walls. There's a psycho-acoustic effect that's such that they don't sound nearly as good when they are put closer to walls than that. So rear firing drivers are a little risky depending on the listening room.
I'm designing some tower speakers now, where I may have a rear firing 3 inch Peerless TG9 mid driver (480HZ - 7kHZ), that can be switched on or off, in it's own camber, separate from the main mid driver in its own chamber on the front of the cabinet. That way I can try it both ways and use it or not.
This guy has created a nice shaded angular array, with Visaton FR58 used as a midtweeter.
Radio & Audio Artikelen scroll to article 41.
It certainly seems worth exploring.
/g
edit: now I see you want them vertically on the same level - not such a good idea. Hank has played with that too, creating a circular array (his article 42).
Vertical and horizontal dislocation plays with different parts of our spatial localisation capabilities, plainly from the way we're constructed. Common sense is not to mess with it carelessly.
Radio & Audio Artikelen scroll to article 41.
It certainly seems worth exploring.
/g
edit: now I see you want them vertically on the same level - not such a good idea. Hank has played with that too, creating a circular array (his article 42).
Vertical and horizontal dislocation plays with different parts of our spatial localisation capabilities, plainly from the way we're constructed. Common sense is not to mess with it carelessly.
Last edited:
Thanks everyone for the responses and further reading. I suppose I should find or write a software tool that would show multi-source lobing with driver directionality, position, and orientation considered, just to get a better feel. But at any rate you all have saved me from trying to build anything along these lines in the foreseeable. Score one for the shark tank, ha. ;-)
-PR
-PR
^^Thank you for linking these articles and projects!
"Nice" polars with LEAP simulation!
https://www.by-rutgers.nl/PDFiles/LSP/LSP-Circular-Array.pdf
"Nice" polars with LEAP simulation!
https://www.by-rutgers.nl/PDFiles/LSP/LSP-Circular-Array.pdf
Cool project there, Juhazi! Thanks for that. Alas LinearX LEAP doesn't seem to be a going concern anymore? Oh well, looks like Octave (freeware version of Matlab) could use a better audio package....
Yes. And, incidentally, one of the reasons why I prefer 3-ways that avoid a crossover point in that frequency range.
Indeed, very interesting project, thanks for posting it, Juhazi! He solved the problem of the drivers being in each others way. In the end it's a good result with moderate effort, it's not perfect though. The highest simulated frequency for the polar plots is 12kHz and at 14k the response drops very fast, at 30° extremely fast. Everyone has to decide for themselves if that's 'good enough', if the limits of the reached compromise is acceptable.
The simulations also show why the original concept of the TO does not work and how important it is to get the sound sources as close as possible.
The simulations also show why the original concept of the TO does not work and how important it is to get the sound sources as close as possible.
Actually, I am wondering it it was woth the effort - uniform directivity above 2kHz is easy to achieve with a waveguide and spl requirement for home listening isn't a problem for modern tweeters.
You are right, that's not a problem. I see just two advantages:
1. You cover all of the mids to the highs with just one driver. That will without doubt sound very homogenous. I've got the Visaton FRS 5 x here, it's surprisingly good. A good tweeter will have still a better resolution and a better spl though, if you look at the simulations, it's only at 86dB on axix, 15° off it's already just 83-84. Most tweeters can do better and it limits your woofer choices unless you go acitve.
2. You can save money. 7 FRS 5 x cost ~70€ and you're covering mids and highs with it. A mid driver and tweeter are likely to be more expensive, especally if you add the crossover parts. Comparing it with the effort, it sounds much less tempting though, you have to work very precisely and it's not really pretty either.
1. You cover all of the mids to the highs with just one driver. That will without doubt sound very homogenous. I've got the Visaton FRS 5 x here, it's surprisingly good. A good tweeter will have still a better resolution and a better spl though, if you look at the simulations, it's only at 86dB on axix, 15° off it's already just 83-84. Most tweeters can do better and it limits your woofer choices unless you go acitve.
2. You can save money. 7 FRS 5 x cost ~70€ and you're covering mids and highs with it. A mid driver and tweeter are likely to be more expensive, especally if you add the crossover parts. Comparing it with the effort, it sounds much less tempting though, you have to work very precisely and it's not really pretty either.
- Status
- Not open for further replies.