Ultimate Solution - a 12 way loudspekersystem

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

Attachments

  • Kopie von HPIM2907bokenslicesm.jpg
    Kopie von HPIM2907bokenslicesm.jpg
    50.5 KB · Views: 278
Well, it's 'different' at any rate. :D

Different, but not unique in its idea though.
When I started off gathering info on the important bits on speaker design I came across the same issues Ronin's talking about. In fact, the speakers I built are still compromised by (among others) the limits in size I set for myself (and even then I ended up with four towers).

I prefer the KISS method, and to reduce the issues in reproduction of music it means every speaker driver should reproduce as less bandwidth as possible with the least amount of excursion. Hence the idea to use one speaker per octave with a high order crossover, thus resulting in a lot of speakers in one box (though not to mention the electronics).
Although I built open baffle to eliminate box resonances I can imagine limiting the excursion makes it easier to use closed/bass reflex designs.

I think the basis of the idea is right (just think of Dynaudio's Consequence) but not limiting myself to anything I'd use multiple open baffle towers with each tower containing MTM like arrangement of speakers. Much bigger in size but keeps the acoustic centers at one point as the phantom center. And use active crossovers of course.

A compromise on that could be to arrange single speakers above and below the tweeter in order of octave like WL-ML-T-MH-WH, and do the other channel upside-down like WH-MH-T-ML-WL.
 
Last edited:
The crossover in this 12 way may not be implementing what Linkwitz refers to as the "correct" topology and, even if it is, the earlier filters in the chain won't have achieved a full 360 degree rotation before the next one comes in (neither will have the last one by 20kHz). I'd have to sim it, but my guess would be the total group delay winds up closer to LR8 in the "correct" topology.
I'd never actually sim'd something so silly :rolleyes: so I gave it a shot. This is the SPL, phase and group delay of 11 parallel electrical bandpass filters using the OP's frequencies. It's an 11-way as he has 2 supertweeters. The lowest bandpass is an LR4-15 highpass and an LR4-30 lowpass. The highest is an LR4-15360 highpass and an LR4-30770 lowpass. The others are spaced an octave apart in between. Notice how the phase wraps through 360 degrees every octave.

attachment.php


attachment.php
 

Attachments

  • silly-spl-phase.png
    silly-spl-phase.png
    9.1 KB · Views: 324
  • silly-gd.png
    silly-gd.png
    4.1 KB · Views: 496
Also note that the average SPL is down about 15dB from the nominal driver sensitivity. That's strictly because of the interference of the filters. Add in the random placement of the drivers and I shudder to think what the acoustic response would be like on and off axis.
 
Also note that the average SPL is down about 15dB from the nominal driver sensitivity. That's strictly because of the interference of the filters. Add in the random placement of the drivers and I shudder to think what the acoustic response would be like on and off axis.

Great simulation Catapult.

Yes, if you have an LR4 filter an Octave away from another LR4 (both of them 6dB down at their corners) then you will be throwing a lot away in the filter. It is possible that there is a better summing configuration and of course the driver delay has to be considered as well, but Octave bandpasses are not the answer for maintaining sensitivity.

Still, the real key is painstaking selection of the crossover components.

David S.
 
I'd never actually sim'd something so silly :rolleyes: so I gave it a shot.
Interesting. I'm not getting my head around the result, though; if the bandpasses are in parallel I wouldn't expect any driver with more than a 360 degree rotation. Since we're being silly, can you share the schematic?

Still, the real key is painstaking selection of the crossover components.
Good point. I was starting to go down the wrong track there by thinking the key was painstaking use of bold and wondering about the sonic degradation from using of italic or underscore. :p
 
Different, but not unique in its idea though.

Indeed. I seem to recall a few other speakers with a ludicrous number of drivers & XOs; mostly dating from 1960s - '70s patents, though a handful were built.

When I started off gathering info on the important bits on speaker design I came across the same issues Ronin's talking about. In fact, the speakers I built are still compromised by (among others) the limits in size I set for myself (and even then I ended up with four towers).

I'm not sure what you happen to regard as being the 'important bits of speaker design', as this will rather depend on how you define / interpret them. 4 towers per se is no big deal.

I prefer the KISS method, and to reduce the issues in reproduction of music it means every speaker driver should reproduce as less bandwidth as possible with the least amount of excursion.

Really? According to whom? While I'm with you on minimising excursion to maximise dynamic BW, using a myriad of drive units that you attempt to blend into a cohesive whole with a massively complex XO is hardly what I would consider condusive to the principle of KISS. Especially when one considers the potential phase related issues (for starters) with multiple XOs in our critical hearing BW.

Hence the idea to use one speaker per octave with a high order crossover, thus resulting in a lot of speakers in one box (though not to mention the electronics).

KISS, huh? ;)
 
Interesting. I'm not getting my head around the result, though; if the bandpasses are in parallel I wouldn't expect any driver with more than a 360 degree rotation. Since we're being silly, can you share the schematic?
I just have the demo version of LspCAD so I couldn't save the file. It was built with active transfer functions so there's no interaction between passive components. In the real world that interaction is a real possibility, maybe making the whole thing resonate as Speaker Dave mentioned. The general layout is like this except 11 of them instead of just 3.

Code:
    |--HP--LP--|
in--|--HP--LP--|--out
    |--HP--LP--|
 
I'm not sure what you happen to regard as being the 'important bits of speaker design', as this will rather depend on how you define / interpret them. 4 towers per se is no big deal.
What I've read so far from both big names like Siegfried Linkwitz and reading between the lines of speaker reviews is that more (cone) surface means less distortion, which is logical and what I remember from hearing planar speakers. I rate that as high as a uniform polar response.
And four towers can be quite a spacehog in average dutch livingrooms. But just wait until I can buy the house I want with giant livingroom :)

Really? According to whom? While I'm with you on minimising excursion to maximise dynamic BW, using a myriad of drive units that you attempt to blend into a cohesive whole with a massively complex XO is hardly what I would consider condusive to the principle of KISS. Especially when one considers the potential phase related issues (for starters) with multiple XOs in our critical hearing BW.
Well, the KISS principle only goes so far but using active crossovers like the DCX2496 helps. :) Simple does not necessarily mean less equipment. :D
Phase related issues are still a hot topic of discussion though and I haven't seen nor heard any conclusive evidence either way. My own experiments have been inconclusive and only phase differences between left and right have made a big impact.

KISS, huh? ;)
*starts Tom Jones on the stereo in the meantime*
My reasoning went something like this: reproduce loads of frequencies cleanly -> use plenty of surface per octave -> use less octaves per driver -> limit bandwidth with active crossover (can do phase shift and delay as well)
It's the trying to get a perfect polar response that limited my set to a 3.5.2 way but now I feel like hearing what a 6-way might be capable of. Damned curiousity...
 
The whole thing in here,smells of "Flat Earthism",once more.Ronin tries to prove that by throwing a number of drivers in a box,and a xover,a speaker will emerge,like the "Lady of the lake".Okay,let him finish the plot,make his measurements,translate from German,and comments will flow.

I am curious,btw,to know the speakers that ,the fridges,were compared to,and found wanting.Ronin doesn't say.Ah,and the rest of the equipment used.


B.L.
 
What I've read so far from both big names like Siegfried Linkwitz and reading between the lines of speaker reviews is that more (cone) surface means less distortion, which is logical and what I remember from hearing planar speakers. I rate that as high as a uniform polar response.

Up to a point, greater pistonic area is certainly a good thing for reproducing lower frequencies and I doubt you'd find anyone arguing against that; you're shifting more air for a given excursion of the powertrain, thus potentially (assuming well designed drivers) reducing distortion. A well designed 15in woofer should shift more air with lower distortion than an equally well designed 8in woofer, assuming you don't ask it to do something daft, like go up to a couple of KHz. However, I think you'll discover that Linkwitz et al, are refering to precisely that: pistonic area in a given BW. They are not refering to 12 disparate drivers covering different frequencies. That's another matter entirely. I haven't recently seen Siegfried creating a 12 way system for e.g. IIRC, 4-way was as far as he went.

No offense, but I suspect you haven't heard a system with a truly lousy polar response. If you had, you'll soon discover why it's important. ;) As you apparantly haven't, I'm relieved for you, because it's not pretty.

And four towers can be quite a spacehog in average dutch livingrooms. But just wait until I can buy the house I want with giant livingroom :)

We're just as badly off in the UK, so you're not alone on that score.

Well, the KISS principle only goes so far but using active crossovers like the DCX2496 helps. :) Simple does not necessarily mean less equipment. :D

Er, I beg to differ on that. While I'm in complete agreement that active XO's done well are going to beat equally well done passive XOs, as far as I'm concerned, simple means precisely that: simple. As in 'uncomplicated.' As in 'not trying to split the audio region into a vast number of zones for no good reason.' And there really is no good reason. Crossovers, especially passive XOs are the hardest part of most multiway loudspeakers to get right; the XO itself isn't necessarily the problem, but achieving a decent integration between different drive units is. 11 of them, with different types of driver is the stuff of nightmares & is 99.999% certain to cause more problems than any well designed drive unit ever has. Especially if stuck right in the middle of our critical hearing BW, and the drivers have been located on the baffle with little or no regard for reasonable integration or system polar response & potential delay times well in excess of the audibility threshold at that particular frequency.

Phase related issues are still a hot topic of discussion though and I haven't seen nor heard any conclusive evidence either way. My own experiments have been inconclusive and only phase differences between left and right have made a big impact.

Minor differences can certainly be processed out depending on severity and the frequencies involved; our brains are remarkable things. Multiple issues throughout the audio band, and particularly in the centre of our critical hearing BW? Good luck. ;)
 
Last edited:
Bovine duckery!

Yes, it's prediction time. My crystal balls are in play.

We will never see a response graph! Look, I might even settle for a photo of the completed inner workings, everything connected and ready to rock! If that fane does 15 - 30 in a sealed enclosure I want some!

Terry :)
 
IIRC, 4-way was as far as he went.
I think that also had to do with keeping the speaker a point source as much as possible by using the tweeter as low as possible (but unfortunately it still keeps the crossover in the sensitive area).

because it's not pretty.
I can imagine what it sounds like after I "activated" my Tannoys and improved the polar response, what a difference just that made.

drivers have been located on the baffle with little or no regard for reasonable integration or system polar response & potential delay times
That's the first issue I thought of after seeing the photo. I'd start off by better lining up the driver layout above and below a single supertweeter (so what if the box is 2 meters high), and when still going the passive route use the Linkwitz topology.
I'm very curious to see what the result would be in both measurements and sound quality. Too bad I don't have the time to build my own experiment. :(
 
the physics of every single electromagnetic speaker in a speakersystem that disables them to reproduce more that four or five tones really exactly same as the original source - so the first reason is physic.

Are you sure about this? I'm not - it's not uncommon for a driver to cover 200Hz to 4kHz (etc) without issue these days. I think your basic premise is flawed, please show me the physics!

But... I think even if it was correct;- the you have created other problems with your design that will overshadow your design regardless:

1) Colouration. Unless it is open back the cabinet will reflect the sound around - eventually it will be heard as a rather serious 'boxy' sound.

2) Focus. I'm not usually big on focus, but by spacing your tweeters that far apart then you have the sound for a cymbal (for instance) arriving from two different places, causing some very odd imaging.. also carpet will eat the treble.

3) Complexity. Audio doesn't like complexity, you will hear the 12 crossovers , and your ears will be confused by the spacing of the drivers. The best sound these days comes from single drivers or perhaps 2 drivers in my experience. Remember we hear all sound via a single ear-drum on each side of our head!!

I think you could improve your design by making a pyramid shape and putting the highest frequencies at the top and working down - in frequency order - to the floor with the bass. You also need an enclosure for every open back unit - can't recall if you had that.

I would be interested to see a design with a wide midrange and fill in tweeter and bass, but at least you are building and experimenting which is all good stuff and more than many do, just not sure it will sound how you imagine.
 
Hmm, I'm not seeing where the excess phase is coming from. Would have to sim it.

An 8th order bandpass (4 high and 4 low) has to swing from 360 degrees positive at the low end to 360 degrees lag at higher frequencies. Generally you'l have half that phase shift at the corner frequencies so: +180 and -180 or a 360 degree swing each section (repeat 11 times = 3960 degrees.). How you have the sections add is not something Ronin is telling us, or is apparently aware of, but you always get the best response if you can get the phase swing of one section to overlay the next. Did we mention that the drivers have phase shift due to their response and physical depth? Did we mention that textbook crossovers don't acount for the driver as a load?

David S.
 
Last edited:
To me this system looks like one of these that emphasise on one aspect - while completely neglecting all others. He is not alone though with such an approach.

Reduction of IMD IS an important parameter but not the only one.

Any single cone is theoretically - and of course to some degree only practically as well - able to handle as many frequencies at once as one could imagine. In the same manner as the air is able to transport many frequencies at once and also like the mechanical parts of our ear are capable of handling. I think the OP's imagination doesn't allow for something like this and therefore all practically existing systems from single driver to four- or five-ways in the world won't work - probably not even when standing in front of one and actually listenenig to it.

Regards

Charles
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.