2. tech-talk
  3. 1287042179

What if you designed your ultimate speaker?

I posted the following the other day as a continuation of my response to a thread entitled The Best Tweeter Design (which explains why it starts out the way it does). However not only was this extended ramble really out of place under that topic, it drew no comment, so I thought I'd repost it under this new heading and try again. (I should also mention that I've never built any speaker, and am not technically qualified to do so.) Please fire/dream away at will!

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

It's always struck me that the presumed need for exotic materials in so-called dynamic (pistonic) tweeters could be eased, if such drivers' physical dimensions were optimized for more limited bandwidths -- in other words, if multiple, crossed-over domes of progressively smaller diameters were used to cover the region above roughly 3KHz (give or take a KHz) that's normally handled by a single circa 1" dome. This would A) ameliorate the conflict between rigidity and low mass that's otherwise necessitated in order to push the resonant breakup mode sufficiently beyond the passband, without resorting to materials any more costly or hard to work with than the ubiquitous aluminum, while B) greatly increasing power-handling capability and C) increasing and smoothing (making more uniform) lateral dispersion with respect to increasing frequency.

Of course multiple drivers, and the crossovers for them, are more expensive than a single one, but exotic diaphragm materials (or horn-loading) can be expensive too (and since when is expense a determining factor in the high end?), and, when it comes to conventional dynamic tweeters, exotics do little if anything in and of themselves to improve power-handling and dispersion qualities. (Horn-loading improves power-handling at the deliberate expense of more limited dispersion, but that's another argument.) I know Linn makes a tweeter array consisting of multiple domes culminating in a diameter around half the conventional size (which I believe use a plastic-film diaphragm material), but I'm not sure if anybody else does anything like this.

Then again, conventional wisdom is that fewer drivers and crossovers sound better, and although I can appreciate the virtues of single-driver speakers in practice, I don't necessarily adhere to this paradigm in theory: I think the problem with crossovers is just the opposite -- i.e., that they're called upon to mate drivers which are too physically dissimilar from one another to merge coherently, and which are operated over too wide a passband to be optimal in terms of dispersion, distortion, and power-handling/dynamics.

If I had my own speaker company with sufficient resources and were making a clean-sheet, full-range, cost-independent design, I'd want to research creating a speaker in which each driver handles only 1/2 an octave, which would mean a 20-way design (there being about 10 octaves in the audioband as normally defined between 20Hz and 20KHz). Why a 1/2-octave design, when that's way more limited in bandwidth than is needed to surpress a diaphragm's own resonant frequency? Because the prevelant distortion product from any induced vibration resulting in a decreasing monotonic sequence is one octave above the fundamental of the input, or the second harmonic. This effect is most notorious in the bass frequencies, where for instance a 40Hz input might yield quite a high percentage of 80Hz in the output (not always seen as a bad thing for certain purposes!), but it pertains at increasing frequencies too, although I'm led to believe in decreasing proportion.

So my concept is, if you want to make a truly low-distortion speaker, one way to achieve this would be to cross-over all the drivers such that the 2nd harmonic of the lowest frequency included in the full-output passband of each is already surpressed by its crossover. This close-cropping of the passbands would also have the benefits of permitting closely matching the physical designs of adjacent drivers, while allowing the size of each to be optimized for smooth, wide dispersion within its passband, and the employment of simpler first-order crossover filters, but without the usual low-order penalties in terms of dynamics or power-handling. And none of the individual drivers would need to be terribly exotic, because the demands placed on each would be minimal. It seems to me the overall result could be more coherent and continuous sounding, with greater effortlessness, lower distortion, more uniform in-room response and a wider listening window (and maybe greater efficiency too) than conventional multi-way or single-driver designs. At least that's my idea. (I'd incorporate a few others too -- maybe below.) Has anybody ever made anything like it?
zaikesman
zaikesman OP
3,017 posts
 
C1ferrari and Mapman: Interesting ideas both. Has either been tried? Is either possible? I would assume at least Mapman's is, and it seems to me that a Walsh driver in particular may stand to benefit from a possibly more physically compact motor, if that in fact pertains to field-coil designs. But I know little about field coils or plasma drivers. I'd be interested to hear more about your concepts and reasoning, if you feel like sharing.

I'd note that both your ideas hew to the more traditional ideal of achieving a full-range 1-way. In theory one of the merits of this approach, among other things, could be that microphones are also full-range 1-way transducers, and you can take the position what we are trying to accomplish with a loudspeaker is an inverse process to the recording microphone. (One supposes the complications associated with the scaling-up of that inverse process are usually thought to involve certain tradeoffs, resulting in the preponderence of multi-way designs.)
tiggerfc
153 posts
 
I agree with C1ferrari. :-)

I heard a set of B&W 800D Sigs several months ago. Problem I heard was I could here the crossover point distinctly from driver to driver. It was terrible and thought it was poorly executed. I don't want to hear several different drivers operating independently. I want cohesion regardless of the number of drivers. But there in lies the problem with 17-18 indepent drivers all crossed over at different points. If this is designed as is the customary tower speaker cabinet then every time I listen to a rapid scale orchestral piece my ears will dance from the bottom to top and down again. But from the stage(in live performance), the sound still originates from the same points. Not to mention they are all seperated by frequency, not by instrument, which the latter could be an interesting implementation if you had the gear to set it up... i.e. A mixer, the original un-mixed master, and a speaker for each channel.

Maybe what you need is an array style cylinder with 9 - 4" woofers and 9 - 1 1/2" tweeters in a spiral formation. I bet that could solve the problem of hearing the crossover points between drivers and still acheive low distortion.

Now, if I designed my own?

The speakers I'm currently designing I kept to two tweeters- an implementation of ribbon and di-pole along with a mid-driver and woofer. Along with a specific crossover frequency that will essentially keep the ribbon quiet until really necessary or the point where I believe standard tweeters fail.

I'll also note that I don't know too much about acoustics but I do know the fundamental characteristics. Not to mention, I've seen the insides of some high-end cabinets and I say "Where's the R&D?" "Where's the cost?" "Why's there cheap MDF in here?" "What's so hard about this design that it cost $20k?" "Huh, I might as well build my own!"

Pretty much said in that order too! But to do it yourself you tend to pay a lot more since you have to pay retail for each piece. I estimate the cost of my own design to push $6,600. Mostly due to the cabinet design, the bracing characteristics that'll allow it to be strong enough to be used as a stilt support for your beach house and that I'll be outsourcing the crossover. Also installing hi-end drivers, no less. The ribbons are pushing roughly $600 each.

Makes me wonder that when I'm finally done with it and plug 'em in whether they'll destroy speakers costing ten times as much at retail. As far as I've seen, no one has ever attempted or mass-produced this design.

And it seems to me that everyone designs something in regards to cost-effectiveness. In every design I see, that is another failing point. These weren't necessarily designed outside of those constraints, but not once did I ask how much it would cost to do a specific thing before deciding to add it in. I came up with the cost estimation AFTER the design was complete. And I cut no corners. I dare any manufacturer or any DIYer to do better. Maybe my design will knock the socks off those AAD 2001s I love so much. We'll see.
c1ferrari
464 posts
 

Hi Zaikesman,

I haven't researched the full-range plasma; however, your observation regarding the microphone intrigues...I can get my head around the mic as a mechanical-->electrical transducer, but with a plasma mic?? Boy, it's time to consult with the theoretical physicists :-)

Thanks, Tiggerfc, it's nice to have company!

Vbr,
Sam
zaikesman OP
3,017 posts
 
Hi C1ferrari: I never thought about a plasma microphone. That's wild, and might well be more executable than a full-range plasma driver. How it would sound compared with conventional microphones, who knows? Very interesting. (What I have thought about before is what a large-panel planar-magnetic or electrostatic microphone array would sound like -- think recording into a pair of Maggies.)

Hi Tiggerfc, you raise a potentially problematic point, one which I had considered but really just don't know how it would play out in the listening: How *physically* coherent would, or could, a 1/2-octave, 17-way design sound?

My contention -- although from the response I've gotten so far it's clearly a counterintuitive one -- is that, purely from a perspective of crossing-over one driver to another, a much more restricted bandwidth per driver, using more-similar adjacent drivers, would actually *help* make any one crossover point *less* audible, and designing and implementing those crossover points *easier* -- not more audible or difficult, as some seem to assume. (Though I strongly suspect this is true as far as it goes, how sonically significant the resulting distortion reduction would be, or how the overall presentation would ultimately sound compared with the regular way of divvying-up duties, I can't say.)

But Tiggerfc isn't mainly arguing with that: He's saying that, in total, the sheer spread of physical separation across all the drivers from top to bottom needed to implement such a design, even if each individual driver is located hard next to its neighbor, would inevitably shoot any chance of getting it all to sound like a quasi-point-source -- as many, if not all, conventional 2- and 3-way designs claim or ostensibly strive to achieve. (And despite any superficial similarity, it wouldn't be a line-source either. And mirror-imaging a doubled array of drivers to simulate a centrally-located point-source, like in an MTM array, or as is sometimes done across an entire 3- to 5-way design using anywhere from 5 to 9 drivers, wouldn't seem to be an option in a 17-way concept for obvious reasons, unless you live in a gothic cathedral.)

Whether or not the sound at the listening position would actually suffer from this presumed effect -- and I'm not sure that it would -- would definitely need to be a concern, maybe one answerable only by making a prototype. (A speaker testing this idea could be roughly prototyped without needing each driver to be individually optimally-sized for the wavelengths within its passband -- you could just use multiples of the same 3 or 4 drivers that you'd use in a conventional design -- or including the low-bass drivers at all, which wouldn't be a concern from the standpoint of frequency-band localization, and would probably be side-mounted on the completed design in part because of that.)
onhwy61's avatar
onhwy61
5,351 posts
 
This is my ultimate speaker design. It may not make any sense at all let alone work as I intend it to.

It's not so much a speaker as a room setup. You have three rooms. Imagine a large rectangular room. Divide that room in two by erecting a wall across the room approximately two-thirds down the length of the room. Then evenly divide the resulting smaller room by erecting a wall perpendicular to the first wall. The two smaller rooms are the left and right chambers and the larger remaining space is the listening room. The key to the design is the common wall separating the chambers from the listening room. It should be extremely rigid and non-resonate yet not too thick in depth. Built into this wall are two floor to ceiling narrow slits. The slits should be symmetrically placed and set wide apart on the wall. In each chamber room you would situate some sort of loudspeaker. It could literally be any speaker, but I suspect it would be better to have a full range speaker with low distortion and capable of high SPL output. In effect the chamber rooms are the loudspeakers. The sound would only enter the listening room via the slits in the wall.

My goal with this design is to try to recreate what music sounds like in a concert hall. I believe the sound that emerges from the wall slits would be time and phase coherent. The slits would function as a line source radiator. Tonally the sound would be rolled off in the treble, hopefully just like in a concert hall. But that could be controlled by EQing the speakers in the chamber. This system should also image and soundstage quite well.

I honestly don't know it this setup would work. One thing in its favor is that it requires no new technology and could be implemented in a DIY manner. The obvious downside is that it requires a large amount of room, but it should be size scalable. The real question is whether what I propose will perform better than simply placing a speaker in a normal listening room?
More to discover