Welcome! Need support, you got it. Or share your ideas and experiences.
Return to Planar Speaker Asylum
64.252.2.149
In Reply to: RE: Apogees and ethics - comments posted by Satie on February 10, 2012 at 13:36:55
Agree that a polymer-backed ribbon can have lower mass than an all-aluminum one. But I thought you liked the sound of the Apogees better than the BG's? Or is that the top-bottom ribbon configuration rather than the material of the ribbon? (Wendell calls everything not foil a quasi-ribbon, and I guess since he invented the term he gets to -- but I still haven't come up with an unambiguous, unclumsy way to distinguish between the various varieties).
The problem with wide spacing of drivers is that not only do they become directional, but the lobes multiply until you have cauliflower pattern -- and the nulls go all the way down to zero, and in the listening area, too (though not necessarily the sweet spot). Perhaps the nulls at the frequencies covered by the tweeter are so narrow that they aren't perceptually significant? Certainly the broad quasi-ribbon tweeters exhibit similar lobing problems, but to a lesser degree, since they aren't 2" wide.
I know of some other speakers that use rear radiators as well. The idea is to maintain a polar response that's fairly constant with respect to frequency; in a conventional box, the bass is omnidirectional, the highs are cardioid, and that reduces the sense of spaciousness. So I understand why Apogee used the rear ribbons, just not why they drove them out-of-phase with the midrange backwave, creating a suckout at the crossover point. Or rather, I suspect as I said that it had to do with their desire to avoid triggering sympathetic resonances in the midrange ribbon, had the front and rear tweeters been driven in phase it would have done that, but with the two out of phase, the forces on the midrange ribbon would cancel. I'm not sure whether Arnie Nudell ran the rear EMIT's in or out of phase with the front ones, but there were fewer of them and as I recall, he aimed for a cardioid pattern because he found it imaged better. Which would mean they were phased as bipoles, but at a lower SPL (which they likely were because there are fewer of them than there are on the front).
Hadn't thought about the fact that the tweeters could be narrower, though really, the Magnepan ribbons are 1/4" and go down to 3000 Hz, and the Apogees are four times as thick, so would the Apogee tweeters have a torsion problem? I was speculating on the flux strength, Apogee had to work with magnets of limited strength, which according to I think Graz is the reason for the low impedance of some early Apogees -- they needed the current to get reasonable efficiency. Not a problem anymore.
I don't think in an MTM arrangement you'd get a dipole null at the tweeter without some kind of relief port on either side of the tweeter, not until the out-of-phase wave had traveled around the barrier, anyway, and that doesn't really happen at the speaker itself because I assume the waves haven't diffracted fully, it seems they don't until you're a couple of meters out. If you left enough space to make dipole nulls, you'd destroy the efficiency of your midrange drivers and I think couldn't use a single magnet assembly. Plus you'd start to get lobing on the mids, you really want to keep the midrange drivers very close together so they operate as one.
With woofers I suppose you could do a W-M-T-M-W arrangement, but below 80 Hz there isn't much reason to and if you left enough room for dipole cancellation at the center you'd halve your effective baffle size.
I think you're right about wanting to keep the bass freqs away from the mids. Lobing isn't a problem at those frequencies, so you only have to deal with lateral image spread -- not ideal, but not too terrible, as those things go.
Follow Ups:
I am not sure that I like the Apogees in the mids and highs more than I like the combination of Neo8s and Maggie ribbon any more. I don't feel that I am losing much by crossing at 250hz, and I very much like not having the torsion screech distraction, even if it is only occasional.
I recently changed the midrange XO to allow a narrower range of HP adjustment, and a more appropriate level adjustment range, and bypassed the HP's Sonicap 1 cap with a foil cap, which changed the perceived FR and smoothed out a bit of grain that makes things that much better. So I am now reworking the tweeter/mid XO to make use of the neo8s further up with less lobing. In the process, I am listening to the mids alone, the bass + mids, Bass + tweeter. Taking your comment in mind about shelving down the treble output rather than tilting up I am experimenting with lower tweeter levels, which effectively raise the XO. There is an interesting level match I reached last night which gives a really interesting improvement in image saturation and air on Billie Holiday's last Columbia recordings, which were two mike stereo from 1958. I need to try this with more recordings and see over what range of adjustments this effect can be maintained.
side note - In order to make the adjustments easier, I am going directly off the DAC so that I don't have to adjust for the tube preamps sensitivity to load impedance - which changes with XO adjustment.
There is one point on BGs + mag ribbon vs. apogees, in that I did most of my Apogee listening at a dealer's where driving electronics were routinely an order of magnitude more costly. Levinson amps and Symphonic line Pre are hard to beat not to speak of the heavy turntables and CD players that each cost more than my entire investment in audio. My home setup has never been that strong on source components, no matter how tweaked, and my amplification, though good or very good, is not "as goo as it gets" as was that at the dealer's. The only thing where I may have the upper hand in my systemm is that the DAC is newer - though the output stage is not discrete but Michaelson's best trickery with 4 dual opamps of the cheap veriety 2 duals per channel - gain and buffer - same as in his better DACs and preamps, but not his top of the line discrete products. Though he does use the same buffer design and op amps for even his terrific sounding trivista DAC (his mini tube nuvistors and trivistors were buffered with op amps).
So bottom line, I don't know anymore whether the Apogees would win out on a level playing field vs bgs+mag ribbon.
Re WTW ot MTM idea, where I was thinking the T would be in a physically separate frame, are you sure there would not be a null? do you really need that much of a gap? I get noticeable dipole cancelation with the space between the bass panels and mid/tweeter frame opened to as little as 1/4", and much more at 1/2" - perceived as a drop in the low end extension of the array. I use a strip of tape to keep the gap closed (touching only the mid/tweet panel).
Re bipole tweeters - I know Nudell did it bipole - I guess for the same reason you quoted.
The Apogee tweeters did have a torsion problem. In my listening it was the only flaw in them that was not addressable by a change in ancilliaries or a trivial mod of the XO or drivers. The way to address it was to raise the XO by an octave or more - leaving the mids to cover another octave - which they did not do that well due to their weight and width, or to sharpen the XO to 3rd or 4th order - which you can do today digitally without phase issues, but not back then.
I see your point on the design considerations forcing the gap to be small enough so there is no null.
Good about the Neo's, since the chance I'll have time to build a midrange ribbon is somewhere between nil and zip. :-)
I'd think that 250 Hz is a pretty ideal crossover point for the Tympanis. AFAIK, there still isn't anything that can beat Tympani midbass. Sure, you're going to get a bit of lateral smearing from the offset, but to my ears smearing is pretty minor at those frequencies. Of course, I haven't heard the Neo 10, but (besides their cost) I remain a bit suspicious of their beaming above 2 kHz. I think I'd rather compromise at 250 Hz than in the midrange.
(BTW, did you know that the Neo 10's use the force of the magnets to bow the perforated metal to accomodate the diaphragm? Just read the patent. They also have pleats on either side of the diaphragm to reduce stresses and improve damping. Also learned something about the failure mechanism of overdriven planars -- apparently, heat leads to differential expansion between the film and the foil, and when combined with large mechanical stresses, the diaphragm can wrinkle, delamination can occur, and conductors can break.)
One of the changes they made in the Tympani IVa was to give it a down-tilted response. I don't think there's a right or wrong here -- it seems that works recorded in a large hall need the downtilt, works recorded in a smaller space don't. But with so many recordings hyped in the highs, a downtilt can't hurt. I didn't know that Billie Holliday made any stereo recordings.
I have the same problem you do with my memories of speakers, though these days, it's more likely to be that amps back then were underpowered and colored. Forex, how much of the magical imagining I remember from the IRS V had to do with the speakers themselves, and how much to do with the big Conrad-Johnsons? And how much with the acoustics? I'll never know.
Not sure about how much of a gap you'd need. I looked up the formula for acoustic impedance of a slit, but realized I didn't know what SPL the ribbon could tolerate so didn't bother running the numbers. It depends not just on the width of the slit, but on the depth of the baffle, and by the time you'd plugged in all the numbers and figured out ribbon resonances and coupling you could have just tried it 20 times over.
BTW, I found a guy (Valvetude) who's running his homebrew 0.75" neo ribbons down to 200 Hz. He gets +/-0.125" excursion from them and says he can play them at deafening levels without distortion. Assuming you can get a sufficiently uniform field, I wonder what the practical limit on this is -- ribbon fatigue? The reason I wonder is that I assume narrower ribbons are less susceptible to torsion, and of course they also have lower mass, better dispersion, and more uniform field strength.
I remember following Vavetude's project as he was posting it. You need to take into consideration his heroic effort with giant magnet structure and fine tuning of every component, surface, location. It is something that you would expect to cost a few thousand a piece and something like most of a Genesis system once you add the appropriate woofers to complement it.
I did not know that the Neo 10 uses the magnets to form the shape of the magnet boards. Very cool idea. Yes, I was aware of the heat failure mechanism. That differs from mylar, which just melts.
I think 250 hz is an ok spot rather than perfect. The ideal would be more like 150-200 depending on how solid the structure holding the mids is.
I have a rather good memory of what I heard, since I got to hear the Centaurs and Slants many times, and got exposure to the Stage and its derivatives sufficient to get a permanent feel for it.
I can say that the Apogees are still somewhat more dynamic sounding than the Neo8 - probably because the ribbons are loose and not stretched as much as the Neo's. I think that limits their output some when you go beyond a certain rather high volume, they compress a little. But otherwise, I think they are just more refined and texturally right. I have heard my Neo8s with a variety of amps, so I have some idea of the sound of the array itself.
I am a little surprised that you can get a more correct imaging with the down tilted top end - as pointed out in the Sonus Faber Amati Futura review from Atkinson. Very interesting.
I guess we just leave the minimum gap size for dipole cancelation aside. From my experience it seems like 1/2" is already over the threshold in the 300 hz area.
I gather that large Neodynium magnet assemblies are insanely expensive right now, I've seen some figures on DIY audio in the $1-$3 thousand range. (Though some have done quite well with factory seconds, etc.)
I'm not sure the Neo's are necessary for sound quality, for anything except the ultra-wide ribbons (and then you wouldn't necessarily need huge ones). The Neos give you great efficiency but as Wendell points out it's cheaper to buy a big amp. I read somewhere that stronger magnets give you lower distortion, but I'm not sure why that would be the case, I'd expect ribbons to be predominantly air damped and distortion to depend on the uniformity of the field rather than its strength, the precision of the air gaps, and the resonances in the ribbon. Am I missing something? Better electrical damping at low frequencies, perhaps?
Ditto Valetude's 250 pounds of steel, I gather he kept adding it until he eliminated most traces of saturation in FEMM but if you don't need the efficiency, I assume you wouldn't have to go that far.
I read last night that Linesource's 2" ribbons are crossed over at 80 Hz? With an 8-pole crossover.
I've heard it said that ribbons have an advantage in dynamics. I'm not sure why. I'd expect the planar-magnetic field to increase distortion at high excursions. I'm not sure about the compliance of the film, does it become nonlinear? Otherwise I'd expect it to store energy, as of course it does, but AFAIK that wouldn't cause compression. Maybe the ear, while it integrates over time to judge frequency response, interprets time smearing as a loss of dynamics or a change in timbre? I'm thinking back to analog synth days, and the effects of the ADSR envelope generators -- stored energy would have the effect of reducing attack and reducing decay, and both of those make substantial changes to timbre.
Do you remember approximately where John Atkinson said that about the Amati's FR and imaging? I just read the review, must have skipped over it (feeling guilty about doing this stuff right now when I should be getting to work).
Well, one fact is and would be that a ribbon in its loose suspension does mostly have air as restriction and load.
Compare that to a membrane that is suspended and stretched!
The stretching in itself is a very strong force that directly affect dynamics as it is actually forced not to move but still does due to flexing in material.
As a perfect example of this is electrostats that really needs a pretty strong suspension that actually forces the membrane to its stationary place between the stators.
This because it would otherwise automatically get attracted to one of the stators and get stuck.
The solution to this is make the membrane area larger to compensate for the lack of motion.
This of course is mainly true in the bass region where large excursion is needed.
The one who succeeded was the one who didn't know it was impossible.
On the other hand, the forces on an electrostatic are essentially limited by the breakdown voltage of air. So there's an upper limit on the force that can be applied to the diaphragm at a given stator spacing, which I think is about 1/10 that of a planar magnetic. Whereas in practice, the low frequency excursion of large planar magnetics seems to be limited by the position of the magnet assembly, which is determined by efficiency and magnet strength. They'll play until they slap the pole piece. So as far as I know, for a large planar, the limitation on LF output is imposed essentially by the cost of the magnets, and of a mechanical assembly of sufficient strength to accommodate them. For a smaller planar like the Neo's thermal damage could also be an issue, and also maybe distortion at large Xmax's caused by the non-uniform field. As far as I know, the compliance of these plastic films is a constant, so I'm not sure where dynamic compression would occur, unless it's a subjective consequence of rising non-linear distortion? And I wouldn't expect most of these concerns to apply at higher frequencies, where excursion is low.
Edits: 02/16/12
What I was talking about was dynamics.
All You say is right when talking about output.
But dynamics are all about having a non restriction or as little as possible. And here a ribbon is unbeatable... except for maybe plasma. hehe
The restriction for a membrane will be exponential and therefore the dynamics will decrease exponentially.
Single ended or double pole piece are another discussion as it is mainly a linearity and efficiency one as such. But both will suffer from restriction from the suspension.
The one who succeeded was the one who didn't know it was impossible.
Why would it be nonlinear, though? I'd expect the deflection to be pretty linear until you get near the elastic limit. Which you don't want to do because you'll stretch or break the diaphragm.
What I mean to emphasize is that suspension equals compression and thus less dynamics.
The one who succeeded was the one who didn't know it was impossible.
Is it? I always assumed it was pretty close to Hooke's Law, that is, linear. In two dimensions, of course, with more compliance laterally than vertically.
"Is it? I always assumed it was pretty close to Hooke's Law, that is, linear. In two dimensions, of course, with more compliance laterally than vertically. "Not really if You look at the function of a suspended membrane witch is not a spring.
You could though try to integrate Hooke's law into the calculation of air volume that is very non linear for a suspended membrane.You probably understand me now? :)
The one who succeeded was the one who didn't know it was impossible.
Edits: 02/16/12
Sure. The next question, then, is which distortion mechanism predominates at high excursions -- the nonlinear displacement, or the nonlinear magnetic field?
You are absolutely right, the mitigating factor is that the membrane becomes more flexible over short time periods as stretching increases temperature in the membrane and reduces resistance to elongation. It is noticeable on the Neo8s as the detail retrieval and dynamic swings after a very loud passage improve and then deteriorate slightly if no further loud passages follow. I should point out that when I use the modified Nuforce class D amps, the dynamics are better and the effect is even more pronounced but dynamics are better overall. I think this better dynamic presentation has something to do with the driver being within the feedback loop since the little amps get way hotter than the energy they are supposed to deliver would imply. Which indicates some sort of electric damping, probably of the return tension on the membranes.
The membrane material in the Neo8 is rather stiff and has little elongation and is responsible for the sheer drop of the bass response as stretch limits on the membrane are approached. It is alot more like kevlar than it is like rubber.
There is definitely energy storage and compression going on with the BGs. But it is not that great so long as you keep bass freq out of it.
The big deal is that the bigger the magnets the smaller the effect of air damping on dynamics. Remember that you have the ribbon sittin in a narrow slot and pushing air into and out of the slot with turbulent air flow and at the kinds of displacements that Valvetude reports that is going to play into damping - and the more magnet you have the better you are able to overcome the air resistance. While you can get "cheap power" to do it instead of expensive magnets, the quality of the "cheap power" in something as revealing as a ribbon is far more significant a cost than it would be on a bass driver.
Torsion is also magnetically/electrically damped, though weakly, which is why that is a limiting factor in ribbons. The torsioning is caused by greater air resistance to motion by the edges of the slot. Once an asymetry initiates the torsion it will not go away till the ribbon's vibration is much reduced in amplitude.
Re Atkinson - it is what he is implying in comparing to the Quads.
http://www.stereophile.com/content/quad-esl-63-loudspeaker-measurements
http://www.stereophile.com/content/quad-esl-989-electrostatic-loudspeaker-john-atkinson-may-2003
Also Iverson's review of the SF Guarneri stand mount speaker actually compares to the Quad ESLs.
While it's true that larger magnets increase electrical damping, air damping so predominates at the higher frequencies at which ribbons operate that I'm not sure how much of an effect electrical damping has, even with strong magnets. More on torsion and edge-gap distortion perhaps, since when that occurs the air is turbulent. But as I think about it, I'm not even sure if it has much of an effect there. Amplifier damping can be compromised by the crossover network, which would leave only self-damping from eddy currents. Since ribbon resonances are below the crossover point, the diaphragm could literally be left to twist in the wind. Another consideration is that the amplifier can only damp the ribbon as a whole. Since torsion is symmetrical, it may not be much affected by amplifier damping at all. Again, you're back to eddy currents. Which would however be stronger in a stronger B field.
The idea is that larger magnets and thicker tweeters end up with larger relative magnetic forces vs. air resistance and damping. Thus less compression.
In the torsioning case, I know from listening and from the basic physics of it that it can't be damped effectively, only prevented - as magnepan do with their little beads - which allows them to have no torsioning and no significant lengthwise resonance.
One of my tweeters is loose, while the other is nice and tight and beaded. The loose one torsions when playing signficant output below 5 khz, but performs just fine with the 10khz XO. The higher frequencies don't seem to be affected by the torsioning when playing at lower XO, but the 2-3 khz range becomes rough and screechy like the Apogees can sound.
Once it starts oscillating the damping would already have proved insufficient, so it makes sense that it wouldn't stop while still driven. But I assume that if you could achieve critical damping, you wouldn't get the torsional oscillation in the first place.
I gather the Apogee ribbons had foam supports to achieve the same thing as Magenpan's adhesive. But of course they still had the torsion problem, I assume due to the width of the ribbon.
Not any measurements done by myself but from the book "Ribbon Loudspeakers Theory and Construction" ISBN: 1-882580-24-X.
Corrugated ribbon have larger/more THD then an uncorrugated ribbon in 3kHz and 10kHz except for 1kHz where the uncorrugated ribbon has slightly higher THD.
IMD is higher for corrugated in 2&5kHz and slightly higher for an uncorrugated ribbon in 5&7kHz.
Interesting I would think.
The one who succeeded was the one who didn't know it was impossible.
I was looking for that book a few days ago, the only copy I could find was going for more than $300. But I saw the web page that had that info. I wonder whether an uncorrugated ribbon would be more susceptible to torsion. Also, whether it would allow sufficient displacement to play at high amplitudes at the lower end of its frequency range.
There are sadly no comparisons in those areas in the book.
But it mentions and shows a research of how a corrugated ribbon does behave as a uniformly moving piston, but has many modes imposed and torsional movement in, what they call 8th resonance mode, that would be 244Hz.
The one who succeeded was the one who didn't know it was impossible.
Edits: 02/17/12
Now that's really ineresting. Wish I could get hold of that book. I'm going to see if I can get it on interlibrary loan.Meanwhile, I found the paper mentioned in the caption:
Edits: 02/19/12
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: