Welcome! Need support, you got it. Or share your ideas and experiences.
Return to Planar Speaker Asylum
64.252.118.104
In Reply to: RE: Should we send a "representative" to Magnepan? posted by emsquare on October 14, 2011 at 15:36:03
Interesting questions. I did ask Wendell about the application of DSP at some point, and he said that right now they had bigger fish to fry, but he didn't rule it out as an eventuality.
How would you fit an amp in a Maggie? Kind of a fun challenge, actually.
Follow Ups:
"How would you fit an amp in a Maggie? Kind of a fun challenge, actually."
I don't think you would want to build it into the panel. That would ruin the tried and true speaker carton that they use. It would just be an accessory like a 3.6 crossover chasis. Of course the real point to it would be the line level crossover accessory that is designed for that model of speaker. Perhaps with a DSP option for room correction like Wisdom. If there was an amp module offered then it might be some savy design that matches the electrical characteristics of the drivers.
You know ... now that I see it ... maybe I didn't think that one through before I wrote it. I know I would seriously consider that option but what percentage of customers would want that nowdays to make it worth supporting? There used to be a simple passive line level crossover option for some of their earlier models. Still, it just seems to me that the logical updated version would include DSP / room correction like the Wisdom system that Wendel mentioned.
There's always the issue of price point -- the Wisdom stuff is excellent, but very expensive -- as well as the fact that many audiophiles are reluctant to put something digital in an all-analog signal path. But I'm a big believer in this, personally, and I wonder about the possibility of cooperating with a company like DEQX. The idea would be to a) make the drivers accessible without major surgery and b) get the right crossover characteristics, inverse transfer characteristic, and alignment algorithms suitable to dipole radiators into the DSP.
Yes it might be a non starter because of the price point that it would have to retail for. It would be a difficult fit for their pricing structure. Still, as the market evolves there may come a time where it is a reasonable option to consider. DSP is till evolving as has been the chipsets that support it. And perhaps more importantly has been the designers figuring out how to make it better.
This is not going to appeal to the all analog aesthetic. I think that most of us understand that you can build a spectacular sounding rig this way. And with some of the underwhelming experiences most of us has had with the advent of digital it is esay to see why people feel this way. And not without reason. That being said I am intrigued by possibilities. It would be exciting to see some blurb from Magnepan about a cooperative with DEQX, Behringer or Meridian to support the next line of 1.x, 3.x & 20.x. Loudspeakers are still a weak link for distortion. But what if you can lower that by an order of magnitude with a well executed DSP engine? Could be interesting.
One of the questions I have is whether DSP can be as effective at controlling planars as it is at dynamics. In general, I'd expect it to be good at compensating systematic errors that can be fixed with an inverse transfer characteristic, but not chaotic behavior. So it could smooth the frequency response, but could it compensate for something like the nonlinearity of a single-ended magnet assembly, or would the chaotic behavior of the diaphragm make it impossible to know where it is? This is one area in which it seems to me that dynamics have an advantage over conventional planars, if you can keep the cones from breaking up.
I toyed some years ago with schemes for local DSP with feedback, e.g., a big printed planar array speaker which would not only allow you to use feedback to control the diaphragm but would allow you to recreate the original sound field at the front of the room. Came up with an elegant way to do the feedback. But that's a huge project and I don't think it could hit a reasonable price point at the volumes of an audiophile product. A rigid planar woofer, though, e.g. carbon fiber with surround, might be more amenable.
Anyway, we'll have to see what Wendell has to say about the practicalities of adding DSP. I think you've identified a bit problem -- the reluctance of audiophiles to put anything digital in the chain. Honestly, I can understand it. It isn't so much that conversion on an analog system can't be transparent -- in my experience it can be -- as that many implementations aren't, and how do you know?
Chaotic behavior as in random flexing (resonance) of the diaphragm and frame? Sorry. Not a term I have come across. I would have nothing to say about that but inverse transfer function should work. You would know better than myself. I would imagine you would have to have a properly equiped speaker lab to pull it off. You might even be able to pull it off in the analog domain. Start with a high rez digital filter that models the trandfer function that is desired. Then input it to one side of a comparitor. Tune a circuit to minimize the difference on the other side of the comparitor till you get a difference of less than 70db or better. I bet you could do it if you find the right nonlinear bias point for a well chosen transistor or FET. The only reason that I mention that is for the devoted analog user. Just a passing thought. But it might appeal to someone who cannot stand the idea of digital processing. I have no problem with it though.
From what I gather reading some conversations on diyAudio, digital domain processing is not standing still. Apparently there are new generations of DAC's available or being developed that are far better than anything used in current audiophile DAC's. What I am trying to figure out is how to get a handle on what the specs imply. Figures like 140 db signal to noise ratios are fairly common with some of this stuff. It's not like you cannot get some benefit from this from Redbook souces. I suppose I will have to educate myself on the landscape of available Hi-Rez digital. What I'm afraid of is the cost of accumulating titles that I actually want to listen to all over again.
There's apparently a paper on chaotic behavior in planar diaphragms -- seems it leads to the generation of subharmonics. Unfortunately, I haven't read it, just seen it mentioned somewhere or other, so I don't now the magnitude of the effect. But what with reflections from the edges of the diaphragm, I do know that you're going to have traveling waves, and these are going to affect the distance of the conductors to the magnets. Also, the diaphragm dishes. Some of this can presumably be incorporated into the transfer function, but even if you could accurately model the speaker you can only use a one-dimensional parameter, voltage, to control a two-dimensional surface. That's why I suspect that, in the absence of cone breakup, a dynamic driver is a better match for DSP -- you have only to control the motion of the voice coil to achieve ideal pistonic motion.
Just speculation, of course. But I've seen some amazing measurements off self-powered speakers, Satie sent me a waterfall plot way back when that was beyond belief.
I've found that redbook can really be surprisingly good. Years ago, I had a good converter at home and tried A/Bing some LP's with and without conversion on my 1-D's. I wasn't able to hear a difference. Of course, live recordings are more demanding than LP's and redbook can apparently be succesfully ABX'd when downconverted from 192/24, but I think the difference is subtle at best. I think it got a bad rap owing to the poor quality of early implementations and bad mixes. So curiously, the reason to get new 192/24 releases typically has more to do with the fact that more of them were mastered with audiophiles in mind than with the bit rate itself. (The effect of the bit depth is easy to calculate. 24 bits is needed for music of the widest dynamic range played at original levels. That immediately rules out most everything we do at home. But 16 probably isn't quite enough, home systems can generally do about 115 dB on peaks so you need about 125 dB of dynamic range.) In any case, I think you're right that advanced converters can help even with redbook, perhaps even more than with high def recordings, because the reconstruction filters have to be better. And spot on about the difficulty of interpreting measurements. There are some good ways to test, e.g., multiple passes through a A/D-D/A cycle. Some converters handle this with aplomb, you can do it 100 times and still can't hear any degradation. Others fail on the first attempt. The question is, which ones?
Ah. That seems like what the term implies. I'm sure that I do not appreciate how much is 'wrong' in theory in comparison to a conventional dynamic driver. Much of what you try to accomplish in manufacturing the perfect cone or dome is discarded with planars. I certainly would not think that the single ended pole piece would work as well as it does. I would think that the nonlinear field would be obvious in the way that they sound. Shows how much I know.
I've read some interesting observations as to what a self powered speaker may have to offer. Perhaps Magepan's would be marvelous candidates for this kind of arrangement. But it probably would not market very well. They have a business model that has kept them in business where so many others have faded away. Let's say they had offered this in a 3.7 special edition this time. And that it had succeeded spectacularly well with glorious reviews as to how it was worth every penny of the additional cost. I'm not so sure that it would end up being that big of a seller. Although their speakers are not exactly cheap they do offer reasonble value in the high end market. We have been opining for some time as to our various wish lists for our dream version of Magnepan. I do think it would be cool if you could have a software plug in correction algorithm though. Not that I would have a way to use it.
I do not disagree with what a good redbook CD can offer. At this time all of my music is in this format and I seldom feel that I am lacking anything with it. Being the jaded gear junkie I still have nagging questions as to what it could have been. I think that most of us feel that the CD format could have been better refined before it was standardised. People that are far more knowledgeable than I name various figures for what could have been a transparent medium. To name one I think it was Bob Meridian(?) that said he felt that 56k at 18 bits would have been just about there. Is he right? I don't know. But I have become aware that signal processing has a lot to offer. We all know that you can't get resolution that isn't on the recording. But you can approximate with better accuracy than I would have guessed.
Thanks for your thoughts on the subject. I'm still trying to get a better bead on how to look at things. It dawned on me for the first time that if you need to preserve signal integrity then you won't gain much benefit from a 125db source through a buffer with 85db S/N ratio. Not that it is the end of the story but it isn't likely be all that it could.
It's just a guess, but I suspect the reason single ended magnet assemblies work is that, since sound is logarithmic, the diaphragm is rarely undergoing large excursions. That would mean that the nonlinearity of the single-ended field would be apparently mostly at very high levels and during peaks, which does tend to correlate with what I remember of the push-pull 1-D tweeters -- that they didn't start to sound congested the way the MMG's tweeters do at high levels. But I could be hearing other effects as well, particularly the fact that the MMG's tweeter shares a diaphragm with the midwoofer so is subject to intermodulation effects -- a loud bass note will move the tweeter into the nonlinear range, and you can sometimes hear the bass muddying the treble.
Agree about marketing. We audiophiles tend to be pretty set in our ways. I think it's easy to understand how that comes about -- I often find myself drawing on listening experiences that are over 30 years old.
The year the CD was introduced, I started a big argument at an AES convention when I suggested that the bit depth and sampling rate were too low. :-) I think about half the audience ended up agreeing with me, the other half said hey, you have to be practical. In fact, the story as I've heard it has more to do with corporate shennanigans on the part of Philips than with technical limitations. Admittedly, this was from somebody at Sony. But 16 bits/96 dB was clearly not enough and there was plenty of data even then to tell us that. And not only was 44.1 kHz marginal, it was a difficult conversion from the already-established 48 kHz AES standard that was already being used in the studio. Today we can do it easily, but the early standards converters didn't sound very good. Fortunately, clever folk were able to ameliorate the 16 bit limit somewhat with noise shaping.
One advantage analog buffers have is that, if they are noisy, the noise is generally of a pretty benign, random sort. Just hiss. Whereas if you listen to a dithered low-level audio signal, it's very rough sounding. And if you eliminate the dither, you lose ambiance and spatial cues. So while, ideally, analog stages wouldn't raise the noise floor of the digital signal, I don't think analog noise is as offensive. Otherwise, we wouldn't be as tolerant of tape hiss as we are.
I wasnt suggesting putting an amp on the speaker - I was thinking of a bryston magnepan active crossover good parts amp pre amp goodness - which would translate to all you need...especially if the magnepan speaker could go low enough to not need a sub
http://followthepeleton.com
That would be like Emerald physic's dipole dynamic speakers with their modified behringer and the custom settings in it.
It works for them.
But we had that early on with the early Tympani models being sold with ARC crossover and amplification.
These days the dealer would be expected to put that together. Or you would need a daft proof product that the user does not need to adjust after the dealer sets it up.
Yes it would be a Magnepan version of that. Perhaps more.
I remember the optional 1st order line level attenuators they made for their Tympani's. There is likely a reason they discontinued those. Probably for the same reason that multi-amping speakers is seldom used anymore. Not exactly plug and play and all too easy to damage something from a simple mistake. Not that I would ever do something like that ... as far as anyone knows.
So you are not telling...
DIY audio was not much of a go since 70s when boomers started having kids so today's audiophile is very finicky and when they pay a gazmillion bucks for a "value" audio setup they don't want to learn any acoustics, electronics, or even geometry. And they don't want to spend time tweaking. If they need a tweak it would have to be ham hand proof.
Besides, boomers are not finding it as easy as it used to be to take out the old solder iron and trying to solder wires and components that seem to never find focus in the graduated multifocal glasses.
OK, I will volunteer one boneheaded move that I made ... Like getting the feed to the drivers reversed. I was mortified when I realized that I was driving the ribbon with the mid-bass amp. It didn't hurt anything but there was a bit of pucker factor there for a minute.
I try to observe the double check axion on such things but didn't catch that one.
You are right, of couse, on your observation on the DIY audio culture. Personally I think it is far more interesting than just throwing money at a rig. Once you start getting some insights into the physics of audio it just keeps getting more interesting. I might not be able to build an F5 with the same distortion figures as First Watt but there is something more satisfying in the exercise. Still working on that one. And it is rather annoying that I cannot focus like I once did. I rarely appreciate being near sighted as opposed to far sighted but for populating or altering circuit boards. I find that to be highly relaxing and enjoyable. And surprising that the alterations work more often than not.
BTW, you were right about the OPA(x)134 op-amps. There were better options available.
Not that I haven't done similar errors, but I do try to be careful. First thing is to start everything from the minimum volume setting at the pre. you are likely to figure out what is going wrong before you blew anything.
That is the point, the audiophile has music listening and component listening as his hobby, he does not need to piggyback the DIY hobby and the science/engineering behind audio. Those are separate hobbies that are far more time consuming and don't provide you with more music per hour.
So what did you use instead of the x134 op amps?
"First thing is to start everything from the minimum volume setting at the pre. you are likely to figure out what is going wrong before you blew anything."Yessir. That's how I caught it. I peg the balance control to one side and slowly bring up the volume to the bass driver to verify. I was puzzled for a bit until I realized what was up. I think I will just say that I was preconditioning the ribbon for maximum joy. Ya... That's the ticket!
"That is the point, the audiophile has music listening and component listening as his hobby, he does not need to piggyback the DIY hobby and the science/engineering behind audio. Those are separate hobbies that are far more time consuming and don't provide you with more music per hour."
And this is fine. Not all things are going to appeal to any given audiophile. If it isn't bringing some joy to your life the what's the point? All too easily can I obsess about the system over what I have it for. I gather a number of ideas from comunities like this that would never have occurred to me otherwise. I hold your Neo-8 array as a highly worthy avenue to explore. It just sounds like fun. I was holding out for them to update their Radia drivers and they have introduced the RadiaPro versions. These use the Kaladex material like the Neo's but aren't quite comparable in performance. At least as far as I know.
As far as the opamps... I really like the LM4562 in the buffer stages and am still using the OPA2134's for the filter stages. I prefer this subjectively to using the LM4562 for everything. I have an interest in trying the current source version of that family but am not sure that the circuit is compatible yet. The Marchand has been the most responsive to modification so far of the electronics that I have. Still, that bring a series of 8 opamps per channel into the chain. Not exactly the elegant minimilist answer that I would like but I have to have something for now.
Edits: 10/19/11
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: