Any 21’s on the horizon with XBL^X? - Page 4 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #91 of 171 Old 07-09-2008, 09:13 AM - Thread Starter
Advanced Member
 
04FLHRCI's Avatar
 
Join Date: Dec 2005
Posts: 777
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Good Morning Dan,

Great to have you contribute to the thread; I'm really looking forward to the driver you and Haskins are working on - the unit affectionately labled "Sicko".

Kevin,

Any further developments you may share? I'm most certainly ready for my pair!

Larry
04FLHRCI is offline  
Sponsored Links
Advertisement
 
post #92 of 171 Old 07-09-2008, 09:16 AM
AVS Special Member
 
armystud0911's Avatar
 
Join Date: Jul 2006
Location: Cincinnati
Posts: 4,423
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Other than this http://www.avsforum.com/avs-vb/showp...0&postcount=66


You folks gotta understand, when we are sitting around with our thumbs up our butts waiting for a driver to launch, what better would we have to discuss on this thread? We've been over all this before, without any new news, we can either choose to let the thread take a break, or get sidetracked.
armystud0911 is offline  
post #93 of 171 Old 07-09-2008, 10:17 AM
AVS Special Member
 
michael hurd's Avatar
 
Join Date: Nov 2007
Location: 49th parallel
Posts: 2,121
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 29
Klippel testing also shows that Le is not a static figure, rather changes with both current and also coil position in the gap.

As the coil moves inward.... the Le goes up... acting like an iron core inductor... and as the coil moves out Le goes down... acting like an air core inductor.

High Le drivers are not the devil.... they simply require a lower crossover point... drivers that have wild Le variation are much "uglier" IMO.

"You take the blue pill - the story ends, you wake up in your bed and believe whatever you want to believe. You take the red pill - you stay in Wonderland and I show you how deep the rabbit-hole goes."
michael hurd is online now  
post #94 of 171 Old 07-09-2008, 10:28 AM
Member
 
Ilkka Rissanen's Avatar
 
Join Date: Jul 2006
Posts: 171
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by michael hurd View Post

As the coil moves inward.... the Le goes up... acting like an iron core inductor... and as the coil moves out Le goes down... acting like an air core inductor.

This is the "typical" behaviour of inductance (Le(x)), though there are drivers which don't follow typical behaviour.

Ilkka Rissanen
R&D engineer
Ilkka Rissanen is offline  
post #95 of 171 Old 07-09-2008, 10:29 AM
AVS Special Member
 
michael hurd's Avatar
 
Join Date: Nov 2007
Location: 49th parallel
Posts: 2,121
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 29
I know... I happen to own some...

"You take the blue pill - the story ends, you wake up in your bed and believe whatever you want to believe. You take the red pill - you stay in Wonderland and I show you how deep the rabbit-hole goes."
michael hurd is online now  
post #96 of 171 Old 07-09-2008, 10:30 AM
Member
 
Ilkka Rissanen's Avatar
 
Join Date: Jul 2006
Posts: 171
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by michael hurd View Post

Klippel testing also shows that Le is not a static figure, rather changes with both current and also coil position in the gap.

Yes, this is the issue that most people don't understand. I blame the manufacturers who (most of them) still are publishing the pretty much useless Le at 1 kHz figure.

Ilkka Rissanen
R&D engineer
Ilkka Rissanen is offline  
post #97 of 171 Old 07-09-2008, 11:38 AM
AVS Special Member
 
Ricci's Avatar
 
Join Date: Mar 2007
Location: Louisville KY
Posts: 5,101
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 219
Quote:
Originally Posted by Ilkka Rissanen View Post

Hi guys,

Some recent tests I have made with a Klippel Analyzer 2 have really opened my eyes with regard to simulations vs actual measured performance. It is really interesting to see the actual reasons for non-linearities instead of just measuring THD.

The AA Tumult and CSS SDX15 are definitely quite different in design and performance, even they both have an XBL^2 motor. The Tumult seems to be more on the lines of what XBL^2 supposed to be with regard to Bl linearity. The culprit is the lack of shorting rings which cause high Le(i) distortion (flux modulation) in mid and upper bass where the current is high. At lower frequencies the Bl and Kms non-linearities are dominant, and those ones are quite small with the Tumult.

I thought that the suspension would have been the culprit of the SDX15. It isn't. It's the Bl non-linearity instead. The XBL^2 motor doesn't work at all how it should be. Inductance based non-linearities are at lower level than with Tumult, but not much better than with regular overhang (say TC2k) design. There seems to be some strange design choices there, too.

I can shed some more light and show the measurements after I have discussed with the people involved.

Does this mean that you have your own, or at-least access to a Klippel analyzer now? I'm interested in the results of the Tumult and SDX tests. I'm willing to bet that the xmax rating of the SDX should be no where near 30mm. Judging from the 2 that I have and your tests.



Mike,
Where did you send your LMS for testing and how much was it if you don't mind me asking? Send me a PM if you don't want to discuss it publically. Maybe I should send in one of my XXX's.
Ricci is offline  
post #98 of 171 Old 07-09-2008, 12:07 PM
AVS Special Member
 
DS-21's Avatar
 
Join Date: Jun 2005
Posts: 3,493
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 162 Post(s)
Liked: 213
Quote:
Originally Posted by michael hurd View Post

High Le drivers are not the devil.... they simply require a lower crossover point... drivers that have wild Le variation are much "uglier" IMO.

Geddes has made some interesting comments along those lines, too, though his idea of a "high inductance" driver is lower than most available subwoofer drivers!

One would think that lower inductance also means lower possible variation.

But one constant I've noticed is that drivers with Faraday shielding always seem to sound cleaner than similar drivers lacking same.

Rings can lower inductance globally, but used right they always reduce inductance variation over frequency and over excursion. Do I have that right?

--
"In many cases there aren’t two sides unless one side is 'reality' and the other is 'nonsense.'" - Phil Plait
Serious Audio Blog 
Multichannel music (and video) urban loft living room system 
DS-21 is offline  
post #99 of 171 Old 07-09-2008, 01:43 PM - Thread Starter
Advanced Member
 
04FLHRCI's Avatar
 
Join Date: Dec 2005
Posts: 777
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
The renders would keep us busy for 15 minutes


Quote:
Originally Posted by armystud0911 View Post

Other than this http://www.avsforum.com/avs-vb/showp...0&postcount=66


You folks gotta understand, when we are sitting around with our thumbs up our butts waiting for a driver to launch, what better would we have to discuss on this thread? We've been over all this before, without any new news, we can either choose to let the thread take a break, or get sidetracked.

04FLHRCI is offline  
post #100 of 171 Old 07-09-2008, 01:51 PM
Member
 
Ilkka Rissanen's Avatar
 
Join Date: Jul 2006
Posts: 171
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Ricci View Post

Does this mean that you have your own, or at-least access to a Klippel analyzer now? I'm interested in the results of the Tumult and SDX tests. I'm willing to bet that the xmax rating of the SDX should be no where near 30mm. Judging from the 2 that I have and your tests.

Yes, I have access to a Klippel analyzer.

And you are being correct with regard to the 30mm claim. I have sent the results to Bob Reimer. After he has reviewed them, I might post them here (him allowing).

Unfortunately I don't have a Tumult (either version) in my hands at the moment.

Ilkka Rissanen
R&D engineer
Ilkka Rissanen is offline  
post #101 of 171 Old 07-09-2008, 02:30 PM
AVS Special Member
 
Ricci's Avatar
 
Join Date: Mar 2007
Location: Louisville KY
Posts: 5,101
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 219
Quote:
Originally Posted by Ilkka Rissanen View Post

Yes, I have access to a Klippel analyzer.

And you are being correct with regard to the 30mm claim. I have sent the results to Bob Reimer. After he has reviewed them, I might post them here (him allowing).

Unfortunately I don't have a Tumult (either version) in my hands at the moment.

Well that sounds like good news (klippel), plus some bad (SDX). I think I hear a can opening in the distance...

Were both of the SDX's used in your tests put through the Klippel or one? My SDX's seem to start loseing control at more like 20mm than 30mm (this using the trusty, infallible eye gauge).
Ricci is offline  
post #102 of 171 Old 07-09-2008, 02:34 PM
Member
 
Ilkka Rissanen's Avatar
 
Join Date: Jul 2006
Posts: 171
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Ricci View Post

Well that sounds like good news (klippel), plus some bad (SDX). I think I hear a can opening in the distance...

You too?
Quote:


Were both of the SDX's used in your tests put through the Klippel or one?

I tested both of them to be sure it wasn't just a defective unit. They both measured almost identically.

Ilkka Rissanen
R&D engineer
Ilkka Rissanen is offline  
post #103 of 171 Old 07-09-2008, 02:35 PM
AVS Special Member
 
ssabripo's Avatar
 
Join Date: Feb 2004
Location: Plantation, Floriduh
Posts: 6,008
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 13
so were are all these folks who jumped on my ass last year when I mentioned that the SDX didn't particularly strike me as a good XBL^2 implementation?

ssabripo is offline  
post #104 of 171 Old 07-09-2008, 02:36 PM
AVS Special Member
 
Ricci's Avatar
 
Join Date: Mar 2007
Location: Louisville KY
Posts: 5,101
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 16 Post(s)
Liked: 219
Oh boy...that's what I'd hoped. Guess i'll get my fishing gear ready for all of these worms...
Ricci is offline  
post #105 of 171 Old 07-09-2008, 02:38 PM
Member
 
Ilkka Rissanen's Avatar
 
Join Date: Jul 2006
Posts: 171
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by ssabripo View Post

so were are all these folks who jumped on my ass last year when I mentioned that the SDX didn't particularly strike me as a good XBL^2 implementation?


Oh come on Sherv! That was just an educated guess on your behalf. You haven't put any of these through Klippel, have you?

Ilkka Rissanen
R&D engineer
Ilkka Rissanen is offline  
post #106 of 171 Old 07-09-2008, 02:41 PM
AVS Special Member
 
ssabripo's Avatar
 
Join Date: Feb 2004
Location: Plantation, Floriduh
Posts: 6,008
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 13
Quote:
Originally Posted by Ilkka Rissanen View Post

Oh come on Sherv! That was just an educated guess from your behalf. You haven't put any of these through Klippel, have you?

correct...just a educated guess given the initial measurements and listening replies I heard from the first few sealed and ported implementations. I've had avalanche and tumult drivers thru my hands, and they never had the issues/problems mentioned back then.

granted, different drivers, and different implementations....as well as different times. But still, something just wasn't right and I said it back then.
ssabripo is offline  
post #107 of 171 Old 07-09-2008, 09:54 PM
Member
 
DanWiggins's Avatar
 
Join Date: Aug 2001
Location: Lynnwood, WA USA
Posts: 50
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by DS-21 View Post

Geddes has made some interesting comments along those lines, too, though his idea of a "high inductance" driver is lower than most available subwoofer drivers!

One would think that lower inductance also means lower possible variation.

But one constant I've noticed is that drivers with Faraday shielding always seem to sound cleaner than similar drivers lacking same.

Rings can lower inductance globally, but used right they always reduce inductance variation over frequency and over excursion. Do I have that right?

Not always. Rings can help until you either saturate the ring or you get too low in frequency. The effectiveness of a ring changes with frequency and power applied. Inductance varies with frequency, position, and power applied. Thankfully THD+N from inductance is relatively "benign" in an audible sense.

I've seen many implementations where the use of rings actually hurt transient behavior of the driver, and increased dynamic (with music or bursts) THD+N; in a static THD+N test it measures beautifully, but doing a THD+N where the test signal is music you get terrible results.

In the more extreme cases, you can actually force dynamic offset in the driver because of an inductive peak that is away from the center/rest position, leading to acoustic rectification because of an inductive problem. In those cases, you would be better off with no rings. Inductance rings and sleeves are definitely an area where the old maxim "a little does a little good, a lot must do a lot of good" does not hold true.

You can do some very trick, low inductance designs if needed; the key is to always consider what is required for a given driver's application. For example here's the impedance plot of an 8" woofer, 16mm one way linear BL (based upon Klippel measurements), 65mm diameter voice coil, with 500W RMS power handling:



Yes, the resonance is at 40 Hz with a peak of 12 Ohms, the minimum (DCR) is 4.5 Ohms, and the peak impedance is 7 Ohms at 20 kHz. And this impedance curve is stable for bursts up to 90V (2 kW peaks) over the entire bandwidth. It's about on par with many tweeters, and even gives some ribbons and planars a run for their money.
DanWiggins is offline  
post #108 of 171 Old 07-09-2008, 11:49 PM
AVS Special Member
 
armystud0911's Avatar
 
Join Date: Jul 2006
Location: Cincinnati
Posts: 4,423
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Am I seeing things here? Is this like the perfect bookshelf driver, all you would need is it and a super tweeter, you could probably run it full range if you wanted to. Where do I buy said mystery driver? You can't post something this too good to be true and not expect a barage of fallow ups.
armystud0911 is offline  
post #109 of 171 Old 07-10-2008, 06:22 AM
AVS Special Member
 
Mark Seaton's Avatar
 
Join Date: Jul 2000
Location: Chicago, IL, USA
Posts: 6,063
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 285
Quote:
Originally Posted by armystud0911 View Post

Am I seeing things here? Is this like the perfect bookshelf driver, all you would need is it and a super tweeter, you could probably run it full range if you wanted to. Where do I buy said mystery driver? You can't post something this too good to be true and not expect a barage of fallow ups.

I wouldn't call it too good to be true, but just a custom design with specific priorities. I expect that is a custom design for an OEM who will offer it in a finished product, but have no direct knowledge on this particular example. It is still an 8" woofer with the directivity that goes along with that size, and the complications of mating to a tweeter at high frequencies.

Of course as Dan mentioned, the question is if this sort of uber-low inductance is needed or even wanted in your use. With many cones, this motor would have a rising frequency response at higher frequencies which could be a non-issue, or could be a hurdle in a specific use. I would bet the above example matches very well to the intended use.

Mark Seaton
Seaton Sound, Inc.
"Make no little plans; they have no magic to stir men's blood..." Daniel H. Burnham
Mark Seaton is offline  
post #110 of 171 Old 07-10-2008, 07:25 AM
Member
 
DanWiggins's Avatar
 
Join Date: Aug 2001
Location: Lynnwood, WA USA
Posts: 50
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Armystud,

Mark pretty well sums it up... There are issues with the frequency response that arise directly because of the non-inductive nature of this driver (Le is 27 uH):



As you can see, the frequency response - while extended to 12 kHz (without a whizzer) - does rise 12 dB from 500 Hz to 5 kHz.

Additionally, the sensitivity is 86 dB in the low end, which kills it as an option for most FR guys, since they are overwhelmingly the "sound of one Watt clapping" type of people (seriously, full-range fans willing to use more than 1-2W of power are a slim segment of the full-range market).

The point here is that XBL doesn't force you to weird inductance, or high inductance. You can do low inductance, and you can do ultra-low. And that each driver tends to be designed for specific design goals which are not often in the interest of the DIY hobbyist.

Realize that most of the offerings from the larger companies are cast-offs from OEM designs. An OEM commissions a factory to design and build a certain product. They do so, and build it for a few few years until the OEM changes the model.

The factory typically takes that design, changes it cosmetically (different basket, different coating on the cone) and then places it into their "stock offerings" catalog. So what you end up with is a lot of other people's solutions, not really your own. Which is also why OEMs tend to just get new designs done; if you're going to build 500+ drivers, it makes economic sense to just do your own from scratch.

Having experience in the DIY and OEM world, I can tell you that the former is a lot harder to deal with because of the demand for the best and the greatest without realizing that it comes with problems which are often diametrically opposed to each other, and the latter is hard because of pricing. Pricing is easier to negotiate; pet theories or demands or ignorance - especially in terms of audio and sonics - is a lot harder to overcome.

And of course pricing is seriously difficult with DIYers, too... You end up with OEM, big-volume and annual commitment quotes being reported/represented as individual, retail prices (like happened in this thread), and then the DIYer getting pissed because the actual retail price is just outrageous and gouging. Never mind that the DIYer will buy one, two, or maybe 4 units. And the distributor and small OEM - who needs to make a profit as well - will buy a few hundred each month. Pretty simple to see who gets the price love there!

I'm sure if the market manifests itself some DIY supplier will take up the challenge, and would be willing to make a driver like this. It won't be cheap, and there will be - by necessity - a significant minimum buy of quantity (good luck getting less than a few hundred speakers made at a time). And there's tooling costs involved, too... All told, rolling out a new driver can run a DIY supplier well in excess of $40,000 per model. That's a lot of sales of a $50 driver to just break even!

And understand that a lot of the problems with the DIY market getting castoffs and being treated as second-tier is because the DIY community demands it. They will take a great driver and completely destroy it because of a perceived fault; even if that fault was incorrectly measured, observed, or flat out wrong.

Sure, driver A may have slightly more THD+N than driver B, but driver A has lower power compression, lower THD+N in dynamic situations, better dynamic stability, easier use in a variety of boxes, more stable product supply. But in the DIYers mind, that difference between 1.5% THD+N for driver B and 2% THD+N for driver A- a totally inaudible difference - is simply case-closed! Driver A must be better because it just is... Missing the forest because of staring at a single tree.

Bottom line: there are tradeoffs everywhere, and usually demanding the best of everything means you end up with the worst amalgamation you've ever seen! Keep in mind that John Dunlavy did some amazing speakers with just $10 tweeters, $12 woofers, and iron core inductors...

It's the execution of the whole, not the individual parts that matter! That holds true for drivers, too... Great technologies (like I think XBL is) are good blocks, but still need to be tempered by the overall performance and goals of the execution and the budget and production requirements.
DanWiggins is offline  
post #111 of 171 Old 07-10-2008, 08:54 AM
AVS Special Member
 
TheEAR's Avatar
 
Join Date: Nov 2006
Location: Montreal,CANADA
Posts: 3,464
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Talking about John Dunlavy,simply amazing speakers.Most natural sounding speakers I have ever heard, they are big ,boxy...with ordinary drivers(at least no exotic overpriced units).

He extracted what sounds like the very best of them and matched them perfectly.

Ask yourself mortal , do you have as much displacement as me ? The answer is no unless you have a Windmere fan sub.
TheEAR is offline  
post #112 of 171 Old 07-10-2008, 10:48 AM
AVS Special Member
 
Kevin Haskins's Avatar
 
Join Date: Feb 2005
Posts: 1,008
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
The rising FR isn't an issue. It is easy to deal with in the crossover. If you have problems dealing with that you probably shouldn't be designing speakers anyway.


In terms of the DIY market. It is driven by some crazy dynamics. It has a lot of problems for larger companies. It is small, time intensive and quirky. What is hot one month isn't the next. What drives demand often has more to do with what the latest "guru" has to say on the forums as it does the quality of the engineering. You can also have something blow-up on you quickly if the right people say the wrong things. Overall it is a high risk low reward part of the market to deal with.

Hey.... why the hell am I doing it anyway??!!

Kevin Haskins
Exodus Audio
Kevin Haskins is offline  
post #113 of 171 Old 07-10-2008, 07:25 PM
AVS Special Member
 
Join Date: Mar 2003
Posts: 4,509
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 10 Post(s)
Liked: 609
Quote:
Originally Posted by DanWiggins View Post

Never mind that the DIYer will buy one, two, or maybe 4 units.

...or maybe 6, or 8, or 12, or 15, or...

Hi Dan,

I understand the concept of a specific driver for a specific use, but this is a DIY subwoofer forum. The parameters are fairly well defined, IMO.

Companies like Parts Express have done very well supplying drivers and accessories to DIYers, so faulting the market yields little sympathy from me.

The fact is that the smaller, higher-end suppliers have been less than stable over the years. There have been many good drivers that would still be in use if they were still available, Tumult being at the top of the list in my book.

Speaking of which, I wanted to ask you: Would there have been a different result if the Tumult that Ilkka tested was wired in series to a single 4 ohm load vs dual 2 ohm? For that matter, what would the difference be, if any, in wiring 2 Tumults series/series to a single 8 ohm load?

The MKI requires a lower crossover because it rolls off quite a bit from 50-100Hz. The MKII, seen below in a close-mic graph with L/T, 12dB boost, .7 Q, allowed me to raise the crossover to 100Hz and a slew of them allows me to enjoy the best TR of any system I know of.

I have asked Kevin, regarding the Maelstrom and would like you to comment about the pole vent diameter difference between MKI and MKII, as well as the 1" difference in motor height, if you get a minute.

Quote:


5. THD+N testing should be taken with a grain of salt; make that a block of salt. Not just in the difficulty of doing it properly (the noise portion is a serious issue), but in terms of the impact of audibility. In fact, a driver with 10% THD+N can often sound a lot cleaner AND better than a driver with 1% THD+N. It's not the level of THD+N that is so important as it is the ratio of the harmonics and the way those ratios scale with power (see Geddes et al).

I agree and always have. Hopefully, it gets more mileage coming from you.

Bosso
LL
bossobass is offline  
post #114 of 171 Old 07-10-2008, 07:49 PM
AVS Special Member
 
KyleLee's Avatar
 
Join Date: Feb 2007
Location: California
Posts: 1,129
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Liked: 12
The biggest complaint about linear subwoofer designs, or linear driver design for that matter is their lack of sensitivity.

i propose for that exact reason, we don't see them but in perhaps less than 1% of drivers designs! i mean linear designs have been progressing for how long now? 30+ years? I have no doubt that XBLL and LMS are the two leading designs right but they still are hurt for sensitivity without adding cost to make up the BL losses.

XBLL and LMS are both below ~25-20% BL losses relative to an overhung coil of the same motor parameters.... not enough to make the market switch over all the way, XBLL seems to be gaining some steam as a licensable design as we are migrate away from tiny magnets into more robust platforms, but its hard to convince someone they have to give up 2dB of their overhung motor to gain 20% more linear xmax. Its even harder to convince a big mfr they have to spend more on the motor to make up BL losses... cheap parts makes you money when you're dealing with volume... no way around it.


Just trying to start some debates...


thoughts?

data-bass.com
@databass_audio
KyleLee is offline  
post #115 of 171 Old 07-10-2008, 08:35 PM
Member
 
DanWiggins's Avatar
 
Join Date: Aug 2001
Location: Lynnwood, WA USA
Posts: 50
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Bosso,

The Le changes with series/parallel are exactly complementary to changes in Re, so you don't see a difference.

As far as the pole vent diameter, it's big enough to not chuff under normal operation, and I think you'll find it quiet in any operation. We optimize pole vent diameters with fluid dynamics to keep noise at least 40 dB below the acoustic output of the driver.

Kyle,

Consider dynamic BL. At rest, you may have less BL (not always; often you can get higher BL, if you balance the motor properly) with an XBL motor. But look at the average BL over stroke. Remember, the effective BL your driver is the average over the entire stroke. So while I may have a dB or two less at small excursions, when the excursion increases my efficiency stays the same (integrated BL over stroke), and the overhung starts to drop.

A prosound monitor to be released at the end of this month uses an XBL motor. At first, they were confused because we did have 1.5 dB SPL less than the existing Peerless woofer they used. But then we measured bursts and transient output. Factoring out the changes in Re (which we measured), as gain increased the XBL motor properly scaled. The Peerless unit did not, to the point where when we were at what was considered medium-loud, the XBL driver had MORE output.

Needless to say, subjective reviews by top-tier recording engineers showed a very audible difference in dynamics and compression...

This is a case where many companies over-emphasize the static cases and measurements when in fact it's the dynamic performance that is most important.
DanWiggins is offline  
post #116 of 171 Old 07-10-2008, 09:53 PM
AVS Special Member
 
KyleLee's Avatar
 
Join Date: Feb 2007
Location: California
Posts: 1,129
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Liked: 12
Dan, that’s not what my results showed.

the work: integral F*dx, or the sum of BLI over dx did not show XBLL or LMS leading overhung for a fixed 5watts.(all 3 coils had a different resistance obviously, voltage would not work for this, we need a fixed power to get the proper current) And in fact, in my cases, it would be impossible because I defined the gaps.. the overhung, lms and xbll coil all leave the gap at the same time…25mm.

Here is why:

i used a 60mm lms coil, a 60mm overhung coil and a 30mm xbll coil with two gaps each 30 mm's apart (edit: 30mm center to center, 20mm between gaps). Lms and overhung use the same motor (10mm gap). The xbll motor was designed to be the same as that motor such that there were exactly 2 (10mm gaps) opposed to one. The gap’s were the same width and the motors used the same t-yoke and magnets. (the xbll t-yoke was taller of course to accommodate the higher gapplate because I used the same number of magnets for formality. (yes you can make an lms coil fit into an xbll gap, its all how you design the coil and the gauge) in my case I used the same wire gauge for all 3 coils. Ignore rocking inertia differences and thermal expansion differences for a this case, its simply a small signal efficiency experiment over the entire displacement.

I will argue you can on longer compare things on a relative bases if you start stretching out the XBLL gaps… you make the coil bigger and you add more resistance and mass so this will start to lower the sensitivity immediately so you can’t just get more xmax for “free” and I didn’t want to show any tradeoffs yet. And of course if you add a third gap, you just divide the flux by 3 and its even less. My point was, I wanted to see what kind of sensitivity I would get with “formal” geometrical definitions for xmax –“coils start to leave the gap is that standard.”

these results showed that xbll and lms had nearly the same xmax (aprox. 20% higher than the overhung coil) As we would expect because they are both linear and we defined the dimensions such that the coils ran out of steam at the same time (25mm) XBLL and LMS were basically tied in that respect and that’s good, it proved I at least achieved one goal.

I think that’s a fair comparison because those relationships define xbll to have the same mathematical xmax as the overhung or lms coil.....necessairly when the last coil winding starts to leave the gap, or enters the gap... either definition applies here

But the sensitivity losses are around 1.5 to 2dB using this standard. No other formal standard exists so until someone comes up with a better one I will argue for this standard which I do believe is why we don’t see as many linear type speakers on the market “yet.”

– efficiency.

I will also argue in favor of your advance.. "dynamic" measures are key!

data-bass.com
@databass_audio
KyleLee is offline  
post #117 of 171 Old 07-10-2008, 11:28 PM
Member
 
DanWiggins's Avatar
 
Join Date: Aug 2001
Location: Lynnwood, WA USA
Posts: 50
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Kyle,

If your XBL gaps have a 30mm rebate between them, and the voice coil is 30mm long, then the motor was designed wrong. The voice coil needs to hang equally in each gap, when at rest. You should, in fact, cover ~55% of each gap at rest. So the result is you integrate at least 55% of one gap, and a good chunk of the fringe field.

Note too that optimization of the depth of the rebate is critical, as you can use that to not only create the dual gaps that the voice coil commutates through, but also to channel more of the total flux not only into the gaps but into the rebate (rather than the external fringe field).

Also note that because you have a shorter voice coil, you can run tighter gaps. For a given angular deflection, you have less radial displacement at the end of the voice coil, meaning you can run a much tighter gap. That of course helps immensely.

Lastly, when using inductance sleeves and the like, the best place to put them is right in the gap. With XBL, you can put it in the rebate and not widen the magnetic gaps at all; that is not possible with other topologies.

Additionally, the geometric definition is invalid; we all know it, that actual measurements, or barring that FEA simulations, are much more accurate. Leaving the gap does not apply, especially for overhungs which are much longer than the gap. The reason is that as you move you still cover the gap, but with short gaps and long voice coils the total fringe field can be 30-40% of the total flux in the system, and loss of coverage of that area will lead to a significant loss in BL prior to the voice coil leaving the gap.

I have taken actual long gap overhung drivers, added XBL, and kept - or in some cases - increased the total efficiency AND the BL - AND the linear stroke - of the driver. The key is that you can shift, often, from a 2 layer to a 4 layer voice coil. Because my voice coil is shorter, I can run tighter tolerances, meaning I can thicken the voice coil. We effectively take windings at the end of the overhung - out in the far fringe field where it is weak - and move them to the gaps and the inner rebate which both contain considerably more flux.

In fact, Nokia did a study that showed the most efficient driver was the one that was evenhung - VC length equal to gap height. This of course makes sense as the driver, at rest and small displacements, essentially integrates all the high-flux region of the gap and doesn't waste a single turn on the lower intensity fringe. XBL - when properly scaled into a driver - approximates a hybrid of evenhung and overhung, from an efficiency standpoint.

Try equalizing the DCR. Use smaller gauge wire for the XBL voice coil, or since your voice coil is half the length, double the layers; you'll get a lot more turns in the gap, and you'll find BL increases.

Fundamentally flux usage efficiency is a measure of getting as much of your copper into the gap as possible. The topology that maximizes that will win the flux usage - and flux stability - race.
DanWiggins is offline  
post #118 of 171 Old 07-11-2008, 12:28 AM
AVS Special Member
 
KyleLee's Avatar
 
Join Date: Feb 2007
Location: California
Posts: 1,129
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 2 Post(s)
Liked: 12
Quote:
Originally Posted by DanWiggins View Post

Kyle,

If your XBL gaps have a 30mm rebate between them, and the voice coil is 30mm long, then the motor was designed wrong. The voice coil needs to hang equally in each gap, when at rest. You should, in fact, cover ~55% of each gap at rest. So the result is you integrate at least 55% of one gap, and a good chunk of the fringe field.

Note too that optimization of the depth of the rebate is critical, as you can use that to not only create the dual gaps that the voice coil commutates through, but also to channel more of the total flux not only into the gaps but into the rebate (rather than the external fringe field).

Also note that because you have a shorter voice coil, you can run tighter gaps. For a given angular deflection, you have less radial displacement at the end of the voice coil, meaning you can run a much tighter gap. That of course helps immensely.

Lastly, when using inductance sleeves and the like, the best place to put them is right in the gap. With XBL, you can put it in the rebate and not widen the magnetic gaps at all; that is not possible with other topologies.

Additionally, the geometric definition is invalid; we all know it, that actual measurements, or barring that FEA simulations, are much more accurate. Leaving the gap does not apply, especially for overhungs which are much longer than the gap. The reason is that as you move you still cover the gap, but with short gaps and long voice coils the total fringe field can be 30-40% of the total flux in the system, and loss of coverage of that area will lead to a significant loss in BL prior to the voice coil leaving the gap.

I have taken actual long gap overhung drivers, added XBL, and kept - or in some cases - increased the total efficiency AND the BL - AND the linear stroke - of the driver. The key is that you can shift, often, from a 2 layer to a 4 layer voice coil. Because my voice coil is shorter, I can run tighter tolerances, meaning I can thicken the voice coil. We effectively take windings at the end of the overhung - out in the far fringe field where it is weak - and move them to the gaps and the inner rebate which both contain considerably more flux.

In fact, Nokia did a study that showed the most efficient driver was the one that was evenhung - VC length equal to gap height. This of course makes sense as the driver, at rest and small displacements, essentially integrates all the high-flux region of the gap and doesn't waste a single turn on the lower intensity fringe. XBL - when properly scaled into a driver - approximates a hybrid of evenhung and overhung, from an efficiency standpoint.

Try equalizing the DCR. Use smaller gauge wire for the XBL voice coil, or since your voice coil is half the length, double the layers; you'll get a lot more turns in the gap, and you'll find BL increases.

Fundamentally flux usage efficiency is a measure of getting as much of your copper into the gap as possible. The topology that maximizes that will win the flux usage - and flux stability - race.



Sorry for the confusion, I implied 30 between each gap (from center to center) <- left out that part meaning 20mm space between the top of the bottom gap and the bottom of the top gap. My FEA results showed i didn’t quite need to cover more than 50% of each gap because I guess as you said, my notched out section gave me enough BL in the void to prevent the double hump with just ~50% gap coverage. I also undercut the t-yoke and overcut it above the gaps and then extended the pole for symmetric results, this probably helped flatten the curve for me, while also lowering the inductance without using shorting rings. I'm not doing any analysis about inductance at this point so im not using them (yet), just BL. (one step at a time) I used a shallow notch on the pole to prevent saturation and boost the b in the middle just a touch to flatten the curve.

I'm trying to prevent talking bout gap tolerances because then we just cant standardize a method of comparisons. It's easy to compare lms to overhung, but then again, if we arbitrarily say we can make the gap tighter because of mass (really mass is the key, or distribution of mass rather a la inertia), then i can say we can make the LMS gap tighter than an overhung because its about ~70% the mass (missing coil volume via lms design) - i don’t want to throw out those nit picky ideas... one wire gauge, fix # of layers, small signal input… sure this is silly, but we gotta start somewhere. Then we can scale things up.

But xbl^2 needs some agreed upon circumstances if we are to compare things. If we are indeed coupling with only 50% (or 55% as you suggest) of the flux lines it cant possibly have the same BL - thats my argument. Of course, in practice XBL^2 motors actually have more flux over the entire gap +- any distance, in other words, because of the extra gap, lets say the overhung had 10 flux lines... the xbl^2 might push to 11 just because there is more steel and the magnetic path is "easier" to get thu... it’s a minor difference but again, i went thu all this analysis to formally determine how much losses we're giving up for linear BL.

gap tolerances depend on a lot of things i mean we could use taller baskets (cone surround dist.) for better alignment but we can argue that for any design because they are mutually exclusive, but at the same time, i can argue, if indeed xbl^2 under the same gap is not as efficient as an overhung design, it would then need more power to reach the same SPL and i would probably want to factor in more room on the ID of the gap for thermal expansion... even if the sensitivity was the same, the xbl^2 coil has less mass and probably will expand more at a given amount of power as a result of that difference.
seriously nit picky argument, but that’s exactly what I’m trying to avoid. but certainly in real life you can indeed make a tighter gap for a short vc... how much so? I have no clue... I guess we would then need to start defining moment of inertia for the voice coil and then come up with some function... i'm really not interested in that kind of detail. I guess the bottom line is, its not quite so simple. Believe me, I’m arguing FOR linear drivers, I would like to see a lot more of them in play. I applaud you for getting the xbl^2 platform into more and more designs, and I hope more engineers look into linear driver design, theory and solutions.

data-bass.com
@databass_audio
KyleLee is offline  
post #119 of 171 Old 07-11-2008, 08:33 AM
Member
 
DanWiggins's Avatar
 
Join Date: Aug 2001
Location: Lynnwood, WA USA
Posts: 50
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Kyle,

I really think you need to consider gap widths and overall geometries; otherwise you are ignoring some of the advantages you get with XBL. You should try optimizing for the same impedance, and ideally the same number of turns (to equalize inductance and mass as well). Remember, small changes in gap width yield HUGE gains. Flux falls off as the cube with distance, so a change of 14% in width will increase the flux in the gap by 150% (to saturation).

And while we may only integrate 50% of the peak flux in the gap, we put more of our wire in that flux. With an overhung - 10mm gap and 60mm coil - only 1/6th of your turns are in that gap. So you have 100% of the gap times 16% of the turns, meaning 16% efficiency in terms of wire and flux usage.

With XBL, you integrate a consistent 50% of the flux, but if you equalize for the turns, you'll find that you also put anywhere from 30% to 50% of the turns in that flux. So you can get 15% to 25%, potentially more output.

Remember, flux is half the equation; the voice coil is the other half. You want to put as many turns as possible consistently in the highest flux regions. This maximizes the coupling of flux to coil. It's not the strength of the flux, but the integration of flux over the coil that matters, which means the way the coil integrates that flux - the turns.

So if you want to compare the motors, you need to equalize the number of turns if possible, as well as the DC resistance, since that will also affect the number of turns when the voice coil is shorter. Otherwise it would be like comparing a 300 HP V8 to a 100 HP I4 engine, but giving the V8 a 1:1 rear end and the I4 a realistic 3.5:1 rear end. In an off-the-line launch, the I4 will dominate.

But get the differential ratios closer, and you'll see that drag-race seriously change...
DanWiggins is offline  
post #120 of 171 Old 07-11-2008, 09:52 AM
AVS Special Member
 
Mark Seaton's Avatar
 
Join Date: Jul 2000
Location: Chicago, IL, USA
Posts: 6,063
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 285
Hi Kyle,

The problem is simple. With the different implementations, you can only make clear comparisons once you assume or lock in many factors. Each user/customer of a given driver will have different priorities. For some, a 10-30% price increase in a driver is a small part of the total package, and performance is of much more importance. In other cases every penny counts, and the application plays a huge part in what level of performance is needed.

I would drop the argument of efficiency, and rather focus on dynamic/high level capabilities, as well as design flexibility. I view motor designs like XBL^2, LMS and other distributed gap/coil techniques as higher order magnetic systems. You can think of this akin to higher order subwoofer/acoustic alignments, as well as electrical filters.

In most basic terms, you have a greater number of defining variables, enabling a wider range of possible shapes for the static BL curve and associated trade offs in other aspects of operation. You have more flexibility in the optimization of the design, as well as more opportunity to create problems. Good design practices can make them very effective, just as I have been designing some very well behaved bandpass subwoofer systems. It's a matter of understanding the many contributing variables and especially those that change in operation. If you can acount for their range of change in the design, you can know and plan for what those variations will be on the system.

In short, these different motor topologies allow us to more easily, directly, or less expensively create drivers with unique combinations of parameters that can be used in different designs. Working with variations of the old ServoDrive mechanisms we were able to achieve even wilder combinations that weren't practical with conventional designs where huge moving masses and motor strength allowed experimenation with Q & Fb combinations normally only seen in IB systems and all the possibilities that affords.

In every case, execution still reigns supreme. It was asked earlier how the LMS-5400 got around many of the suggested limitations. The answer is simple. Lots of metal, lots of magnet, and lots of copper. All of those things cost money and add lots of weight. I recall the earlier/smaller LMS driver being less impressive, with low motor strength and high Le. Similarly extravagent efforts can be made with many different motor designs including XBL^2, overhung & underhung within some limits. The result generally costs more, and often steps well beyond readily available parts, meaning a lot more up front investment.

Mark Seaton
Seaton Sound, Inc.
"Make no little plans; they have no magic to stir men's blood..." Daniel H. Burnham
Mark Seaton is offline  
Reply DIY Speakers and Subs

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off