Testing 1, 2, 3, testing...

[Jed]

Looking at the tests of Nextel W15 in HH and KT magazines, I can only ask myself what they were thinking. The Nextels are probably marketings brainchild in order to have something to compete with Scan Speak.

Please be a bit more specific about what the tests revealed. I'd be interested to know how they compare to the 882. I'm using the W15 nextel from 280-3K. Pretty nice sounding but I think Accuton is going to be my next step. I have a pair of C44s, and maybe next I'll try the C79. Also, one must listen to a driver before saying its just a marketing decision because the nextels do sound nice- ultimate in resolution?. . . that is debatable and something worth investigating. I plan on doing it in the near future.

Jed

[capslock]

Well, HobbyHifi hasn't tested the new Peerless drivers yet, so I can't compare.

W15LY001
CSD: significant ringing (1 ms) at 900, 1500 and 6000 Hz, broadband decay needs about 0.2 to 0.4 ms more than a good metal cone such as W15 or L12 or, for that matter, RS125.

impedance: blurps at 900 Hz and 2.5 kHz

harmonics at 90 dB / 1 m:
2nd: broad peak of almost 3% centered at 900 Hz, almost 1% at 3.7 kHz, other than that oscillates between 0.1 and 0.3% above 100 Hz

3rd: oscillates around 0.1% above 150 Hz with the exception of 0.25% at 1 kHz

5th: below 0.02% above 100 Hz


W18NX001

CSD: similar 0.3 ms - ish broadband decay, ringing at 800 Hz is more pronounced, other than that, CSD does not look as messy as for its smaller sibling

impedance: small blips at 800 and 3 kHz

distortion:
2nd: below 0.3% above 100 Hz with the exception of nearly 1% centered around 800 Hz

3rd / 5th look fine

[capslock]

Klang & Ton 1/06 had both the W15LY and 830882:

Impedance:
882 stays below 15 Ohm out to 20 kHz whereas W15 shows sharp increase above 4 kHz, ending it around 22 Ohms at 20 kHz, blips at 950 and 3 kHz wheres 882 looks clean

CSD: even with the smooting that KT uses, the W15 looks a lot messier above 2 kHz

distortion at 95 dB:
2nd: W15 has about 45 dB average separation above 120 Hz but only 30 dB (interestingly, this peak is almost lost in the noise at 85 dB);
882 hat 40 dB 150 to 250 Hz, 50 to 55 dB out to 3.5 kHz, 45 above

3rd: W15 has only 45 dB 1 to 4 kHz, 55 to 60 dB elsewhere above 200 Hz;
882 has 55 to 60 above 250 Hz, maybe 51 around 1.5 kHz

5th: very similar at 60 to 65 dB for both

[BobEllis]

Jon, Have you had a chance to run your Focal tweeters through the wringer? Curious to see if my affection for them is justified.

Bob

[cotdt]

other tests also confirm slight energy storage at 800Hz for the W18NX, but it is not bad at all. in general it measures well and sounds great subjectively too.

[AJINFLA]

I do not think it's all marketing, some people choose their units based on what they like to hear.

I would consider that to be marketing. One doesn't have to go very far on a internet speaker forum to hear someone blathering on about paper cones being the best, etc (strangely, they never seem to mention why it is that if paper is so great a material, why are they not using paper tweeters also? But I digress). SEAS makes these Nextel drivers for them. That is all marketing IMHO.
They also know that hard cones (metal/ceramic,etc.) are more expensive to use commercially (more xo parts) and more difficult to work with - easier for the amateur designer to screw up sound wise (usually the same folks blathering about preferring paper after "trying" both :roll: ). Always better to blame the driver for bad sound, poor imaging, etc than one's own incompetence :??.
Different strokes (or coloration) for different folks.

cheers,

AJ

p.s. I like the Accuton (hard material,suprise,suprise) drivers. I just wish some large manufacturer like HiVi made them for half the price .

[noah katz]

capslock,

"A conductive former moving in the static B field of the motor will experience viscous (i.e. proportional to velocity) damping due to the formation of eddy currents. For the same former and voice coil, the damping will be lower if the motor is weaker."

Ah, I see; thanks for the explanation.

"I've tried dozens of different topologies with the OW1 and they just never could project female voices like the C12 can."

Curious if you've ruled out the freq response as the primary contributor to this.

Thanks

[JonMarsh]

I would consider that to be marketing. One doesn't have to go very far on a internet speaker forum to hear someone blathering on about paper cones being the best, etc (strangely, they never seem to mention why it is that if paper is so great a material, why are they not using paper tweeters also? But I digress). SEAS makes these Nextel drivers for them. That is all marketing IMHO.
They also know that hard cones (metal/ceramic,etc.) are more expensive to use commercially (more xo parts) and more difficult to work with - easier for the amateur designer to screw up sound wise (usually the same folks blathering about preferring paper after "trying" both :roll: ). Always better to blame the driver for bad sound, poor imaging, etc than one's own incompetence :??.
Different strokes (or coloration) for different folks.

cheers,

AJ

p.s. I like the Accuton (hard material,suprise,suprise) drivers. I just wish some large manufacturer like HiVi made them for half the price .

Yeah, we need someone to Klone Accuton the way Usher has Kloned Scanspeak.


Now, you know you're really old when you can remember back when most of the tweeters WERE made of paper, even things some people thought were pretty good, like Allison's. My "idol" back then was when MB came out with a 3/4" hard dome plastic tweeter that actually went up beyond 20 kHz. The only way I could tell it did, compared with others, was measuring the rise time. Of course, those were the days when my main goal in life was building a speaker that did the best job possible of reproducing a 1-10 msec pulse waveform. Pulses were God... :B Needless to say, a lot of folks justifiably thought I was a little wacky. To do a really decent job took a 3/4" MB tweeter, an Audax dome mid, an Audax Bextrene (I think) 6" mid woofer, and a carefully selected 10" woofer- I don't remember now if those were sourced from CTS or Eminence.

[AJINFLA]

You know there is one rather large manufacturer that espouses the "naturalness" of paper cones and still uses it for tweeters. Or twiddlers, to be exact :W.
Me, I'm with the other Dr., Dr. Toole that is. Pistonic behavior please. Thanks.

cheers,

AJ

[Jed]

c

"I've tried dozens of different topologies with the OW1 and they just never could project female voices like the C12 can."

Curious if you've ruled out the freq response as the primary contributor to this.

Thanks

I think there are a lot more contributors to the differences in sound between OW1 and C12 than FR, even when using same slope and crossing over at identical frequencies. To be honest, the FRs were pretty close, but the sound.. . IMHO I prefer the C12 for dynamics, speed, presence, and detail in my system. I can't wait to try the accuton mids. The OW1 was a nice tweeter until I replaced it with an accuton. :T

[capslock]

other tests also confirm slight energy storage at 800Hz for the W18NX, but it is not bad at all. in general it measures well and sounds great subjectively too.

Yes, the W18L is better behaved than the W15L, that's what I wrote. Especially the W15L seems to combine the worst of both worlds (midrange cone edge reflections, bending distortion, amplified harmonics, sharp FR peaks).

[JonMarsh]

Especially the W15L seems to combine the worst of both worlds (midrange cone edge reflections, bending distortion, amplified harmonics, sharp FR peaks).

Yeah, that's EXACTLY what I want in my next system.

Maybe I'll turn that idea over to Evil Twin and see if he'll design a "killer" system with that part for all the guys over at Madisound who love paper cones..... :B OW1 tweeter, too, and 2nd order crossovers!
:rofl:

[Paul H]

Sounds excellent Jon - why don't you go all out and do a 1st order crossover - higher slopes just get in the way of the natural pure sound .... :B

Paul

[Jed]

Yeah, that's EXACTLY what I want in my next system.

Maybe I'll turn that idea over to Evil Twin and see if he'll design a "killer" system with that part for all the guys over at Madisound who love paper cones..... :B OW1 tweeter, too, and 2nd order crossovers!
:rofl:

I'm a bit embarassed that I tried out a system with RS225, W15L, and OW1 for awhile with. . .yup. . . you guessed it, 2nd order crossovers. Replacing the OW1 with the accuton, which was a recent event, was monumental to say the least. Anybody want to buy a pair of W15LYs to try out? :B I still want to hold on to the W15LYs for a bit so that I can hear the difference directly with a midrange that is much cleaner- C79 or C89 used with a low distortion tweeter. It looks like the W15LY runs into problems when played pretty loud and I rarely go over 85DBs. I recently sold a pair of W15CH magnesium drivers because I thought they were the problem with some harshness I was hearing, but now I'm thinking it was distortion from the OW1 (crossed LR4 2.3K).


Jed

[JonMarsh]

Jed,

My little joke wasn't a poke at you, Jed. Let's face it- sorting all these performance issues out is a complex and ongoing task. OTOH, I suppose there are guys who like the paper cone drivers that think I'm just as obstinate and one sided as we sometimes think they are.

What does make me scratch my head is when folks equate the Vifa XTWH09 wood pulp cone driver with the resolution of metal cone parts, but don't pay any attention to the cone resonances, frequency repsonse irregularities and mechanical resonances showing up in the impedance curve (300Hz, 400Hz, 1200 Hz, 3000Hz), and the rather medicore distortion performance. And are willing to pay $100 each for the privilege!

That's a real puzzler to me. :huh: :duh: :unsure: :nonod: :rant:

[cjd]

So, I should, like, sell the OW1's I have? (I did not buy them...)

Maybe I really do need to mate them up with these $0.49 mid-woofers and do a 3-way that'll probably tickle the fancy of those certain types, and maybe make them wonder when I tell 'em what I paid for it. Have to find woofers that fit the bill though. Anyone have a line on particularly inexpensive but still decent performing paper or even poly woofers?

C

[Jim Holtz]

You're in luck...

So, I should, like, sell the OW1's I have? (I did not buy them...)

Maybe I really do need to mate them up with these $0.49 mid-woofers and do a 3-way that'll probably tickle the fancy of those certain types, and maybe make them wonder when I tell 'em what I paid for it. Have to find woofers that fit the bill though. Anyone have a line on particularly inexpensive but still decent performing paper or even poly woofers?

C

If you want inexpensive, check out Dennis Murphys designs with the OW1 and GR Research M130 at $25 each. Lots of chatter about paper drivers in this thread but the fact is, these have a superb midrange and when ever one of Dennis's designs based on that woofer or a CA15 are at a DIY event, they get the listeners choice award. Maybe it's the distortion everyone likes. :huh:

Jim

[AJINFLA]

I'm no fan of paper cones, but just to keep things in perspective, just about every CD I own was probably mastered using them. Maybe those guys don't want to hear every detail either, just in case there might be some nastiness that would cause them extra work. No sense in fixing what 99% wont hear. In that sense, paper/plastic rules.

cheers,

AJ :W

[Jim Holtz]

I'm no fan of paper cones, but just to keep things in perspective, just about every CD I own was probably mastered using them. Maybe those guys don't want to hear every detail either, just in case there might be some nastiness that would cause them extra work. No sense in fixing what 99% wont hear. In that sense, paper/plastic rules.

cheers,

AJ :W

AJ,

Actually, I'm a big hard cone fan and detail freak too. I was just trying to add a little perspective to the discussion. I do have to admit, the M130's have a really nice midrange. I can't imagine using a poly driver in any of my designs however. Paper is the least detailed I'd use.

I wasn't kidding about the MBOW1's though. They always rank at the top of the budget class and could easily hold their own against many of the mid class designs at the DIY events. The bass isn't great but the mids are very nice. I also like Hiquphon tweeters so I guess I'm biased there too.

Jim

[Dennis H]

I think people generalize too much about cone materials. All paper/poly/metal drivers are not alike. Certainly, none of the better drivers are pure paper or pure poly but rather some sort of composite. Peerless has managed to do pretty well with the old HDS (poly sandwich) and the new HDS (paper/nomex.)

[cjd]

Yeah, I agree on the cone mats...

I was just imagining entering a system with OW1's in a budget category...

C

[cotdt]

well not all soft cones have bad resolution. i'm listening to the Peerless Exclusive 7" 830883 right now which has a soft paper cone that seems to flex easily, but it seems to match the resolution of Seas Excels! But at the same time it draws less attention to itself with a more "laid-back" sound.

[TacoD]

I've used various units (paper, kevlar, carbon, paper/carbon, paper/kevlar, HDA, polyprop, mineral filled poly, magnesium, alu, ceramic) and I have to say that each category has very good performers and very bad ones.

I think it's up to the designer to make something of it, I rarely see a DIY design which has all bases covered. Therefore I do not think there is a best material or a best topology. The sum of all parts, cabinet, x-over, driver, electrical components is the most important thing. Also I do not take people serious who are voicing on cheap electonics or only voice on one amlifier/ source.

A well executed design with any driver will bring a smile on my face. All those people talking about units without actually building/ trying something are filtered by my system. The design proces is about making compromises. Most of the time people are losing the big picture and go frenzy on small details.

I do not want to insult anybody! . I think Scanspeak/ Seas are a step ahead, they have shown in the past they can make low distortion drivers and now they are investigating other leads.

just my 2 cents (that is 2 euro cents... :T )

[Jed]

I've used various units (paper, kevlar, carbon, paper/carbon, paper/kevlar, HDA, polyprop, mineral filled poly, magnesium, alu, ceramic) and I have to say that each category has very good performers and very bad ones.

I think it's up to the designer to make something of it, I rarely see a DIY design which has all bases covered. Therefore I do not think there is a best material or a best topology. The sum of all parts, cabinet, x-over, driver, electrical components is the most important thing. Also I do not take people serious who are voicing on cheap electonics or only voice on one amlifier/ source.

A well executed design with any driver will bring a smile on my face. All those people talking about units without actually building/ trying something are filtered by my system. The design proces is about making compromises. Most of the time people are losing the big picture and go frenzy on small details.

I do not want to insult anybody! . I think Scanspeak/ Seas are a step ahead, they have shown in the past they can make low distortion drivers and now they are investigating other leads.

just my 2 cents (that is 2 euro cents... :T )

Well said Taco! I think and will add that the measurement process should enhanse the selection process not deter one from making a design decision altogether. However, I think we have to set a variety of parameters that we find acceptable/not acceptable and go from there. I think that is what Jon is doing with all his new measurements. I give him a lot of credit for testing so many drivers! Also, we all have different philosophies about what the speakers role in a system is, is it meant to reveal every nuance without coloration, or is it more of an instrument that has a set of characteristics that are pleasing to listen to (distortion)? I think there is not one path to happiness in audio, and the more we experience and learn the better we are able to substantiate our opinions. Some may even be able to call their opinions a fact.. . . in time.

Jed

[TacoD]

You're right when you say that measuring drivers is an important part of the design process, I do not deny that. In my opinion it is also important which parameters you are measuring (distortion, sensitivity, radiation). The most important thing about measuring is interpreting your results and not loosing yourself on only one aspect. But hey that's engineering and nobody told you that it was easy.

I almost forgot, the MOST important thing is making a loudspeaker you like (do not forget the "Y" in DIY) and listen to your favorite music. And if you have good taste others will like your loudspeakers too :T).

[David G]

Being an unfashionable fan of paper & poly cones, there's couple of points I'd like to make in their defence.

1. Distortion below 1% is generally considered inaudible, so I wouldn't get too hung up on it. Most decent drivers of any material have HD below 1%. I agree though, that 3% of the W15LY is getting a bit too high.

2. The slightly better CSD plots of rigid cones is all but lost once you add the steep filter. Don't believe me - compare a couple of stereophile tests of similar priced designs - a 2005 model NHT XdS high-order active speaker with 5.5" magnesium bass/mid, versus a 2003 1st order passive 8" polyconed Dynaudio Special 25 :

http://stereophile.com/standloudspeakers/1105nht/index4.html

http://stereophile.com/standloudspeakers/605dynaudio/index3.html

Scroll down to the bottom in each and compare the CSDs.
In case you missed it, the boring old 8" poly cone with 1st order crossover performs just as well, if not better, than the high tech 5.5" magnesium cone with steep, active filters.

Have a read of John Krekovskys paper on the stored energy of steep filters.

http://www.musicanddesign.com/Stored_energy_2.html

A 4th order LR adds about 0.3mS to the CSD compared to 1st order.
Don't judge a driver by it's cone material, they're much more complex than that. Also, the filter/driver combination is what's important if you want to compare CSDs.

[Dennis H]

Scroll down to the bottom in each and compare the CSDs

CSDs (and SL's ETC curves) are just another way of looking at the frequency response curve. Flat frequency response yields nice looking plots with those other ways of looking at it. The math was hashed out ad-nauseum a while back over at the madisound board. It all gets back to the notion that drivers are minimum-phase devices and all the different ways of looking at their response are just fun with mathematics.

[JonMarsh]

CSDs (and SL's ETC curves) are just another way of looking at the frequency response curve. Flat frequency response yields nice looking plots with those other ways of looking at it. The math was hashed out ad-nauseum a while back over at the madisound board. It all gets back to the notion that drivers are minimum-phase devices and all the different ways of looking at their response are just fun with mathematics.

That's all true EXCEPT one key point which seems to be overlooked in that discussion- a driver is ons signal in, MANY signals out- it's like a bi-furcated/tri-fucated on and on network- the farfield (1 meter) response at any one point is NOT the total sum of the driver output, and just like some crossover networks, you can jiggle the parameters so that on that axis point things look more or less OK amplitude wise, and still have time smear and energy storage- only by looking at the total energy output of the driver can the equivalency to an electrical network model hold true. Hence, at some points on axis, the different points of the cone vibrating out of phase add up to a smoothed response, while moving off axis reveals a different response profile (like looking at the on axis response of butter worth crossovers versus the lobing response and power response).

That's part of why I do these quite nearfield measurements (2-3") as well as ultra nearfield measurements (1/2") It turns up more of then "dirt", as a diagnostic tool. Witness, the RS225, and W22 issues around 1500-1700 Hz.

[JonMarsh]

Jon, Have you had a chance to run your Focal tweeters through the wringer? Curious to see if my affection for them is justified.

Bob

There's going to be some more "vintage" testing (after I get back from the APEC2006 conference- starts this weekend, runs till Thursday), including the Focal TC120dx2, Eton 8-800, and Accuton C88-6. (I just picked up an NOS pair of the latter at a good price to experiment with)

[Mark K]

Being an unfashionable fan of paper & poly cones, there's couple of points I'd like to make in their defence.

1. Distortion below 1% is generally considered inaudible, so I wouldn't get too hung up on it. Most decent drivers of any material have HD below 1%. I agree though, that 3% of the W15LY is getting a bit too high.

2. The slightly better CSD plots of rigid cones is all but lost once you add the steep filter. Don't believe me - compare a couple of stereophile tests of similar priced designs - a 2005 model NHT XdS high-order active speaker with 5.5" magnesium bass/mid, versus a 2003 1st order passive 8" polyconed Dynaudio Special 25 :

http://stereophile.com/standloudspeakers/1105nht/index4.html

http://stereophile.com/standloudspeakers/605dynaudio/index3.html

Scroll down to the bottom in each and compare the CSDs.
In case you missed it, the boring old 8" poly cone with 1st order crossover performs just as well, if not better, than the high tech 5.5" magnesium cone with steep, active filters.

Have a read of John Krekovskys paper on the stored energy of steep filters.

http://www.musicanddesign.com/Stored_energy_2.html

A 4th order LR adds about 0.3mS to the CSD compared to 1st order.
Don't judge a driver by it's cone material, they're much more complex than that. Also, the filter/driver combination is what's important if you want to compare CSDs.

Hi David,

I think you're not quite appreciated some of the subtlies here. You're right that a steep filter addes stored energy compared with a more gentle one, but, it does this in a much narrower frequency range This is absolutely key. Sure, if you had perfectly flat FR drivers, that were extended in both directions 2 octaves to each side, then lower order xovers would be the best choice.

That's just not how real world drivers are. The reason metal drivers can be better than ploy/paper is that they are generally more pistionic and thus have better stored energy through the bulk of the passband. So a paper/poly has minor amounts of stored energy through a wide frequency spectrum, whereas a steep slope metal combo would only have this limited to a small region in the neighborhood of the xover point. And listening tests suggest that the latter is a better combo. See the papers by Dr. O'Toole linked on my site/page on linear distortion.

Also, generalizing about what level of distortion is audible depends on the type of distortion. Higher order disto is easily audible at ~ -60dB, or 0.1%.

[AJINFLA]

Speaking of Dr. Toole begs the question Mark, what does DST have against pistonic behavior? The SS9800 is the only tweeter left, no mids/midbass and the Peerless titanium subs, which I assumed would be the reason for axing the (2yr old?) MA26WR, have disappeared from the website. Oh yeah, there is the one SS alu sub also. What gives?

Cheers,

AJ

p.s. the (optional) waveguide faceplates I had asked Ken for, I've pretty much given up on. Thats why there is D-I-YYYY...

[David G]

Hi David,

I think you're not quite appreciated some of the subtlies here. You're right that a steep filter addes stored energy compared with a more gentle one, but, it does this in a much narrower frequency range This is absolutely key. Sure, if you had perfectly flat FR drivers, that were extended in both directions 2 octaves to each side, then lower order xovers would be the best choice.

That's just not how real world drivers are. The reason metal drivers can be better than ploy/paper is that they are generally more pistionic and thus have better stored energy through the bulk of the passband. So a paper/poly has minor amounts of stored energy through a wide frequency spectrum, whereas a steep slope metal combo would only have this limited to a small region in the neighborhood of the xover point. And listening tests suggest that the latter is a better combo. See the papers by Dr. O'Toole linked on my site/page on linear distortion.

Also, generalizing about what level of distortion is audible depends on the type of distortion. Higher order disto is easily audible at ~ -60dB, or 0.1%.

G'day Mark
Your Dr Toole link doesn't appear to work.
A few points. John K's waterfall tests clearly show wideband storage over 2 octaves above & below the 2kHz crossover point in his tests. I wouldn't call that a narrow range.
Secondly, why aren't the stereophile tests revealing the "superiority" of rigid diaphragms in CSD tests? They seem pretty thorough in their measurements.
As for high order distortion, I agree about the audibility of higher order harmonics, but I don't think rigid cones are necessarily superior in this regard.
Here's another test showing the CSD of the Sonus Faber Cremona with the Scanspeak 15w4530 paper midrange.

http://www.stereophile.com/floorloudspeakers/304sonus/index3.html

[Dennis H]

http://www.linkwitzlab.com/faq.htm#Q17

Q17 - Why do you not show waterfall response plots for the PHOENIX when your test equipment is capable of producing them?

A17 - My reason for a measurement is to glean relevant information from the data presentation. The waterfall plot (cumulative spectral decay or amplitude-time-frequency presentation) is full of processing artifacts due to the necessary windowing operation performed on the impulse response. This masks information and makes it difficult to interpret what you see. The time axis is usually too short (3 ms) due to lack of anechoic measurement conditions to present anything of relevance below a few kHz. Add to this dilemma the auto-ranging of amplitude relative to the highest peak in the initial frequency response. This has the effect that a tweeter with a 22 kHz resonant peak of 10 dB pushes the whole presentation 10 dB closer to its floor so that the spectrum appears to decay very rapidly. The plots have limited usefulness and must be read with considerable understanding of how they are generated, but they provide attractive advertising and magazine graphics.

[David G]

Dennis
This thread has been pointing out issues in the CSD of the w15LY and other drivers. Now that I have highlighted other CSDs showing damped drivers with measurements at least as good as the rigid materials, you're now saying that CSDs aren't really valid anyway. You can't have it both ways.
Also, in the case of dipoles, CSD is more difficult to measure which I think is the context in which SL was referring. Also, stereophile's tests all appear to have the same measurement conditions and scales.

[Dennis H]

You can't have it both ways.

I don't want it both ways. As SL says CSDs "have limited usefulness and must be read with considerable understanding of how they are generated, but they provide attractive advertising and magazine graphics." Pretty pictures, look good in a magazine, not much useful information. The only one promoting their use here is you.

[David G]

Do you have a better scientific measurement?

[Dennis H]

The thing is, the CSD isn't a "scientific measurement." It's a calculation derived from the impulse response and it's more subject to errors than most other calcs -- almost to the point of making it worthless. The much simpler frequency response is less subject to error than the CSD although the same caveats apply -- if you derive the frequency response from the impulse response you need to thoroughly understand how the measurement was done and how the subsequent calculations were done.

[cotdt]

wouldn't measuring square waves be a better measurement than CSD's? it would be easier to read and understand

[dlr]

SL's take on this relates only to his use of MLSSA

http://www.linkwitzlab.com/faq.htm#Q17

MLSSA does use auto-ranging and is a problem for MLSSA, but not all software is limited as is MLSSA. In a thread at the Mad board, he made this comment. I pointed out that other software, LAUD to be specific, has manually set ranging, both scale and peak. I can set up a CSD of 120db range if I care (of course it is in the noise floor at that point). But the point was that the argument relating to auto-ranging is not valid.

BTW, he did not respond at all, not unlike him at times. His interest in the topics on the boards was always limited.

The point about the limited window is valid, though, so I do find it to not be very useful below about 1KHz or so.

dlr

[dlr]

The window used affects more than the CSD

The thing is, the CSD isn't a "scientific measurement." It's a calculation derived from the impulse response and it's more subject to errors than most other calcs -- almost to the point of making it worthless. The much simpler frequency response is less subject to error than the CSD although the same caveats apply -- if you derive the frequency response from the impulse response you need to thoroughly understand how the measurement was done and how the subsequent calculations were done.

The FR is subject to the same errors as the CSD. Remember, the FR is simply the convolution of the entire time-span of the window selected. The CSD is the convolution of ever-decreasing time samples. There's nothing more error-prone in the CSD vs. the FR.

The low end of the FR for a 3msec time sample can't be taken to be very reliable as well, either. It's more a matter of understanding how to interpret the CSD. As John showed, the CSD of an ideal bandpass will show the stored energy issue. A driver's CSD is simply an alteration from an ideal bandpass CSD.

The 3ms issue does certainly intrude, though, in both the CSD and the FR. One can fairly easily "see" the CSD from seeing the FR response non-linearities after a while. Switching to a "square" window rather than a tapered window also improves both the FR and CSD in the low end. I still like to use the CSD at times to get an indication of the length of time of a decaying ridge.

But if one prefers to use only MLSSA, they're going to be limited in a CSD analysis. That's why I posted a response to SL. Too bad there was no indication if he read it. He did reply to other posts afterwards, so I could only conclude that he didn't want to address the topic of CSD ranging limitations in MLSSA.

[Mark K]

G'day Mark
Your Dr Toole link doesn't appear to work.
A few points. John K's waterfall tests clearly show wideband storage over 2 octaves above & below the 2kHz crossover point in his tests. I wouldn't call that a narrow range.
Secondly, why aren't the stereophile tests revealing the "superiority" of rigid diaphragms in CSD tests? They seem pretty thorough in their measurements.
As for high order distortion, I agree about the audibility of higher order harmonics, but I don't think rigid cones are necessarily superior in this regard.
Here's another test showing the CSD of the Sonus Faber Cremona with the Scanspeak 15w4530 paper midrange.
http://www.stereophile.com/floorloud...us/index3.html

Hi David,

I just have to disagree. I see the curve that you're talking about(john k's) but I disagree about the meaning of your comment about stored energy. The ear does not process FR anomalies in the same fashion it might a phase shift. The stored energy d/t phase issues looks the same, but it is NOT treated the same by the ear. I guess we'll have to agree to disagree, but a ruler flat FR/pistionic driver is audibly superior to one that's not, all things being equal. That's what the data says. Sure, I'd rather have a transient perfect system, but I won't sacrifice FR flatness for this.

I don't really think that metal is inherently better than any other material. But I do think that, in general, they are flatter in the passband and this is a desireable characteristic. But don't get me wrong. I'm looking forward to testing some revelators and suspect they will be very, very good.

[JonMarsh]

Hi David,

But don't get me wrong. I'm looking forward to testing some revelators and suspect they will be very, very good.

And I'm looking forward to testing some more Accuton's- especially those with the new motor design. Their cones, especially the models with the ears or dampening sections, seem to be pretty well behaved- less issues to deal with than many of the metal cone drivers.

Mark, I think you'll find the nonlinear LF distortion of the Revelators to be excellent. Very good motor design. But even the 7" has a big blip in the impedance curve at 800 Hz befor starting the usual ScanSpeak climb in the upper mids. The slitting may distribute the lower chaotic cone modes over a wider range (or the blip may be the suspension, as some speculate), but either way, there are problems.

Same with the 8" woofer, but in that case the dip in the FR due to the first flapper cone mode is quite visible in even the farfield response. (and $340?!?) It's really a fairly impessive woofer, with a nice trade-off of sensitivity and longish Xmax (9 mm) and impedance (8 ohms) (probably great for a ported three way), but it's not a midwoofer. It's just a baby versiopbn of the 26W/88161T, which has it's first major cone mode at 700 Hz. But has the same lovely motor.

Think what you could have if you could combine Scanspeak motors with Eton cones. :T (an 8-800 (9" woofer) has the notch type cone mode at 1400, not 700, and it's peak at 3.2 kHz; but has a rather indifferent motor. Tunnel vision in the designers)

~Jon

[Jed]

Think what you could have if you could combine Scanspeak motors with Eton cones. :T (an 8-800 (9" woofer) has the notch type cone mode at 1400, not 700, and it's peak at 3.2 kHz; but has a rather indifferent motor. Tunnel vision in the designers)

~Jon

Eton is going to have to do something to stay competitive in the DIY market. Then again, it's probably such a small segment that they don't really care. I imagine, but don't know for sure, that folks like Avalon get special treatment and better motor designs than what is offered to us. When was the last time we saw a new driver from Eton? 1990?
:roll:

[Mark K]

And I'm looking forward to testing some more Accuton's- especially those with the new motor design. Their cones, especially the models with the ears or dampening sections, seem to be pretty well behaved- less issues to deal with than many of the metal cone drivers.

Mark, I think you'll find the nonlinear LF distortion of the Revelators to be excellent. Very good motor design. But even the 7" has a big blip in the impedance curve at 800 Hz befor starting the usual ScanSpeak climb in the upper mids. The slitting may distribute the lower chaotic cone modes over a wider range (or the blip may be the suspension, as some speculate), but either way, there are problems.

Same with the 8" woofer, but in that case the dip in the FR due to the first flapper cone mode is quite visible in even the farfield response. (and $340?!?) It's really a fairly impessive woofer, with a nice trade-off of sensitivity and longish Xmax (9 mm) and impedance (8 ohms) (probably great for a ported three way), but it's not a midwoofer. It's just a baby versiopbn of the 26W/88161T, which has it's first major cone mode at 700 Hz. But has the same lovely motor.

Think what you could have if you could combine Scanspeak motors with Eton cones. :T (an 8-800 (9" woofer) has the notch type cone mode at 1400, not 700, and it's peak at 3.2 kHz; but has a rather indifferent motor. Tunnel vision in the designers)

~Jon

I agree the accutons look very promising. I've thought about spinging for the updated tweeter.

I understand what you're saying about the cone modes/impedance glitch. I think where you and I differ is that I think if you can get the system final response flat and the cone modes don't impact the on and off axis FR too much, then the more advanced motor and better nonlinear distortion may be a better tradeoff. Especially in a two way, where a lot of excursion is demanded of the midwoofer.

[Jed]

It appears that even the C79 has an impedance glitch at 800HZ. Perhaps those "ears" are the cause, even though they benefit the FR above 2K as opposed to the C89. Any speculation as to which of these 2 drivers will be the best performer as a mid in a 3-way? Jon you mentioned some possible combinations of RS225 C79, C13 or RS270, C89, C24. I think the highest the C89 should go is 2K based on the published distortion which is probably why you recommended the C24 as opposed to a C12 or C13? I'm about to spring for some accuton mids, already have the C44s but want something that is clean down to 250HZ or so. I guess I could save up for a C90 and retire from audio/be done with it. But man- $460 a pop!