Lynn Olson said:
Charles Hansen said:
For those too young to know about the B110 from KEF I provide some pics and a quick CSD measurement.
As can be seen by the CSD this ancient polyprop-driver easily can compete with current metal cones like from Seas/ Excel (or large format 4" compression drivers ?) when it comes to membrane breakup resonance.
Actually this particular B110 is a tuned version from the Linn Kan.
That small foot print speaker was really impressive at its days – very good imaging, timing and tonal balance – though it had kind of issue with piano music.
There were several modifications made by Linn at this speaker. Some can be seen at the pictures
- coating at both sides of the membrane (seems to bee very effective for the dust cap resonancies Lynn mentioned)
- strips of lead at the basket and at the magnet
- coating of the basket
- holes through the membrane for voice coil ventilation - a technique copied by Peerless later on as can be seen here:
http://www.diyaudio.com/forums/showthread.php?postid=1279267#post1279267
-----------------------------------
Some rather amazing CSD behaviours I've put together earlier:
Though these CSD reflect speakers in cabinets – similar effects may be seen at speakers without cabinets once we would measure with higher resolution according to the faster decay of a speaker alone.
http://www.diyaudio.com/forums/showthread.php?postid=1202358#post1202358
When the stimulus is turned off - the stored energy is released in rather complex patterns of "hidden resonances".
Some effects are more clearly visibly with a sonogram. This is basically a CSD measurement seen from the top and displaying the "height of the mountains" in a color code.
What I observed, I haven't seen elsewhere or read about in such clarity before:
- different slopes for different frequencies
- build up of resonances at certain frequencies
- non linear slopes that are "walking down in steps"
- energy swapping between adjacent frequencies
- resonancies smoothly changing frequency with time of decay
Linn, could you tell about the enclosure your Eaton CSD has being taken with ?
Greetings
Michael
Planar CSD
Other than the raggedness above ~11khz or so, does this look to be pretty good behavior? Down 30dB at ~ 1.25 msec...
John L.
Other than the raggedness above ~11khz or so, does this look to be pretty good behavior? Down 30dB at ~ 1.25 msec...
An externally hosted image should be here but it was not working when we last tested it.
John L.
That B110 is just as bad as I remember. I hated those things - they required equalization up the yin-yang, burnt up the paper voice-coil former if you looked at them cross-eyed, and were the most inefficient and peaky driver I've ever used. That the BBC got good sound of them in the LS3/5A is nothing less than a miracle - but that still doesn't mean they were a good driver.
The driver I really liked in Audionics TL-M 200 was the one just below the B110 - a Richard Allan 7" cast-frame midbass driver with a Bextrene cone. Now that was a quality driver - much better in fact than the wretched B110 just above. Unfortunately, I "inherited" the TL-M 200 project from a designer that high-tailed it out of town and left no forwarding address (he was smart!), and I was stuck with the speaker as-is, and had the make the existing driver complement (B139, RA. B110, and T27) somehow work all together.
Talk about being thrown into the deep end of the pool. These days, I would ***t-can the drivers and tell the management they'd just have to write off the cost of the useless prototype boxes - not *my* fault the previous designer didn't have a clue what he was doing. That's why I still have a bit of a chip on my shoulder about so-called "genius designers" and their half-witted driver selections. In my first speaker project, I had to clean up after one of these clowns, and it wasn't any fun.
I guess this experience is also the reason I don't cut driver builders any slack. They get away with the stunts they do because most speaker designers don't hold them accountable for the dumb mistakes they make. I know from hard experience there are very real limits to how much you can work-around driver faults - and when you look at the sum total of hours-to-design and the net results, it's better to harass the driver builders and look as hard as possible for alternatives (your negotiating power is also a bit better if you're aware of the alternatives, and the exact ways in which the competitor has a superior product).
Regarding the earlier mention of compression drivers and direct-radiator domes, one point I didn't make is that phase-plugs and horn/waveguides do not change the inherent sound of the diaphragm material. If you don't like plastic-dome tweeters, you shouldn't expect a plastic-dome compression driver to sound all that different. Similarly, aluminum, titanium, and beryllium dome tweeters will have noticeable kinship to their compression-driver equivalents. This is just one of those things you have to deal with.
Although the phase-plug and horn/waveguide don't do that much to damp diaphragm breakup and resonances, one huge difference is efficiency and headroom - about a 50X to 100X difference. A typical dome tweeter is about 90 dB/metre efficient, and the highest I've ever seen is 95 dB/metre. By contrast, 109 to 111 dB/metre is the norm for compression drivers with the appropriate horn/waveguide.
The rated pink-noise power capacity of an Altec 288-G is 15 watts (modern SR drivers are several times higher) with an efficiency of 110 dB at four feet. Let's convert that to 111 dB at one metre for ease of calculation. A direct-radiator dome tweeter is doing pretty well if it is 91 dB/metre efficient. For the dome tweeter to put out the same level of acoustic watts as the 288-G, it would have to accept 1500 watts! Well, maybe it could, but only for a few milliseconds, only once, and it would be pretty distorted during that brief interval.
By contrast, the Altec rating is a commercial rating intended for continuous use in an application where driver failure results in considerable expense and inconvenience for the theater-owner. That's rather different than the rather optimistic power rating for consumer products, where the number seems to be picked out of the air.
So the tradeoff becomes 100X the headroom vs diaphragm breakup happening one to two octaves lower than the direct-radiator. We can now buy very good supertweeters from several vendors, and can measure and design crossovers with far more precision (not true back in the Fifties). Hmm, which way should we go?
The driver I really liked in Audionics TL-M 200 was the one just below the B110 - a Richard Allan 7" cast-frame midbass driver with a Bextrene cone. Now that was a quality driver - much better in fact than the wretched B110 just above. Unfortunately, I "inherited" the TL-M 200 project from a designer that high-tailed it out of town and left no forwarding address (he was smart!), and I was stuck with the speaker as-is, and had the make the existing driver complement (B139, RA. B110, and T27) somehow work all together.
Talk about being thrown into the deep end of the pool. These days, I would ***t-can the drivers and tell the management they'd just have to write off the cost of the useless prototype boxes - not *my* fault the previous designer didn't have a clue what he was doing. That's why I still have a bit of a chip on my shoulder about so-called "genius designers" and their half-witted driver selections. In my first speaker project, I had to clean up after one of these clowns, and it wasn't any fun.
I guess this experience is also the reason I don't cut driver builders any slack. They get away with the stunts they do because most speaker designers don't hold them accountable for the dumb mistakes they make. I know from hard experience there are very real limits to how much you can work-around driver faults - and when you look at the sum total of hours-to-design and the net results, it's better to harass the driver builders and look as hard as possible for alternatives (your negotiating power is also a bit better if you're aware of the alternatives, and the exact ways in which the competitor has a superior product).
Regarding the earlier mention of compression drivers and direct-radiator domes, one point I didn't make is that phase-plugs and horn/waveguides do not change the inherent sound of the diaphragm material. If you don't like plastic-dome tweeters, you shouldn't expect a plastic-dome compression driver to sound all that different. Similarly, aluminum, titanium, and beryllium dome tweeters will have noticeable kinship to their compression-driver equivalents. This is just one of those things you have to deal with.
Although the phase-plug and horn/waveguide don't do that much to damp diaphragm breakup and resonances, one huge difference is efficiency and headroom - about a 50X to 100X difference. A typical dome tweeter is about 90 dB/metre efficient, and the highest I've ever seen is 95 dB/metre. By contrast, 109 to 111 dB/metre is the norm for compression drivers with the appropriate horn/waveguide.
The rated pink-noise power capacity of an Altec 288-G is 15 watts (modern SR drivers are several times higher) with an efficiency of 110 dB at four feet. Let's convert that to 111 dB at one metre for ease of calculation. A direct-radiator dome tweeter is doing pretty well if it is 91 dB/metre efficient. For the dome tweeter to put out the same level of acoustic watts as the 288-G, it would have to accept 1500 watts! Well, maybe it could, but only for a few milliseconds, only once, and it would be pretty distorted during that brief interval.
By contrast, the Altec rating is a commercial rating intended for continuous use in an application where driver failure results in considerable expense and inconvenience for the theater-owner. That's rather different than the rather optimistic power rating for consumer products, where the number seems to be picked out of the air.
So the tradeoff becomes 100X the headroom vs diaphragm breakup happening one to two octaves lower than the direct-radiator. We can now buy very good supertweeters from several vendors, and can measure and design crossovers with far more precision (not true back in the Fifties). Hmm, which way should we go?
auplater said:
The gated time window is much too short at 2.17 mSec. If possible, try to re-measure with at least 7~8 mSec interval. This will require the annoyance of the two-foot-high pillow pile, but it is unavoidable if you want a usable CSD. I can assure you the CSD looks *very* different with a longer time window. If you cannot get a clean time window of at least 5~6 mSec, I regret to inform you that you cannot perform CSD's or reasonably detailed FR measurements.
What gets my attention is the rough FR at the back of the CSD. Although not very high in resolution, it is not too smooth in overall peak-to-dip magnitudes, which should appear in the CSD - and doesn't, thanks to premature truncation of the time window. Rest assured that uneven frequency response always has a visible impact on the time domain - provided the measurement is allowed to show it.
On the other hand, the ups-and-downs could be completely benign, and nothing more than simple comb filtering from the large vertical radiating area. If this is the case, measuring at different distances and microphone heights (close-up, 1 meter, 2 meters) will reveal that the bumps-and-dips move around with measuring distance. If the bumps-and-dips stay the same, that's not so good - that means the large diaphragm itself is going into breakup modes, and there's nothing you can do about it. Suppressing modes from large plastic diaphragms is not trivial - the electrostatic folks have spent decades on that one.
Regarding the B110 measurements shown above, that is nearly as bad as it gets. If the time interval were longer, and the 1/12 octave smoothing removed (the protocol I use), then it would appear considerably worse, and the fine structure of the multiple resonances in the 4~5 kHz region would start to appear. When those multiple, highly directional resonances are excited by female choir, the B110 can really shriek at you - the notorious "shhh" coloration of so many Bextrene speakers.
I guess the reason I'm so touchy about these obnoxious 1~5 kHz colorations is that I've spent so much time trying to get rid of them - from what I can tell, magazine reviewers can't seem to hear them at all, and some demented audiophiles actually think that is what "resolution" and "detail" sounds like.
P.S. The Eton measurements were taken with a larger-than-IEC flat baffle, with the Eton mounted about 20% off-center, the driver centerline about 2 meters off the floor, and the flange of the driver flush with the baffle. Unless the test driver is mounted in a near-spherical enclosure, the test box will contaminate the time response and make it very difficult to see driver artifacts.
CSDplot
Thanx Lynn. Always good to get an eval. from someone with extensive experience in the field. This is ~10 yr. old data I pulled from somewhere on the net years ago (probably John Whitacre's study), so I doubt anyone's done much more exploring this driver.
I wondered about the gating vs. resolution issue...I notice many CSD's offered don't include the res. info... looks like the Eton is gated @ 1.511 msecs?? and the KEF @ 2.97 msecs...? so does that make these waterfalls less than useful?
The BG's sure sound pretty mello and sweet in my setup, tho... so I wonder about the usefulness of the CSD's vs. what I hear...
John L.
Thanx Lynn. Always good to get an eval. from someone with extensive experience in the field. This is ~10 yr. old data I pulled from somewhere on the net years ago (probably John Whitacre's study), so I doubt anyone's done much more exploring this driver.
I wondered about the gating vs. resolution issue...I notice many CSD's offered don't include the res. info... looks like the Eton is gated @ 1.511 msecs?? and the KEF @ 2.97 msecs...? so does that make these waterfalls less than useful?
The BG's sure sound pretty mello and sweet in my setup, tho... so I wonder about the usefulness of the CSD's vs. what I hear...
John L.
Could you elaborate on how the CSD would vary with a longer window? Are there any comparisons that can be shown to help us better understand the influence? Normally I would expect the differences to occur at the lower frequencies and at extented time normally at much lower levels. The higer frequencies and the shorter time duration would remain unchanged.Lynn Olson said:
plots vs. hearing
here are some freq. response plots presented in various formats... any of them mean anything at all wrt. what one actually hears??
Rd75
RAAL tweeter
ATC Neoplanar
Peerless 6 1/2" mid woofer
and a polar plot to boot..
ATC Neoplanar
or is all this data just for naught as far as designing and subsequent listening goes??
thanx
John L.
here are some freq. response plots presented in various formats... any of them mean anything at all wrt. what one actually hears??
Rd75
An externally hosted image should be here but it was not working when we last tested it.
RAAL tweeter
An externally hosted image should be here but it was not working when we last tested it.
ATC Neoplanar
An externally hosted image should be here but it was not working when we last tested it.
Peerless 6 1/2" mid woofer
An externally hosted image should be here but it was not working when we last tested it.
and a polar plot to boot..
ATC Neoplanar
An externally hosted image should be here but it was not working when we last tested it.
or is all this data just for naught as far as designing and subsequent listening goes??
thanx
John L.
Hi Lynn
not sure whether your negative recption of my B110 posting is due to my interference with a long time vendetta of yours as a reviewer and some specific companies or anything else.
Anyway, it delivers hard data valid in the range presented. No academic study about the B110 was intended nor promised elsewhere though.
My subjective and very positive comments about the Linn Kan are widely shared and explicit do not apply to the bigger speakers of that time like the Sara and the Isobaric, where you can HEAR the effects of energy transfer between one and the other bass speakers as kind of crude beats ("Schwebung" in German).
Rogers on the other hand were designed to be studio monitors afaik not as consumer product and they for sure were not the only company that succeeded in balancing out downsides of the material they had available.
If you crank up resolution in terms of less smoothing – yes - you will see more artefacts. But you also will have to make a LOT more measurements to make sure what they are telling you – and even than the interpretation of its sonic relevance will be highly subjective.
To my knowledge no one ever has put together a set of patterns that clearly can identify – from analysing a CSD measurement for example - different materials of cones surroundings baskets VC formers etc.
If you crank up resolution in terms of larger time window – no – there wouldn't be that much more to see, except at the low frequency end where invalid data is cropped in ARTA.
And no, no need to stude the subjet for centuries to grasp the main issues displayerd clearly on CSD measurement.
What my B110 CSD's are really telling can be read between the lines: no progress concerning cone break up during the last centuries - regardless of cone material.
One has to accept that crossing at 2,5kHz requires a speaker of less than 4" - 6" maximum – maybe even less, when liking low order networks.
This may be hard to digest even more so if combined with the desire of achieving low IM which restricts the usable bandwidth covered by a single speaker at high output even more.
With your approach of high output at low IM and low order XO you are defiantly dancing along the borders of physics .
Among others, this makes your thread "juicy" and interesting for me and makes me contributing from time to time.
Yes the KEF B110 was an inferiour speaker - in what you outlined (guess why I have ONE spare).
And no, the KEF B110 was an excellent speaker - you will be hard pressed to find one that is substantially better in fast decay BELOW it's cone breake up resonnace.
Greetings
Michael
not sure whether your negative recption of my B110 posting is due to my interference with a long time vendetta of yours as a reviewer and some specific companies or anything else.
Anyway, it delivers hard data valid in the range presented. No academic study about the B110 was intended nor promised elsewhere though.
My subjective and very positive comments about the Linn Kan are widely shared and explicit do not apply to the bigger speakers of that time like the Sara and the Isobaric, where you can HEAR the effects of energy transfer between one and the other bass speakers as kind of crude beats ("Schwebung" in German).
Rogers on the other hand were designed to be studio monitors afaik not as consumer product and they for sure were not the only company that succeeded in balancing out downsides of the material they had available.
If you crank up resolution in terms of less smoothing – yes - you will see more artefacts. But you also will have to make a LOT more measurements to make sure what they are telling you – and even than the interpretation of its sonic relevance will be highly subjective.
To my knowledge no one ever has put together a set of patterns that clearly can identify – from analysing a CSD measurement for example - different materials of cones surroundings baskets VC formers etc.
If you crank up resolution in terms of larger time window – no – there wouldn't be that much more to see, except at the low frequency end where invalid data is cropped in ARTA.
And no, no need to stude the subjet for centuries to grasp the main issues displayerd clearly on CSD measurement.
What my B110 CSD's are really telling can be read between the lines: no progress concerning cone break up during the last centuries - regardless of cone material.
One has to accept that crossing at 2,5kHz requires a speaker of less than 4" - 6" maximum – maybe even less, when liking low order networks.
This may be hard to digest even more so if combined with the desire of achieving low IM which restricts the usable bandwidth covered by a single speaker at high output even more.
With your approach of high output at low IM and low order XO you are defiantly dancing along the borders of physics .
Among others, this makes your thread "juicy" and interesting for me and makes me contributing from time to time.
Yes the KEF B110 was an inferiour speaker - in what you outlined (guess why I have ONE spare).
And no, the KEF B110 was an excellent speaker - you will be hard pressed to find one that is substantially better in fast decay BELOW it's cone breake up resonnace.
Greetings
Michael
data
Interesting that Lynn's response was so negative...I've never heard any combing from this single driver, nor have I measured any. And the frequency variations are not all that large and may in fact be due to measurement artfacts (although I wonder what the mechanism for combing from a single driver is, especially since I don't hear any).
Some of this thread reads like an old Stereo Review by Julian Hirsch...
I s'ppose we all have our own baseline and bias wrt using data to reach an end.
John L.
Interesting that Lynn's response was so negative...I've never heard any combing from this single driver, nor have I measured any. And the frequency variations are not all that large and may in fact be due to measurement artfacts (although I wonder what the mechanism for combing from a single driver is, especially since I don't hear any).
Some of this thread reads like an old Stereo Review by Julian Hirsch...
I s'ppose we all have our own baseline and bias wrt using data to reach an end.
John L.
Well, all I can say that listening to a driver in a finished speaker, and attempting to use it in a loudspeaker, are two different experiences. When the loudspeaker is a finished product, there are many ways to conceal substantial faults from the auditioner. I can think of several:
1) Tweeter IM distortion is concealed by much louder output from the midbass driver beneath it. You can actually have a tweeter that is quite obviously distorting, and not really hear this for what it is, until you disconnect the midbass. Then the source of the intermittent lack-of-transparency and odd coloration becomes very obvious as a fizzing, kazoo-like noise from the tweeter.
2) Listen to a driver full-range on a baffle at moderate levels with no crossover, and compare it to known rapid-CSD-decay driver. The colorations will be most obvious with pink-noise, and somewhat less so with full-scale symphonic and choral works. The CSD colorations may not be audible at all with rock, blues, or jazz quartets. The audibility of overhangs and resonances in the CSD is a direct function of spectral density in the frequency range of the colorations.
3) Take a driver with a substantial CSD resonance and build a notch filter to compensate, then toggle the network with pink-noise, symphonic, choral, rock, blues, and jazz quartets (in order of audibility). You may find that the notch filter will be the best choice for noise, symphonic, and choral, and less desirable for rock, blues, and jazz, where it may degrade the sense of impact and dynamics. What then? Do you give the user a "coloration" switch?
These options are not available to the reviewer or owner unless they want to start tampering with the finished product and void the warranty. The consumer has to take the product as-is, and like it or not. The internal design choices are not always audible in the finished result. The person who designed it, though, will probably be very aware of the shortcomings that remain, even if they don't choose to tell anyone else about it.
I used to take magazine reviewers seriously until I started meeting them, and heard their systems in their own houses. Reviewers write well, but it stops there. They have no better taste than the average audiophile (yes, I'm including the big-name reviewers here), and their knowledge of audio technology is about comparable to a hifi salesman working on commission. That's why so many product reviews uncritically repeat absurd statements lifted from advertising literature - the reviewer (and publisher) really don't know any better, so in it goes. Besides, it gives the review a shiny gloss of engineering know-how, leavening the miniature psychodrama that opens the "feature" reviews in the high-profile magazines.
During the brief time I wrote reviews for Positive Feedback magazine, it was so stupefyingly dull that I went off on an absurdist tangent and started mixing in science-fiction and UFO stories to liven things up. If you think hifi reviews are dull reading, try writing one! It's worse - and unlike a magazine you can throw away when you're through with it, you still have the deadline looming over you and XYZ words needed to fill the space. A year of this was all I could take. So I have a little sympathy for reviewers, having paid a few dues myself - and now I can tell when the editors of the magazine are routinely re-writing what the reviewer originally wrote. (This is standard practice for at least one of the high-profile magazines.)
I'll return to the point that the best way to get to know a driver is measure and audition it in your own design. If it works for you, great! That saves a lot of time. Most of the time, though, I'm fault-finding mode, trying to understand the underlying operating principle of a driver, and what it's good at, where it falls down, and the methods available to improve performance (if any).
About the CSD. I use it as an analytical tool. If it doesn't reveal resonances, standing waves, reflections in the cabinet, and diffraction, it's not much use for debugging purposes. Increasing the time window increases the frequency resolution, and this is very desirable to see where the driver problems might be.
But the CSD is not a simple go-nogo instrument; comb filtering from a large surface area like an electrostat or a line source can appear similar to driver breakup, but they are completely separate things, and sound completely different too. A diffuse source may not measure that well, but they don't sound that bad in real life - just big. Resonances, on the other hand, sound like what they are - colorations from a driver that is breaking up. That's why a single CSD doesn't tell the whole story, or even part of it. It's just another measurement tool.
There are lots of ways to make a CSD reading useless - early room reflections, poor choice of mike position, reflections from a badly designed mike-stand, reflections off the test equipment or even the test person. If these unwanted reflections are left in, they turn the CSD display to unreadable gibberish.
I'm guessing here, but I suspect that's why many published CSD's are derived from such short time windows - there were problems with the original time measurement, so the tester shortened the window, shorter and shorter, until the CSD finally cleaned up. That's why I like to see the CSD published alongside the original time data from which it was derived - that tells a lot about the test condition.
1) Tweeter IM distortion is concealed by much louder output from the midbass driver beneath it. You can actually have a tweeter that is quite obviously distorting, and not really hear this for what it is, until you disconnect the midbass. Then the source of the intermittent lack-of-transparency and odd coloration becomes very obvious as a fizzing, kazoo-like noise from the tweeter.
2) Listen to a driver full-range on a baffle at moderate levels with no crossover, and compare it to known rapid-CSD-decay driver. The colorations will be most obvious with pink-noise, and somewhat less so with full-scale symphonic and choral works. The CSD colorations may not be audible at all with rock, blues, or jazz quartets. The audibility of overhangs and resonances in the CSD is a direct function of spectral density in the frequency range of the colorations.
3) Take a driver with a substantial CSD resonance and build a notch filter to compensate, then toggle the network with pink-noise, symphonic, choral, rock, blues, and jazz quartets (in order of audibility). You may find that the notch filter will be the best choice for noise, symphonic, and choral, and less desirable for rock, blues, and jazz, where it may degrade the sense of impact and dynamics. What then? Do you give the user a "coloration" switch?
These options are not available to the reviewer or owner unless they want to start tampering with the finished product and void the warranty. The consumer has to take the product as-is, and like it or not. The internal design choices are not always audible in the finished result. The person who designed it, though, will probably be very aware of the shortcomings that remain, even if they don't choose to tell anyone else about it.
I used to take magazine reviewers seriously until I started meeting them, and heard their systems in their own houses. Reviewers write well, but it stops there. They have no better taste than the average audiophile (yes, I'm including the big-name reviewers here), and their knowledge of audio technology is about comparable to a hifi salesman working on commission. That's why so many product reviews uncritically repeat absurd statements lifted from advertising literature - the reviewer (and publisher) really don't know any better, so in it goes. Besides, it gives the review a shiny gloss of engineering know-how, leavening the miniature psychodrama that opens the "feature" reviews in the high-profile magazines.
During the brief time I wrote reviews for Positive Feedback magazine, it was so stupefyingly dull that I went off on an absurdist tangent and started mixing in science-fiction and UFO stories to liven things up. If you think hifi reviews are dull reading, try writing one! It's worse - and unlike a magazine you can throw away when you're through with it, you still have the deadline looming over you and XYZ words needed to fill the space. A year of this was all I could take. So I have a little sympathy for reviewers, having paid a few dues myself - and now I can tell when the editors of the magazine are routinely re-writing what the reviewer originally wrote. (This is standard practice for at least one of the high-profile magazines.)
I'll return to the point that the best way to get to know a driver is measure and audition it in your own design. If it works for you, great! That saves a lot of time. Most of the time, though, I'm fault-finding mode, trying to understand the underlying operating principle of a driver, and what it's good at, where it falls down, and the methods available to improve performance (if any).
About the CSD. I use it as an analytical tool. If it doesn't reveal resonances, standing waves, reflections in the cabinet, and diffraction, it's not much use for debugging purposes. Increasing the time window increases the frequency resolution, and this is very desirable to see where the driver problems might be.
But the CSD is not a simple go-nogo instrument; comb filtering from a large surface area like an electrostat or a line source can appear similar to driver breakup, but they are completely separate things, and sound completely different too. A diffuse source may not measure that well, but they don't sound that bad in real life - just big. Resonances, on the other hand, sound like what they are - colorations from a driver that is breaking up. That's why a single CSD doesn't tell the whole story, or even part of it. It's just another measurement tool.
There are lots of ways to make a CSD reading useless - early room reflections, poor choice of mike position, reflections from a badly designed mike-stand, reflections off the test equipment or even the test person. If these unwanted reflections are left in, they turn the CSD display to unreadable gibberish.
I'm guessing here, but I suspect that's why many published CSD's are derived from such short time windows - there were problems with the original time measurement, so the tester shortened the window, shorter and shorter, until the CSD finally cleaned up. That's why I like to see the CSD published alongside the original time data from which it was derived - that tells a lot about the test condition.
Re: Re: plots vs. hearing
I agree. The test data shows a benign, smooth-measuring driver - about what you'd expect from a good-quality polypropylene driver of fairly modern vintage. Can you tell what it sounds like from this data? Not really.
In my experience, some types of drivers seem to sound much better than they measure, and some types sound worse. Paper cones, for example, usually sound better than what see in the FR and CSD, although the time response might be better than you expect. Conversely, polypropylene can have an odd "plastic" coloration that just doesn't seem to show up in any measurements I've seen.
I seem to have stepped on some toes with my comments about the B110. Hey, the LS3/5A is one of my favorite speakers. But the B110 was a dog to design with, I can tell you that, one of the most time-consuming drivers I've ever worked with. If a driver takes many, many months of tedious twiddling of filters and subjective auditioning, I count that against it, even if the final result is reasonably gratifying.
The BBC had good reason to get away from all the problems of Bextrene drivers and pioneer the use of polypropylene - they were wanting unit-to-unit consistency and freedom from resonance, and Bextrene drivers have severe resonances that require substantial amounts of hand-applied damping compound, which knocks down efficiency and introduces a new set of problems with manufacturing quality control. (Hand-applied anything is a problem in manufacturing, much less applying a compound directly to the cone, where it affects all of the most important measurements of the driver.)
Is polypropylene "better" in the absolute sense than Bextrene? Well, it's easier to design around, is more consistent, and is more efficient by several dB. But in the absolute sense? I'm not sure the question is answerable, since it comes down to personal esthetics.
soongsc said:
I agree. The test data shows a benign, smooth-measuring driver - about what you'd expect from a good-quality polypropylene driver of fairly modern vintage. Can you tell what it sounds like from this data? Not really.
In my experience, some types of drivers seem to sound much better than they measure, and some types sound worse. Paper cones, for example, usually sound better than what see in the FR and CSD, although the time response might be better than you expect. Conversely, polypropylene can have an odd "plastic" coloration that just doesn't seem to show up in any measurements I've seen.
I seem to have stepped on some toes with my comments about the B110. Hey, the LS3/5A is one of my favorite speakers. But the B110 was a dog to design with, I can tell you that, one of the most time-consuming drivers I've ever worked with. If a driver takes many, many months of tedious twiddling of filters and subjective auditioning, I count that against it, even if the final result is reasonably gratifying.
The BBC had good reason to get away from all the problems of Bextrene drivers and pioneer the use of polypropylene - they were wanting unit-to-unit consistency and freedom from resonance, and Bextrene drivers have severe resonances that require substantial amounts of hand-applied damping compound, which knocks down efficiency and introduces a new set of problems with manufacturing quality control. (Hand-applied anything is a problem in manufacturing, much less applying a compound directly to the cone, where it affects all of the most important measurements of the driver.)
Is polypropylene "better" in the absolute sense than Bextrene? Well, it's easier to design around, is more consistent, and is more efficient by several dB. But in the absolute sense? I'm not sure the question is answerable, since it comes down to personal esthetics.
Measurements, and PHL 1120 as a midrange
Speaking of measurements, and how they can be sometimes tricky to interpret:
Regarding the PHL 1120 that was named as a midrange
[Side note: Lynn, Aleksandar from RAAL seems to recommend the PHL -- in particular the PHL 1120 above -- and not the PHY (or PHY-HP, whatever you want to call them) -- as a midrange. Quite a bit of difference]
That will make three of us.
Is it just me or in the measurments above the ripples above 2.5k are the talesign of cone breakup ? Did anyone noticed that with the PHL 1120 ?
OTOH other detailed measurements of that very same driver -- namely these -- look very different.
... And suggest (from the impedance plot) that the driver is good down to about 400 Hz (or 300Hz, but maybe that's pushing it ?)
Anyone care to comment on that ?
Speaking of measurements, and how they can be sometimes tricky to interpret:
Regarding the PHL 1120 that was named as a midrange
[Side note: Lynn, Aleksandar from RAAL seems to recommend the PHL -- in particular the PHL 1120 above -- and not the PHY (or PHY-HP, whatever you want to call them) -- as a midrange. Quite a bit of difference]
Bratislav said:
That will make three of us.
Is it just me or in the measurments above the ripples above 2.5k are the talesign of cone breakup ? Did anyone noticed that with the PHL 1120 ?
OTOH other detailed measurements of that very same driver -- namely these -- look very different.
... And suggest (from the impedance plot) that the driver is good down to about 400 Hz (or 300Hz, but maybe that's pushing it ?)
Anyone care to comment on that ?
"I agree. The test data shows a benign, smooth-measuring driver - about what you'd expect from a good-quality polypropylene driver of fairly modern vintage."
I hope this isn't directed at the data I posted. There are no polypropylene drivers in this data. the 6 1/2" mids are peerless/tymphani
http://www.tymphany.com/datasheet/printview.php?id=12
chosen to match the efficiency of the BG 75 72" planars, as I measured and chose to match their sensitivity (they use PEt film with AL bonded vc)
The neoplanars use mylar film with screened conductive ribbons... and are attenuated to match the lower sensitivity of the other drivers.
http://209.85.175.104/search?q=cach...uctinfo.pdf+BG+RD+75&hl=en&ct=clnk&cd=5&gl=us[/url]
and I chose them after careful listening. They are used as widerange mids, from 600 Hz to ~10 Khz,, with gentle L/R xovers to the peerless mids, which run up to ~ 1.5Khz matched to the shelved BG response with a xover network and BSC at ~230 Hz.
The high end is handled first by the planars running free from ~600Hz up, reinforced from ~7Khz up with a front firing EAS10Th panasonic leaf tweeter and a rear firing peerless dome. All these were carefully "tweaked" after installation, both by measurement and by listening, to "voice" the individual and combined driver profiles to create a convincing and smooth lifelike image in a large arc at 15' with low distortion and smooth response and imaging that is extremely convincing to above 105dB(C)avg. for HT and symphonic work. All this started out after several weeks of design, measurement, and "mules" to test various configs, until the (almost) finished design emerged. CSD's never came into play, other than I looked at a few for the drivers, saw nothing unusual (yes, even at extended windowing)...
the image is very stable throughout the room, with almost no variation across ~ 90 degrees subtended front, piano, solo guitar, and solo female vocalsare frighteningly realistic (jump factor as in "someone's actually in the room playing!")
Thanx for your input... you've confirmed what I thought might be going on here...
John L.
I hope this isn't directed at the data I posted. There are no polypropylene drivers in this data. the 6 1/2" mids are peerless/tymphani
http://www.tymphany.com/datasheet/printview.php?id=12
chosen to match the efficiency of the BG 75 72" planars, as I measured and chose to match their sensitivity (they use PEt film with AL bonded vc)
The neoplanars use mylar film with screened conductive ribbons... and are attenuated to match the lower sensitivity of the other drivers.
http://209.85.175.104/search?q=cach...uctinfo.pdf+BG+RD+75&hl=en&ct=clnk&cd=5&gl=us[/url]
and I chose them after careful listening. They are used as widerange mids, from 600 Hz to ~10 Khz,, with gentle L/R xovers to the peerless mids, which run up to ~ 1.5Khz matched to the shelved BG response with a xover network and BSC at ~230 Hz.
The high end is handled first by the planars running free from ~600Hz up, reinforced from ~7Khz up with a front firing EAS10Th panasonic leaf tweeter and a rear firing peerless dome. All these were carefully "tweaked" after installation, both by measurement and by listening, to "voice" the individual and combined driver profiles to create a convincing and smooth lifelike image in a large arc at 15' with low distortion and smooth response and imaging that is extremely convincing to above 105dB(C)avg. for HT and symphonic work. All this started out after several weeks of design, measurement, and "mules" to test various configs, until the (almost) finished design emerged. CSD's never came into play, other than I looked at a few for the drivers, saw nothing unusual (yes, even at extended windowing)...
the image is very stable throughout the room, with almost no variation across ~ 90 degrees subtended front, piano, solo guitar, and solo female vocalsare frighteningly realistic (jump factor as in "someone's actually in the room playing!")
Thanx for your input... you've confirmed what I thought might be going on here...
John L.
Well, the hardest part of the voicing would be getting a good subjective match between drivers with very different emission shapes and sizes, and it sounds like Mr. Auplater has already done that. It's normal to see ripples in the response of line radiators, but it isn't audible all that much from what I can tell. The radiator just sounds "big", which of course it is.
As for my new speaker, in the simplest manifestation, it would be nothing more than a classic 12~15" studio monitor in a big box with a top-mounted horn with a large-format compression driver and a time-delayed supertweeter. The crossovers would similarly be at the traditional frequencies of 700~880 Hz and 6~8 kHz. The antecedents of the Lansing Iconic, Altec A5, and Klipsch Cornwall/Chorus go back to the Thirties, Forties, and Fifties.
The more adventurous version that I'll finally be building this summer will have an open baffle instead of the traditional big box, and an array of bi-amped bass-fill drivers to complement the OB rolloff. If the OB thing doesn't pan out, I'll revert to the more traditional big box, but I'm hoping I won't have to do that.
As for my new speaker, in the simplest manifestation, it would be nothing more than a classic 12~15" studio monitor in a big box with a top-mounted horn with a large-format compression driver and a time-delayed supertweeter. The crossovers would similarly be at the traditional frequencies of 700~880 Hz and 6~8 kHz. The antecedents of the Lansing Iconic, Altec A5, and Klipsch Cornwall/Chorus go back to the Thirties, Forties, and Fifties.
The more adventurous version that I'll finally be building this summer will have an open baffle instead of the traditional big box, and an array of bi-amped bass-fill drivers to complement the OB rolloff. If the OB thing doesn't pan out, I'll revert to the more traditional big box, but I'm hoping I won't have to do that.
