Retsel said:
I'm with you - to a certain point.
First of all we have to gain sort of feeling about CSD *without* Cassandra yelling "its no science you do!" - as it certainly isn't used enough to having found a consensus among people.
And - its not that much used as it tells some "truth" about what's happening *after*.
I find it interesting to look at things that are not completely determined in its meaning - this basically is the curiosity / obsession of the scientist as well as of the busybody...
Low decay in CSD is certainly high on the list as otherwise we better would craft bells or something.
*But* if we look at the decay "generated" by radiating areas that are not infinitely small (point source) due to filters that optimise IR at a certain point – well *that* could become interesting!
Michael
gedlee said:
Bummer.
(..I understood the contribution of the waveguide and primary resonance for the driver, I just thought there might have been an additional contribution.)
The result doesn't look good to me. It isn't bad, (and certainly far better than most large sd drivers up that high in freq.), but it isn't good either. At least in this respect, most *good* tweeters will exceed this level of performance. (..you can toggle through Zaph's tweeter "mishmash" for comparisons.)
Since my OX is being gored here, as an example of something that introduces more "distortion" that is pleasing, let me point you to a blink comparison of the same aluminum cone, treated and untreated, by John K himself.
I trust Johns work and all of you should wander over to this particular spot and blink the results. Please look carefully at the resonant node and how it decays, not that it takes the same time, but that the EnABL'd driver shows the actual ringing of the major node at 9 kHz.
http://www.diyaudio.com/forums/showthread.php?postid=1461231#post1461231
or here http://planet10-hifi.com/johnK-test/
For the rest of the information in these two plots, interesting as it is, they provide me with NO information that would be useful in determining what else to do to these drivers to remove the noise in and thus from the driver, that veils information 60 db down, that is also first arrival signal, in real music reproduction circumstances.
For me, a real exploration of what all of the driver altering technologies provide and the need to alter the design of or even create a driver design, I need a much deeper look than just this surface view. Is it possible to perform this sort of layered signal test and extract the data Michael?
Bud
I trust Johns work and all of you should wander over to this particular spot and blink the results. Please look carefully at the resonant node and how it decays, not that it takes the same time, but that the EnABL'd driver shows the actual ringing of the major node at 9 kHz.
http://www.diyaudio.com/forums/showthread.php?postid=1461231#post1461231
or here http://planet10-hifi.com/johnK-test/
For the rest of the information in these two plots, interesting as it is, they provide me with NO information that would be useful in determining what else to do to these drivers to remove the noise in and thus from the driver, that veils information 60 db down, that is also first arrival signal, in real music reproduction circumstances.
For me, a real exploration of what all of the driver altering technologies provide and the need to alter the design of or even create a driver design, I need a much deeper look than just this surface view. Is it possible to perform this sort of layered signal test and extract the data Michael?
Bud
I'm basing my comment on Zaph's on cautions about comparing data in his graphs. Every time he makes some change in how he measures things, he talks about not comparing the old to the new.
I personally think that Zaph focuses too much on the wrong things in tweeter measurements, and not on what really counts. He focuses a lot on things like distortion, smooth response, etc. However, a smooth response can be had with Dr. Geddes tweeter, as is evident from his system response, so while the tweeter, with no crossover, might not be completely smooth, it will be in the end. As for distortion, I don't think the kinds that get so focused on matter. A waveguide and compression tweeter offer far greater sensitivity, and as a result, are capable of far greater output without distortion to a far lower frequency. This allows designs like Dr. Geddes, which can play at realistic levels, without a crossover at 3khz. Then there is what Dr. Geddes will tell you as the most important factor, the power response. Most of those tweeters with better CSD's are direct radiating designs with terrible power responses. As you look more and more off axis the response becomes more and more "yucky". The power response of the waveguide/compression driver allows a very smooth and even repsonse over a wide listening area.
While my experience is just that, my experience, and my views are based on limited experiences, It has been my experience that direct radiating tweeters can not play at realistic levels without audible compression, unless crossed over fairly high. Speakers with crossovers over 2khz have, in my experience, not had as deep or holographic a sound stage as speakers crossed below that point. I can give no technical reason why that is, so while very subjective, its all I got for crossover point. However, power response issue is very important for both qualitative and technical reasons. Again, speakers using direct radiating tweeters have, in my experience, had a less smooth power response when measured over a wide enough angle, and it creates issues in what you hear at the listening position. I really like the ability of controlling smoothing the off-axis response, and avoiding 1st reflections that way. I much prefer the sound of a speaker whose sweet spot is that much wider and smoother as well.
I personally think that Zaph focuses too much on the wrong things in tweeter measurements, and not on what really counts. He focuses a lot on things like distortion, smooth response, etc. However, a smooth response can be had with Dr. Geddes tweeter, as is evident from his system response, so while the tweeter, with no crossover, might not be completely smooth, it will be in the end. As for distortion, I don't think the kinds that get so focused on matter. A waveguide and compression tweeter offer far greater sensitivity, and as a result, are capable of far greater output without distortion to a far lower frequency. This allows designs like Dr. Geddes, which can play at realistic levels, without a crossover at 3khz. Then there is what Dr. Geddes will tell you as the most important factor, the power response. Most of those tweeters with better CSD's are direct radiating designs with terrible power responses. As you look more and more off axis the response becomes more and more "yucky". The power response of the waveguide/compression driver allows a very smooth and even repsonse over a wide listening area.
While my experience is just that, my experience, and my views are based on limited experiences, It has been my experience that direct radiating tweeters can not play at realistic levels without audible compression, unless crossed over fairly high. Speakers with crossovers over 2khz have, in my experience, not had as deep or holographic a sound stage as speakers crossed below that point. I can give no technical reason why that is, so while very subjective, its all I got for crossover point. However, power response issue is very important for both qualitative and technical reasons. Again, speakers using direct radiating tweeters have, in my experience, had a less smooth power response when measured over a wide enough angle, and it creates issues in what you hear at the listening position. I really like the ability of controlling smoothing the off-axis response, and avoiding 1st reflections that way. I much prefer the sound of a speaker whose sweet spot is that much wider and smoother as well.
pjpoes:
While I agree power response, power handling are important, I would not dispute Zaph measurements of distortion and smoothness.
A very flat response between 1k-8kHz is extremely important. While some people suggest that power response is more important, in my view, due to a plenty of early reflections in most rooms, the pursuit of even power response will end up in flat response within the first 10-20ms. Flat response is the goal.
I have a Dynaudio Esotar T330D, a legendary soft dome tweeter with 94dB sensitivity and is capable of low distortion with 1000W transient. It was the top of the range and the most expensive tweeter made by Dynaudio. It is terribly difficult to work with due to its non-flat response. I spent 6 months working on it with passive XO with no good result. In the end, I was forced to use 3 opamps to shape the response to be very flat. I only installed the circuit yesterday, and believe that I have now finally made it working. Using 3 opamps to shape it to be ruler flat is not funny. But I have no other choice, and it is now a revelation (much better sonically, even with the use of opamps). Should it be as flat as the SS6600 the 3 opamps can be saved. Everything else in my XO/EQ is so simple so the re-shaping of the flat response consists of more than half of the circuit.
There are numerous factors determining the sound quality of the speakers. If I select the 1st, most important factor, I would say it is the flat response.
Regards,
Bill
While I agree power response, power handling are important, I would not dispute Zaph measurements of distortion and smoothness.
A very flat response between 1k-8kHz is extremely important. While some people suggest that power response is more important, in my view, due to a plenty of early reflections in most rooms, the pursuit of even power response will end up in flat response within the first 10-20ms. Flat response is the goal.
I have a Dynaudio Esotar T330D, a legendary soft dome tweeter with 94dB sensitivity and is capable of low distortion with 1000W transient. It was the top of the range and the most expensive tweeter made by Dynaudio. It is terribly difficult to work with due to its non-flat response. I spent 6 months working on it with passive XO with no good result. In the end, I was forced to use 3 opamps to shape the response to be very flat. I only installed the circuit yesterday, and believe that I have now finally made it working. Using 3 opamps to shape it to be ruler flat is not funny. But I have no other choice, and it is now a revelation (much better sonically, even with the use of opamps). Should it be as flat as the SS6600 the 3 opamps can be saved. Everything else in my XO/EQ is so simple so the re-shaping of the flat response consists of more than half of the circuit.
There are numerous factors determining the sound quality of the speakers. If I select the 1st, most important factor, I would say it is the flat response.
Regards,
Bill
The sample rate must be pretty low during the original measurement.mige0 said:
Could do ?but I doubt there is enough information left in the result to allow for serious discussion.
1ms is only very few samples ?magnifying by 10 or so - is reading tea leaves (even more as already)
Michael
But you currently have the slice shift set to "2". Setting that to "1" would give more resolution.
gedlee said:
I think I already have.
Even off-axis a tweeter can perform well (CSD included).
CD isn't a requirement for good sound, nor is a compression driver with waveguide. As for power/thermal compression - it all depends on how loud the listener likes it.
I would however agree that your design can provide advantages, and depending on the preference of the listener - possibly enough advantages to opt. for such a design.. BUT this still doesn't void the fact that at lower levels a good tweeter outperforms the solution you presented, (at least on axis), with respect to it's overall decay character. (..of course I'd also not treat this one aspect as a "nail in the coffin", but rather one modest design limitation to factor-in.)
I would also agree with pjpoes that for "realistic" spl's, at listening distances greater than 2 meters, that many/most designs utilizing single "average" tweeters (operated lower in freq.) sound compressed. On the other hand I don't think many critical music listeners listen at "realistic" levels (except for the occasional short term dynamic).
Some people really don't like the "feel" thing. But I would like to give one example:
There are many explanations on how a person balances himself in varous activities, and you can analyze it to the end. But if you don't get up and do it to experience the "feel", you will never be able to do that particular act.
There are many explanations on how a person balances himself in varous activities, and you can analyze it to the end. But if you don't get up and do it to experience the "feel", you will never be able to do that particular act.
soongsc said:
He's right and you're right. The CSD is created by software making successive FFTs on the impulse response, successively shifting the start time marker. But the key to this is as he said, it is not simply successive FR curves, it is the decay after the signal is removed. Any particular FR curve in the CSD is meaningless without being in the context of the CSD. It's the set that counts, not any single curve.
That said, I also see very little use in CSDs for much more than detecting resonance ridges and even those is observable in the SPL response.
Dave
ScottG said:
I would disgaree on both counts.
Even Linkwitz now agrees that CD is a highly desirable trait. You seem to be alone in this view.
Thermal compression is always present and it may actually be a constant percentage that is independent of level.
ScottG said:
Uh, maybe that's because most speakers sound so bad at higher levels. Most people who listen to my speakers comment on how unoppressive the higher volumes are. They sound just as clean and crisp as the lower levels, just louder. When this doesn't happen, people just don't turn it up.
gedlee said:
1. ..and the sky is "blue".
2. A "highly desirable trait" and a "requirement for good sound" are not the same thing. Moreover I don't necessarily disagree that CD isn't a worthwhile pursuit.
3. correction: add-in *Significant* to thermal compression on my previous post.
4. No. It may be *a* reason, and occasionally it maybe *the* reason. Overall from a broad spectrum of listeners (casual and critical) that I've encountered, it's simply that most people don't like to listen that loud to music that often - especially the compressed drek that epitomizes much of the music on offer today. The possible exception to this is a critical listener listening to "classical" music with very wide dynamic ranges (.."realistic" when compared to live performances). Even then I believe that most listener's tend to listen a little less loud than what you would find in a typical concert hall, and because of the still fairly low overall average to the music - the effects of compression are overall low.