What is often overlooked is the following:
If only one driver device of a certainly model is tested, series spread (scattering) can falsify the test result. I remember that a test magazine (I think VOICE-COIL) always examined several driver devices of the same model in order to better rule out such undesirable effects.
In general - with titanium, beryllium and ceramic material, this risk of spread is always significantly higher than with fabric phenol, cevlar or carbon fiber - at least according to a developer from this company:
http://www.kurtmueller.com/index.php?id=25
OTOH, the perception of perfect sound quality is currently completely different among young people than among those who are familiar with it from the 60s, 70s and 80s.
Irrespective of this - lately I've been getting the impression that more and more detailed knowledge from the past is being lost and is therefore no longer included in new designs, because the old, experienced guys are more and more missing from the old, well-known manufacturers due to age and state of health - such as e.g. also at Celestion, Klipsch and some other (I don't know anything about speakers because one usually can't build drivers in a diy process. But I'm pretty sure that one would have achieved much better results with the Celestion Axi2050 if one had implemented a coating of diaphragm like the old Audax super tweeter TW074, which also use a metal diaphragm - the advantage of the improved heat dissipation would of course be lost, which is of course a disadvantage for PA applications).
If only one driver device of a certainly model is tested, series spread (scattering) can falsify the test result. I remember that a test magazine (I think VOICE-COIL) always examined several driver devices of the same model in order to better rule out such undesirable effects.
In general - with titanium, beryllium and ceramic material, this risk of spread is always significantly higher than with fabric phenol, cevlar or carbon fiber - at least according to a developer from this company:
http://www.kurtmueller.com/index.php?id=25
OTOH, the perception of perfect sound quality is currently completely different among young people than among those who are familiar with it from the 60s, 70s and 80s.
Irrespective of this - lately I've been getting the impression that more and more detailed knowledge from the past is being lost and is therefore no longer included in new designs, because the old, experienced guys are more and more missing from the old, well-known manufacturers due to age and state of health - such as e.g. also at Celestion, Klipsch and some other (I don't know anything about speakers because one usually can't build drivers in a diy process. But I'm pretty sure that one would have achieved much better results with the Celestion Axi2050 if one had implemented a coating of diaphragm like the old Audax super tweeter TW074, which also use a metal diaphragm - the advantage of the improved heat dissipation would of course be lost, which is of course a disadvantage for PA applications).
Last edited:
Am I wrong or can REW now also do IMD measurements (both dual, tri tone and multitone) through the RTA module? I am doing them with REW, since I have been using the mac it is inconvenient for me to use ARTA even though it continues to have several arrows in its bow (STEPS for example).
Yes, as I have found since writing that post you quoted. When I first looked at the multiple tone generator when it first appeared perhaps 3 years ago, there were few options other than dual tone. In September 2021, more options became available for mutitone signals. I have used the multitone output recently, and it seems to work quite well, although I need to experiment with it more to produce plots that I can say "here is the modulation distortion performance".
I actually think the measurement is going to turn a few heads, since it seems to correlate to subjective listening performance.
Note that I originally asked for an "intermodulation distortion" test capability within REW a couple of years ago, but no response was made since then. So I am pleasantly surprised.
Chris
Hi
I am not very experienced with IMD measures, I have also recently started using them with some consistency. Although I have studied how to perform them with a minimum of proficiency I still need to gain experience on how to interpret them. The graphs provided by REW are less easy to interpret than the Kippel graph you included in the post you sent me. Nevertheless, I believe that the graphs you can produce with REW contain all the necessary information.
See here for some examples.
Franco
https://www.diyaudio.com/community/threads/beyma-6mcf200nd-measurements.397781/page-2
I am not very experienced with IMD measures, I have also recently started using them with some consistency. Although I have studied how to perform them with a minimum of proficiency I still need to gain experience on how to interpret them. The graphs provided by REW are less easy to interpret than the Kippel graph you included in the post you sent me. Nevertheless, I believe that the graphs you can produce with REW contain all the necessary information.
See here for some examples.
Franco
https://www.diyaudio.com/community/threads/beyma-6mcf200nd-measurements.397781/page-2
Note that J. Crowe had to try several different instrumentation microphones to get acceptable multitone IM performance. The more popular USB mikes fell down rather badly. This is described in more detail in his blog posts, which are worthy of close reading.
You more commonly see multitone IM measurements used for testing electronics, where it can expose circuit weaknesses that are hidden from single-tone harmonic distortion measurements. There are circuits that look very clean from a harmonic-distortion perspective that fall on their face in certain portions of the spectrum ... typically, transformer-coupled circuits can get in trouble below 100 Hz, and slewing distortion can be visible above 2~5 kHz. Class AB crossover distortion is very apparent on multitone IM distortion measurements. Although I'm not a big fan of Audio Science Review and their single-number chasing, the multitone IM tests are by far the most valuable part of their test protocol.
We're seeing the same things in loudspeaker drivers, which might be linear in some parts of the spectrum but get dirty in other parts. This is particularly significant since IM distortion dominates loudspeaker nonlinearity (low frequencies cross-modulate with higher frequencies, which leads to spectral "thickening"). In compression drivers, problems with the surround become visible, as well as assorted noises from leaks.
You more commonly see multitone IM measurements used for testing electronics, where it can expose circuit weaknesses that are hidden from single-tone harmonic distortion measurements. There are circuits that look very clean from a harmonic-distortion perspective that fall on their face in certain portions of the spectrum ... typically, transformer-coupled circuits can get in trouble below 100 Hz, and slewing distortion can be visible above 2~5 kHz. Class AB crossover distortion is very apparent on multitone IM distortion measurements. Although I'm not a big fan of Audio Science Review and their single-number chasing, the multitone IM tests are by far the most valuable part of their test protocol.
We're seeing the same things in loudspeaker drivers, which might be linear in some parts of the spectrum but get dirty in other parts. This is particularly significant since IM distortion dominates loudspeaker nonlinearity (low frequencies cross-modulate with higher frequencies, which leads to spectral "thickening"). In compression drivers, problems with the surround become visible, as well as assorted noises from leaks.
Hi Lynn, I know the test done by J Crowe which I also follow, if I remember correctly the microphone Joseph selected as the best (ACO Pacific) has a cost has a cost 10 times more than the one I use (UMIK1). Unfortunately, the UMIK has not been tested so I do not have a reference compared to those tested by Joseph. I have to say one thing: on some particularly high performing speakers at the IMD measurement I get measurements with a rather low level of distortion, which, if the microphone had an inherent IMD level like the ones Joseph tested, could not happen.
It would be interesting to find a microphone with sufficient performance for IMD measurements without spending more on the microphone than on the speaker, after all for many of us it is a hobby and an expense like that for the ACO Pacific is hardly sustainable.
Franco
It would be interesting to find a microphone with sufficient performance for IMD measurements without spending more on the microphone than on the speaker, after all for many of us it is a hobby and an expense like that for the ACO Pacific is hardly sustainable.
Franco
Before we write off the Axi2050... Is it just me or does something look a little odd with some of Mr Crowe's older measurements - on many of the drivers he has measured, the 2nd, 3rd and 4th harmonics all seem to be at the same level across a sweep. See eg JA6681B measurement at
https://josephcrowe.com/blogs/news/yamaha-ja6681b-compression-driver
Not what I'd expect - the famous ETF shootout is much more plausible to me.
For the axi2050,
https://josephcrowe.com/blogs/news/celestion-axi2050
the % figures on the harmonic distortion sweep charts don't make sense... typo wrt each of the 3 SPL levels?
This looks more plausible to me:
https://forums.melaudia.net/showthread.php?tid=11647
I'd be surprised to see such high IMD from a driver with low THD given the mathematical link between them.
Unit variation / diaphragm alignment also a big question - and to his credit, JC has mentioned and revisited drivers accordingly.
I could easily be wrong but it would be good to double check some of this stuff.
https://josephcrowe.com/blogs/news/yamaha-ja6681b-compression-driver
Not what I'd expect - the famous ETF shootout is much more plausible to me.
For the axi2050,
https://josephcrowe.com/blogs/news/celestion-axi2050
the % figures on the harmonic distortion sweep charts don't make sense... typo wrt each of the 3 SPL levels?
This looks more plausible to me:
https://forums.melaudia.net/showthread.php?tid=11647
I'd be surprised to see such high IMD from a driver with low THD given the mathematical link between them.
Unit variation / diaphragm alignment also a big question - and to his credit, JC has mentioned and revisited drivers accordingly.
I could easily be wrong but it would be good to double check some of this stuff.
From Joseph what he says in a youtube video devoted to microphones he should have purchased the ACO Pacific mic (which I tend to trust, given the high quality of the mic and since he uses ARTA and a very good mic preamp) about nine months ago.
Often when comparing measurements on the same speaker made by different people you see even very different results due probably to different measurement conditions. For example if you measure distortion indoors and not outdoors and you change the distance between the microphone and speaker the distortion values can also change a lot. I would like to hear the opinion of someone more knowledgeable explaining why the axi2050 can perform so differently when measured by two different people (moreover it would seem with the same software!). The effect of the horn in my experience is not so radical as to distort the measurements in this way.
Often when comparing measurements on the same speaker made by different people you see even very different results due probably to different measurement conditions. For example if you measure distortion indoors and not outdoors and you change the distance between the microphone and speaker the distortion values can also change a lot. I would like to hear the opinion of someone more knowledgeable explaining why the axi2050 can perform so differently when measured by two different people (moreover it would seem with the same software!). The effect of the horn in my experience is not so radical as to distort the measurements in this way.
Hi
Back to the original content of the forum: to the drivers already listed in my opinion two other drivers should be added, one brand new and very promising:
B&C DCX354
https://bcspeakers.com/en/products/coaxials-hf/1-4/8/dcx354-8
and one slightly less new B&C DCX 50
https://bcspeakers.com/en/products/coaxials-hf/2-0/8/DCX50-8
The DCX 354 although less performing from the point of view of SPL and power handling was developed from the already very performing DCX464 but with some added more recent knowledge, I look forward to reading some measurements about it or being able to listen and measure it in person.
Back to the original content of the forum: to the drivers already listed in my opinion two other drivers should be added, one brand new and very promising:
B&C DCX354
https://bcspeakers.com/en/products/coaxials-hf/1-4/8/dcx354-8
and one slightly less new B&C DCX 50
https://bcspeakers.com/en/products/coaxials-hf/2-0/8/DCX50-8
The DCX 354 although less performing from the point of view of SPL and power handling was developed from the already very performing DCX464 but with some added more recent knowledge, I look forward to reading some measurements about it or being able to listen and measure it in person.
I guess the DCX 354 would need quite some EQing effort. And I don't know whether it has a lot of advantages over an ordinary 1.4" driver at (even very loud) home listening levels. The fact that it can go low is an advantage admittedly but does it still have any advantage when crossed at 1 kHz or higher ? They are too expensive to just buy them out of curiosity so I wait for reviews......
Regards
Charles
Regards
Charles
I am feeling like there is kind of an elephant in the room, or even a few to be honest.
First off, people look at impulse response, frequency response, harmonic distortion and IMD. Still, drivers who measure fairly identical, or one driver that measures significantly better than another, seem to be outperformed in listening tests quite often. I believe we do not see as clear as pattern of this in direct radiating drivers.
We know about some factors that do play a significant role. For example, since the diaphragm break up stays behind the phase plug and the cavity resonances, it is really hard to see if this affects the performance of the driver at these specific parameters. That is not the same as saying that "yes, we know of a different measurement method that reveals this clearly", but we kind of know at least some of the things going on.
We also know there is a mathematical model to determine the optimum back acoustic impedance for a given driver and LF extension. We have known this since at least the mid 30's when they extended a 2 inch WE driver down to 34 Hz in a giant horn just by tuning the back chamber. I do not know, however, what it sounded like, and very few people do.
From time to time, we see people talking about drivers properties (and here is the elephant in the room) just by experience, but unfortunately being unable to connect the dots of the measurements in a way that we actually see a clear pattern as to why one driver can sound so different than another while seemingly being very nearly identical. Especially low end performance seem to vary a lot. Off course, we have the bias factor here and we know it can be quite big. But still, it is like people do talk about something connected to horns and drivers that is very real, easily reproducable and still without a clear way to connect these things to a scientific approach.
Is it just me, or do we have some uncovered ground in terms of figuring out the real link between compression driver "sound type" and what really goes on inside the driver from an objective point of view? For quite some time, I had my bet on the motor, but after checking how well the popular TAD drivers are designed compared to some modern competitors, I just had to remove that part from the equation.
What do you think guys, am I missing something obvious?
First off, people look at impulse response, frequency response, harmonic distortion and IMD. Still, drivers who measure fairly identical, or one driver that measures significantly better than another, seem to be outperformed in listening tests quite often. I believe we do not see as clear as pattern of this in direct radiating drivers.
We know about some factors that do play a significant role. For example, since the diaphragm break up stays behind the phase plug and the cavity resonances, it is really hard to see if this affects the performance of the driver at these specific parameters. That is not the same as saying that "yes, we know of a different measurement method that reveals this clearly", but we kind of know at least some of the things going on.
We also know there is a mathematical model to determine the optimum back acoustic impedance for a given driver and LF extension. We have known this since at least the mid 30's when they extended a 2 inch WE driver down to 34 Hz in a giant horn just by tuning the back chamber. I do not know, however, what it sounded like, and very few people do.
From time to time, we see people talking about drivers properties (and here is the elephant in the room) just by experience, but unfortunately being unable to connect the dots of the measurements in a way that we actually see a clear pattern as to why one driver can sound so different than another while seemingly being very nearly identical. Especially low end performance seem to vary a lot. Off course, we have the bias factor here and we know it can be quite big. But still, it is like people do talk about something connected to horns and drivers that is very real, easily reproducable and still without a clear way to connect these things to a scientific approach.
Is it just me, or do we have some uncovered ground in terms of figuring out the real link between compression driver "sound type" and what really goes on inside the driver from an objective point of view? For quite some time, I had my bet on the motor, but after checking how well the popular TAD drivers are designed compared to some modern competitors, I just had to remove that part from the equation.
What do you think guys, am I missing something obvious?
I think that when you look at his objective, the price of microphones starts to go up exponentially (i.e., $1500-$3000 USD, etc.). Instead, I believe it is much easier to raise the drive level by 10 dB or so to see the modulation distortion trends with more affordable microphones. Modulation distortion levels are said to be a strong function of drive level (at least proportional to diaphragm displacement--and probably more sensitive to increasing levels) according to PWK, et al. Trying to measure at too low a level (as I think the referenced "blog entry" is attempting) is not the way I'd approach the issue.
I do think that separate thread on:
(1) measuring modulation distortion and
2) discussion of where the actual audibility thresholds are,
...is probably a better approach. Arguing about Troy Crowe's blog entries isn't a very good use of time here, I believe. There are many more issues to discuss on that subject. Suffice it to say that the Axi2050 is reviewed quite well by everyone else that I've read, and in my direct experience it is a much better driver than was apparently discussed there. No need to focus on that one author.
Chris
I agree: if the crossover point is around 1000Hz there is no need to consider a driver like this that competes with other products such as AXI or BMS coaxials.
Franco
I agree. Much eludes us, and although I for my work am attached to a scientific approach therefore to objectivity rather than subjectivity we must admit that our theoretical and measurement tools have obvious limitations that therefore make listening and subjective approaches very useful. Our ear even with its limitations, taking into consideration that we do not build scientific instruments but rather objects dedicated to pleasure, remain in my opinion those who should have the last word. We should not overestimate our scientific-technical abilities: we must admit that we still do not know how to connect all the dots to draw the line. In my case certainly not only because of lack of good measuring instruments but also because of limited competence.
Chris,
I had the same thought about the low level measurements. Something about it doesn't make sense. Intuitively.
However, I am not prepared to dispute the findings from Troy no matter how many 'other reviewers' have positive impressions from the driver. The measurements are real and cannot be brushed aside IMHO. One thing I would say for sure, his idea of pushing the crossover freq to 300Hz on a tiny biradial horn was ill conceived. I would probably not go < 600Hz on that. I think the K402 horn may be a much better match.
and also don't forget that mesurement must be taken under 1 Wattt , becausein home use we don't gererally listen over 110db
if you take a look at all the brands advertising they all have the best compression drive you could get , but at the end only a few are models are higly praised by the pro or hifi audio consumer . only time will tell ..
(Sorry for the following, which I believe is necessary--in terms of its length.)
This is ultimately true. However, how do we calibrate our individual hearing capabilities so that others can rely on them? In this thread alone, instances of disbelief in someone else's hearing capabilities (or probably more to the point...what they like to hear) are apparent.
To illustrate the problem, there was one instance where I listened to a variety of bass bins (different numbers of woofers in reflex boxes) in the same listening room, as compared to a horn-loaded bass bin, all using the same top end (K-402 with modified B&C D75, carefully EQed to flat response using an anechoic chamber). The bass bins were initially EQed anechoically, then dialed into the room to account for boundary gain, etc. There were approximately a dozen guys in the room, all owners or potential owners of these type of loudspeaker systems throughout all the trials. They all heard what I heard and they all chose their favorite sound.
However, while almost all the others chose the reflex boxes (four 15" woofers was the biggest, and most liked example), my own preference was for the horn loaded bass bin (the one used in the first-gen Jubilee). Why? Lower modulation distortion and more realistic dynamics. Everyone else save one or two others went for the "impact". This was a problem that couldn't be "explained away".
The way I looked at it, the others liked the sound of jukeboxes and rock-style amplified deep bass concerts and home stereo systems that they probably only listened to as children and adults. My background was live acoustic instrumentation (mostly classical, but also other acoustic genres), not recorded.
[One parent is a classically trained organist that I was obliged to listen to in large churches and auditoria settings during extended pipe organ practice sessions as a child. I also imprinted on the sound of the acoustic instruments during piano accompaniment of classically trained professional and amateur musicians. These were my baseline--not jukeboxes and garage rock bands.]
There were also other interesting instances: some in the room couldn't hear the difference between 2" titanium dome drivers that chatter above 14 kHz, and others that could easily hear it and its absence using beryllium diaphragm or dual ring radiator drivers (properly dialed in).
Note especially that I'm talking about those individuals that have spent their professional lives in the recording and/or sound reinforcement industry--who I would not trust to pick out truly neutral or perhaps "more realistic sounding" loudspeaker types, but rather, their own personal standard of "acceptable sound quality".
So there is the fundamental problem.
The way I see it, we really can't talk to each other and really understand what each other are really saying unless we also show measurements, because with measurements, we have a chance to gauge whether the other guy hears or likes to hear what we personally like--or not (i.e., not 100%, but with better certainty).
As the quality level of today's loudspeakers and recordings keeps increasing, we need measures that correlate to subjective preferences. I believe that modulation distortion measurements (which would have easily knocked out the differences in reflex bass bins vs. horn loaded) is one such measure, as well as flat phase and excess group delay response in the bass bin example.
I see "circle of confusion" including the listeners themselves as the state of the art gets better and better, regardless of Floyd Toole's insistence to the contrary. Flat on-axis amplitude response (or even power response) are no longer sufficient to distinguish between loudspeakers, room acoustics, electronics, and "personal taste" (biases).
Chris
Last edited:
One experience that has stuck with me as a reference when assessing any loudspeaker system was being asked to sit next to the conductor of the BBC Philharmonic Orchestra, during sound check in case any of the mics were causing a problem for the musicians. That was a real eye opener to not just the audio experience but the physical sensations of that many acoustic instruments played in unison.