Hi,
Recently Stereophile magazine published a summary of the Recommended Components for 2012. Go to this site for the recommended Class A Full Range Loudspeakers;
2012 Recommended Components Loudspeakers | Stereophile.com
All these speakers have one thing in common, all of them are very expensive, most of them are
priced in excess of US$20,000 per pair retail price, beyond the reach of most consumers esp. DIYers. Can DIYers learn from the reviews from these speakers and apply them in their design? Hence I am writing this thread.
I expect most of these speakers are designed to meet the audio needs of "majority" of consumers who can afford to buy these speakers, and may not reflect the "taste" of DIYers here. It should be remembered that there are many ways to design a pair of great sounding speakers to suit individual taste but sound "ordinary" to others.
Obviously there are many factors governing the "sound" of these speakers, e.g., Freq.
Response, Drivers' phases/impedance/sensitivity/material
(paper,polypropylene,etc.)/directivity/power response, Enclosure, etc. There are too many
variables here to consider on which parameters are more important in determining the speakers'
sound. Here I will focus on two very important measurable parameters, Freq. response, and Phase/Impedance. During the course of speakers reviews from Stereophile, John Atkinson had produced many graphs on all these speakers. I would expect some degree of consistencies coming from the measurements derived from these graphs since they were measured by the same person.
Find attached a summary of Freq. Resp and Impedance/Phase graphs from these speakers.
I would like to made the following observations:
Frequency Response Curve
---------------------------------
None of the Freq. response curves are "ruler" flat. There are still DIYer designers that design
their speakers to be ruler flat. One important feature which worth noting is a 5dB Bass bump around
70-100Hz. This is a common feature among these speakers. Furthermore the all time
classics speakers Roger LS3/5A has this feature too. Besides this feature the speakers are "relatively" flat(exceptions are from Verity and Wilson). Relatively flat is an important feature and has been proven to be a prerequisite by a number of technical papers published by AES. "Tilting" the response curves upwards and downwards(around the tweeter end) will determine of the overall characteristics of the speakers from individual manufacturers.
Impedance and Phase Curve
----------------------------------
There have many discussions on minimising the drivers phase especially around the x-over
frequencies. It is important to note that two manufactures(B&W and Kef) have the driver phases exceeding +/- 45 deg at around x-over. Is phase minimising that important?
The drivers impedances swing erratically across the freq. bandwidth, this is not surprising,
what is surprising is that are no attempts to level them.
Invitation
------------
There are many knowledgeable and competent speakers designers here may like to make further comments on these curves/graphs that I have summarised here esp., Do's and Don't when designing speakers to meet the freq/phase response curves. This will benefit new and current DIYers who are learning how to design their first set or subsequent pair of speakers.
Please discuss.
Recently Stereophile magazine published a summary of the Recommended Components for 2012. Go to this site for the recommended Class A Full Range Loudspeakers;
2012 Recommended Components Loudspeakers | Stereophile.com
All these speakers have one thing in common, all of them are very expensive, most of them are
priced in excess of US$20,000 per pair retail price, beyond the reach of most consumers esp. DIYers. Can DIYers learn from the reviews from these speakers and apply them in their design? Hence I am writing this thread.
I expect most of these speakers are designed to meet the audio needs of "majority" of consumers who can afford to buy these speakers, and may not reflect the "taste" of DIYers here. It should be remembered that there are many ways to design a pair of great sounding speakers to suit individual taste but sound "ordinary" to others.
Obviously there are many factors governing the "sound" of these speakers, e.g., Freq.
Response, Drivers' phases/impedance/sensitivity/material
(paper,polypropylene,etc.)/directivity/power response, Enclosure, etc. There are too many
variables here to consider on which parameters are more important in determining the speakers'
sound. Here I will focus on two very important measurable parameters, Freq. response, and Phase/Impedance. During the course of speakers reviews from Stereophile, John Atkinson had produced many graphs on all these speakers. I would expect some degree of consistencies coming from the measurements derived from these graphs since they were measured by the same person.
Find attached a summary of Freq. Resp and Impedance/Phase graphs from these speakers.
I would like to made the following observations:
Frequency Response Curve
---------------------------------
None of the Freq. response curves are "ruler" flat. There are still DIYer designers that design
their speakers to be ruler flat. One important feature which worth noting is a 5dB Bass bump around
70-100Hz. This is a common feature among these speakers. Furthermore the all time
classics speakers Roger LS3/5A has this feature too. Besides this feature the speakers are "relatively" flat(exceptions are from Verity and Wilson). Relatively flat is an important feature and has been proven to be a prerequisite by a number of technical papers published by AES. "Tilting" the response curves upwards and downwards(around the tweeter end) will determine of the overall characteristics of the speakers from individual manufacturers.
Impedance and Phase Curve
----------------------------------
There have many discussions on minimising the drivers phase especially around the x-over
frequencies. It is important to note that two manufactures(B&W and Kef) have the driver phases exceeding +/- 45 deg at around x-over. Is phase minimising that important?
The drivers impedances swing erratically across the freq. bandwidth, this is not surprising,
what is surprising is that are no attempts to level them.
Invitation
------------
There are many knowledgeable and competent speakers designers here may like to make further comments on these curves/graphs that I have summarised here esp., Do's and Don't when designing speakers to meet the freq/phase response curves. This will benefit new and current DIYers who are learning how to design their first set or subsequent pair of speakers.
Please discuss.
Attachments
I've heard only one of the A-list speakers, and it's horrible....
The bass rise/hump on Stereophile measurements should be there, because they measure the woofers in the near-field. But it worries me that the bass extension of all those speakers are more or less ...well .. not present. That rising bass should continue down to 20 Hz I think
The bass rise/hump on Stereophile measurements should be there, because they measure the woofers in the near-field. But it worries me that the bass extension of all those speakers are more or less ...well .. not present. That rising bass should continue down to 20 Hz I think
Last edited:
I think the low frequency response (say below 150 Hz) is unusable to compare these speakers with.
The big bump at 60 Hz is quite surely an undamped room mode, and indicates that their listening room is not too big
Seems that the Vandersteen active bass has been measured in near field, and it shows that this speaker has a lot of out out at low freq.
Looking at the Impedance curves, it is also clear that most of the other speakers should have a lot of low end ... many of them seems to be tunes for a little over 20 Hz.
I therefore also think the lack of bass, which will probably also be what you hear in their room, is due to a low pressure zone at the measurement /listening position (if these are the same).
That the peak at 60 Hz is "only" 5 db is actually quite good
From 150 ad up I think things are more comparable. Here you see frequency responses of say +- 2.5 db .... that is actually also quite respectable.
Thanks for putting the measurements together ... quite something to think over
Best regards Baldin
The big bump at 60 Hz is quite surely an undamped room mode, and indicates that their listening room is not too big
Seems that the Vandersteen active bass has been measured in near field, and it shows that this speaker has a lot of out out at low freq.
Looking at the Impedance curves, it is also clear that most of the other speakers should have a lot of low end ... many of them seems to be tunes for a little over 20 Hz.
I therefore also think the lack of bass, which will probably also be what you hear in their room, is due to a low pressure zone at the measurement /listening position (if these are the same).
That the peak at 60 Hz is "only" 5 db is actually quite good
From 150 ad up I think things are more comparable. Here you see frequency responses of say +- 2.5 db .... that is actually also quite respectable.
Thanks for putting the measurements together ... quite something to think over
Best regards Baldin
Nothing on power distribution or limits, in room power response, controlled or constant directivity, baffle step, zobel compensation, if any. Furthermore there's no mention made of driver polar response or the distribution of harmonic distortion products, intermodulation distortion or comb filter effects at crossover points, etc., etc.
Frequency and phase are but the first page in a large volume.
Frequency and phase are but the first page in a large volume.
John Atkinson's measurements are quite thorough, considering doesn't have an anechoic chamber at his disposal. His measurements usually include:
- Sensitivity estimation
- Impedance
- Cabinet vibration measurements
- Near-field measurements for low frequencies
- Gated measurements for mid and high frequencies.
- Off-axis measurement to 90 degrees horizontal and 15 degrees vertical
- In-room spatial average
- Step response
- CSD based on gated measurement
The greatest uncertainty is in the way he adds the near-field curves and the gated far-field curve together. Speakers that are flat under anechoic conditions show a mid-bass bump in Atkinson's measurements due to his measurement technique. He himself acknowledges this uncertainty. I'm not a great fan of the enclosure vibration measurements, because I just don't know how to reliably correlate them with sound quality. If I were him I wouldn't include the step response either - they don't show much relevant information and are easily misinterpreted by lay readers.
John Atkinson is a self-proclaimed 'believer' when it comes to many aspects of sound reproduction, but when it comes to speakers he does seem to know what he's talking about.
See this article:
Measuring Loudspeakers, Part One | Stereophile.com
When it comes to the A list, all speakers are very expensive and hence are likely to look, feel and thus actually sound better than less expensive ones. It's just the power of suggestion doing its magic. Based on the measurements both the Revel and KEF look to represent the state of the art of conventional consumer loudspeaker design. If I had that kind of money to spend on speakers I wouldn't buy any of the others. Better performance is available for considerably less money.
EDIT: I've had a closer look at the list. The Revel Studio 2 seems similarly good as the Salon 2. Atkinson's frequency response measurements of the Vivid Giya don't look as smooth as those measured by Soundstage magazine. Soundstage does have an anechoic chamber at their disposal (the one of the Canadian National Research Council), so I'm tempted to take their measurements over Atkinson's. Usually there is reasonably good correspondence between Atkinson's measurements and those made at the NRC, though.
- Sensitivity estimation
- Impedance
- Cabinet vibration measurements
- Near-field measurements for low frequencies
- Gated measurements for mid and high frequencies.
- Off-axis measurement to 90 degrees horizontal and 15 degrees vertical
- In-room spatial average
- Step response
- CSD based on gated measurement
The greatest uncertainty is in the way he adds the near-field curves and the gated far-field curve together. Speakers that are flat under anechoic conditions show a mid-bass bump in Atkinson's measurements due to his measurement technique. He himself acknowledges this uncertainty. I'm not a great fan of the enclosure vibration measurements, because I just don't know how to reliably correlate them with sound quality. If I were him I wouldn't include the step response either - they don't show much relevant information and are easily misinterpreted by lay readers.
John Atkinson is a self-proclaimed 'believer' when it comes to many aspects of sound reproduction, but when it comes to speakers he does seem to know what he's talking about.
See this article:
Measuring Loudspeakers, Part One | Stereophile.com
When it comes to the A list, all speakers are very expensive and hence are likely to look, feel and thus actually sound better than less expensive ones. It's just the power of suggestion doing its magic. Based on the measurements both the Revel and KEF look to represent the state of the art of conventional consumer loudspeaker design. If I had that kind of money to spend on speakers I wouldn't buy any of the others. Better performance is available for considerably less money.
EDIT: I've had a closer look at the list. The Revel Studio 2 seems similarly good as the Salon 2. Atkinson's frequency response measurements of the Vivid Giya don't look as smooth as those measured by Soundstage magazine. Soundstage does have an anechoic chamber at their disposal (the one of the Canadian National Research Council), so I'm tempted to take their measurements over Atkinson's. Usually there is reasonably good correspondence between Atkinson's measurements and those made at the NRC, though.
Last edited:
I have visited a number of the Stereophile reviewers and gone through the review process with products many times. Heres the way it usually goes. You deal with a particular reviewer and set up the system at their house. They do a subjective review based on their impressions and certainly influenced by their biases and preferences. Most manufacturers spend a good deal of time schmoozing and convincing the particular reviewer that your product is uniquely great, that you are the best intentioned, smartest product developers and this particular product is the greatest thing since sliced pizza.
The reviewers vary in their technical knowledge and their ability to withstand the manufacturer's spin. They may have their own biases towards particular makers or design approaches. I never saw or heard of any money changing hands or any other favors other than the long term loan of very expensive gear (outside of the Japanese magazines, another story).
After the review is underway and when their listening impressions are firm the product goes off to John Atkinson for measurements. Now John is a very bright guy and has measured more speakers than many big name designers. He is not a designer but he has a good idea of what is important. Read his AES paper on the subject and it is pretty good, comparing his measurements with the general listening impressions of his reviewers. He also acknowledges me in the paper so he must be a really smart guy.
Depending on which staff reviewer had the product there may be a correlation between measurements and subjective impressions, or there may not. Some of the reviewers are pretty good and some are totaly amateur. Some fell for the Bull of the manufacturer and others saw through it. This creates a big variability between the measurements and the subjective impressions, "sighted bias", Floyd would say. In those cases John will have to tap dance around between what he measured and what the particular reviewer's system appraisal was.
I always find this the most entertaining parts of the reviews.
As to the measurements, anechoic axial frequency response is always the most import factor by far. I agree that Floyd's book is the new bible in this regard and he will tell you that axial response flatness, axial response smoothness, power response smoothness (to reveal the presence of resonances) and bass extension are the most important measurable parameters.
Other factors such as impedance curve flatness or phase angle, impulse response ringing, cabinet measurements and the like are secondary.
Some of those curve look pretty good to me (Sony, KEF, Ariel, Revel) and the others are pretty flawed (e.g. Wilson). In the end the reviewers are being reviewed as much as the speakers.
By the way, the low frequency part of John's curves are taken near field and would therfore represent the 2 pi response. They tend to bump up because most designers design towards a 4 pi flat rather than 2 pi flat curve. This has nothing to do with any in-room resonances.
David S.
The reviewers vary in their technical knowledge and their ability to withstand the manufacturer's spin. They may have their own biases towards particular makers or design approaches. I never saw or heard of any money changing hands or any other favors other than the long term loan of very expensive gear (outside of the Japanese magazines, another story).
After the review is underway and when their listening impressions are firm the product goes off to John Atkinson for measurements. Now John is a very bright guy and has measured more speakers than many big name designers. He is not a designer but he has a good idea of what is important. Read his AES paper on the subject and it is pretty good, comparing his measurements with the general listening impressions of his reviewers. He also acknowledges me in the paper so he must be a really smart guy.
Depending on which staff reviewer had the product there may be a correlation between measurements and subjective impressions, or there may not. Some of the reviewers are pretty good and some are totaly amateur. Some fell for the Bull of the manufacturer and others saw through it. This creates a big variability between the measurements and the subjective impressions, "sighted bias", Floyd would say. In those cases John will have to tap dance around between what he measured and what the particular reviewer's system appraisal was.
I always find this the most entertaining parts of the reviews.
As to the measurements, anechoic axial frequency response is always the most import factor by far. I agree that Floyd's book is the new bible in this regard and he will tell you that axial response flatness, axial response smoothness, power response smoothness (to reveal the presence of resonances) and bass extension are the most important measurable parameters.
Other factors such as impedance curve flatness or phase angle, impulse response ringing, cabinet measurements and the like are secondary.
Some of those curve look pretty good to me (Sony, KEF, Ariel, Revel) and the others are pretty flawed (e.g. Wilson). In the end the reviewers are being reviewed as much as the speakers.
By the way, the low frequency part of John's curves are taken near field and would therfore represent the 2 pi response. They tend to bump up because most designers design towards a 4 pi flat rather than 2 pi flat curve. This has nothing to do with any in-room resonances.
David S.
I used to find the same with our local magazine back when, and some of the reviews were quite good but now and then the reviewer might take more than one paragraph to discuss his new halved tennis balls. The second half of the review may as well have been for a different product.
Had to fix grammar error in title.. 
I always enjoy reading the measurements of all the equipment that passes through John Atkinson's hands. Most of the time I completely ignore the subjective review.
The multiway Revel's tend to have superb measured performance and in pretty much all important areas. If I had the money and no desire to build speakers their products would be at the top of my list.
It often surprises me how poor some massively expensive loudspeakers measure. You'd figure any decent designer would want to keep the on axis response flat, some will argue for 'simplicity', yet don't pick or engineer drivers that support that kind of philosophy, B&W being one of them.
The multiway Revel's tend to have superb measured performance and in pretty much all important areas. If I had the money and no desire to build speakers their products would be at the top of my list.
It often surprises me how poor some massively expensive loudspeakers measure. You'd figure any decent designer would want to keep the on axis response flat, some will argue for 'simplicity', yet don't pick or engineer drivers that support that kind of philosophy, B&W being one of them.
I think the point at which the on-axis response converges on being virtually the only important issue depends on the design (maybe because it is different or maybe because it is wrong, but still). It is possible for off-axis radiation to become the significant issue even if only over a certain range.
The great speakers have usualy poor WAF Since 99% of audiophols are badly married (and deservedly so )only 1% or less enjoy good sound. Of course that unhappy 99% is going to rationalize but in the end they are sentenced to bad sound anyway hehe somehow I'm not sympathetic to those poor bastards ...
Since 99% of audiophols are badly married (and deservedly so )only 1% or less enjoy good sound. Of course that unhappy 99% is going to rationalize but in the end they are sentenced to bad sound anyway hehe somehow I'm not sympathetic to those poor bastards ...What do Great Sounding Speakers have in Common?
..I'm not sure that all of the "class A" speakers are in fact Great Sounding, rather they managed to get the votes necessary for that particular article at that particular time from what is overall a very limited selection. In fact I wouldn't be at all surprised if at least a few reviewers felt quite differently.
As for any objective/measured correlation, I don't think you'll find one - at least not when you compare the objective results with all the other classes. The fact is that some MUCH *less* expensive speakers in other categories are objectively better performing, and often enough a LOT better performing (..by any of the measurements/displays taken/provided).
I'd normally agree with Dave's emphasis with "flat on axis" anechoic response, but even that doesn't achieve dominance (..as emphasized by any number of "class A" speakers). Here is a particularly good example of a "rule breaker" (..as if there were in fact "rules"):
Voxativ Ampeggio loudspeaker Measurements | Stereophile.com
Lousy anechoic response (..unless it's averaged into oblivion), while still having an excellent in-room average.. go figure.
As for ridiculous pricing and the like - I don't think that's entirely true either. Niche enthusiast level products in small production runs for people who will spend their *discretionary* income/savings on these products. Ferrari profits to more than a billion a year with exactly this same model. For most people owning a Ferrari (particularly new production) - is ridiculous ..but for a few it's a worth it.
I'm not even sure if most are ridiculous just on the basis of pricing vs. parts/assembly/marketing et al.. (..though I *think* I'd have a stronger argument there).
..I'm not sure that all of the "class A" speakers are in fact Great Sounding, rather they managed to get the votes necessary for that particular article at that particular time from what is overall a very limited selection. In fact I wouldn't be at all surprised if at least a few reviewers felt quite differently.
As for any objective/measured correlation, I don't think you'll find one - at least not when you compare the objective results with all the other classes. The fact is that some MUCH *less* expensive speakers in other categories are objectively better performing, and often enough a LOT better performing (..by any of the measurements/displays taken/provided).
I'd normally agree with Dave's emphasis with "flat on axis" anechoic response, but even that doesn't achieve dominance (..as emphasized by any number of "class A" speakers). Here is a particularly good example of a "rule breaker" (..as if there were in fact "rules"):
Voxativ Ampeggio loudspeaker Measurements | Stereophile.com
Lousy anechoic response (..unless it's averaged into oblivion), while still having an excellent in-room average.. go figure.
As for ridiculous pricing and the like - I don't think that's entirely true either. Niche enthusiast level products in small production runs for people who will spend their *discretionary* income/savings on these products. Ferrari profits to more than a billion a year with exactly this same model. For most people owning a Ferrari (particularly new production) - is ridiculous ..but for a few it's a worth it.
I'm not even sure if most are ridiculous just on the basis of pricing vs. parts/assembly/marketing et al.. (..though I *think* I'd have a stronger argument there).
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.