BTW I completely agree with you about the frequency response measured I wouldn't probably be that brave either to make something that has a peak resonance at 3.5KHz, at least this is what my logit tells me.
However the set of results from Stereophile and the one provided by B&W don't match and again, I agree with you that 3d party is more reliable or at least not biased, but there could as well be some mistake on those measurements I don't know.
All I can say is that listening proves these measurement somehow wrong.
Looking at those curves would make me think of a poor work and something that isn't going to sound any good or at least not coherent for sure.
I can however ensure you that I haven't heard such an accuracy on the tonal balance of the voice especially as I heard it with these speakers and like I told you I have a direct way to compare few recordings I have with the real stuff and in way too many occasions with other speakers and systems the voice was either too bright or too mellow.
I invite you to go listen to those speakers with nothing but the best audio gears and in the proper room and try not to think about the measurements.
I would be interested to know your opinion on that.
Ending my thought I can't disagree with your objective point, but at the same time I don't find a real explanation as to why this wouldn't reflect listening and I can maybe only conclude that for some reason the set of measurements made by stereophile weren't accurate or I am missing something else.
However the set of results from Stereophile and the one provided by B&W don't match and again, I agree with you that 3d party is more reliable or at least not biased, but there could as well be some mistake on those measurements I don't know.
All I can say is that listening proves these measurement somehow wrong.
Looking at those curves would make me think of a poor work and something that isn't going to sound any good or at least not coherent for sure.
I can however ensure you that I haven't heard such an accuracy on the tonal balance of the voice especially as I heard it with these speakers and like I told you I have a direct way to compare few recordings I have with the real stuff and in way too many occasions with other speakers and systems the voice was either too bright or too mellow.
I invite you to go listen to those speakers with nothing but the best audio gears and in the proper room and try not to think about the measurements.
I would be interested to know your opinion on that.
Ending my thought I can't disagree with your objective point, but at the same time I don't find a real explanation as to why this wouldn't reflect listening and I can maybe only conclude that for some reason the set of measurements made by stereophile weren't accurate or I am missing something else.
Thank you for those replies as that helped a lot.
First of all though I see the art, or the human touch part of it being in the creative side of the music. The hi-fis role is to reproduce this as faithfully as it can, art therefore, not being part of the reproduction chain.
Now as to the ignoring the ears, I am not saying that I am. What I've said is that a peak at that frequency is very much audible and if it were notched out you would hear it. It's presence in and of itself may not be offensive to the ears, in some cases it may even be preferred as it will add emphasis to certain parts allowing you to more easily here certain bits of the music. Just like a bit of extra boost in the bass can be very pleasing to the ears, as can a bit of lift in the top octave by providing more 'air' to the sound. These things wont get you bad reviews, far from it, subjectively they should do very well, but by definition any deviation from flat means a deviation away from being completely neutral. The B&Ws might be used by some recording studios because that peak allows them to more easily hear certain aspects of the music, but to say that they are completely neutral would be wrong.
Do I like completely neutral sounding loudspeakers? On the whole yes, but I do prefer a bit of bass lift and a slightly downward sloping response. Sometimes I like a bit of lift in the treble but it depends on the dispersion characteristics of the loudspeaker in general.
From lots of experience however, I do know that my ears don't tolerate peaks in the 1-4kHz band particularly well. When my hyperacusis started I figured it was because of the loudspeakers I was designing at the time had some flaw in around the presence region. Playing around with the 1-4kHz band helped significantly with this, with any peaks being a really bad idea. I then later realised that it wasn't the loudspeakers that were to blame, it was actually my left ear. Needless to say what I've found helps with this is very low distortion drivers, with very well integrated drivers with a smooth set of off axis curves and a flat on axis response with a bit of lift in the bass. Having the midrange handled by the FST driver makes the loudspeakers a lot easier to listen to, too. But it's as you said, simply having a flat frequency response isn't the whole story, as other flat speakers have me running for the off switch. So while I need a combination of parameters all to be right for it to be easy enough to listen to, having a peak in the 1-4kHz region will mean that no matter how right everything else is, it wont sound good to my ears and likewise flat with other parameters being off wont work either.
Building the DSP and using absolute digital delay to time align the drivers properly also helped with this. Allpass filters also helped vs using asymmetric slopes for the phase alignment, but nothing helped as much as the true digital delay. It was a bit of a gamble buying the FST direct from B&W as I already had very good, low distortion, midrange drivers, but the FST just made things even easier to listen to.
First of all though I see the art, or the human touch part of it being in the creative side of the music. The hi-fis role is to reproduce this as faithfully as it can, art therefore, not being part of the reproduction chain.
Now as to the ignoring the ears, I am not saying that I am. What I've said is that a peak at that frequency is very much audible and if it were notched out you would hear it. It's presence in and of itself may not be offensive to the ears, in some cases it may even be preferred as it will add emphasis to certain parts allowing you to more easily here certain bits of the music. Just like a bit of extra boost in the bass can be very pleasing to the ears, as can a bit of lift in the top octave by providing more 'air' to the sound. These things wont get you bad reviews, far from it, subjectively they should do very well, but by definition any deviation from flat means a deviation away from being completely neutral. The B&Ws might be used by some recording studios because that peak allows them to more easily hear certain aspects of the music, but to say that they are completely neutral would be wrong.
Do I like completely neutral sounding loudspeakers? On the whole yes, but I do prefer a bit of bass lift and a slightly downward sloping response. Sometimes I like a bit of lift in the treble but it depends on the dispersion characteristics of the loudspeaker in general.
From lots of experience however, I do know that my ears don't tolerate peaks in the 1-4kHz band particularly well. When my hyperacusis started I figured it was because of the loudspeakers I was designing at the time had some flaw in around the presence region. Playing around with the 1-4kHz band helped significantly with this, with any peaks being a really bad idea. I then later realised that it wasn't the loudspeakers that were to blame, it was actually my left ear. Needless to say what I've found helps with this is very low distortion drivers, with very well integrated drivers with a smooth set of off axis curves and a flat on axis response with a bit of lift in the bass. Having the midrange handled by the FST driver makes the loudspeakers a lot easier to listen to, too. But it's as you said, simply having a flat frequency response isn't the whole story, as other flat speakers have me running for the off switch. So while I need a combination of parameters all to be right for it to be easy enough to listen to, having a peak in the 1-4kHz region will mean that no matter how right everything else is, it wont sound good to my ears and likewise flat with other parameters being off wont work either.
Building the DSP and using absolute digital delay to time align the drivers properly also helped with this. Allpass filters also helped vs using asymmetric slopes for the phase alignment, but nothing helped as much as the true digital delay. It was a bit of a gamble buying the FST direct from B&W as I already had very good, low distortion, midrange drivers, but the FST just made things even easier to listen to.
One reason you may not hear it is because it takes adequate energy and exposure to that energy to excite the cone into ringing.
Maybe during music playback it is not resonating to the same extent that it does when doing frequency sweeps at constant voltage
Or maybe you find the sound more enjoyable or musical
Maybe during music playback it is not resonating to the same extent that it does when doing frequency sweeps at constant voltage
Or maybe you find the sound more enjoyable or musical
Last edited:
That ringing may only be presented to him as a peak, if the speakers has a lot of toe in, if his current setup is straight ahead with very little toe in, it may not be audible . Also an overdamped room will tame , i had the same Issue listening to the new Wilson sasha's , heavy ringing in the upper mids on piano's , changing the toe and moving them away from the walls , helped, but it was still there, the owner is now changing cables and doing room treatment.
When drivers have such an anomaly , you dont have to measure, listening test , done to the driver in free space with out xover when voicing will tell all tales.
Measurements will confirm what the ear detect....
Not necessarily. It depends on how the measurements are carried out and whether they are representative of conditions while listening to music. Most measurements do not take this into account.
Cone resonance is not corrected by cables. It is corrected by treating the cones eg ENABL process or application of viscoelastic coatings in strategic places on the cone
Last edited:
Well to make a comparison to cancer, you get benign and malignant resonances and if they are from the cone/surround/spider/motor then they almost always show up regardless of the drive level.
Benign resonances generally only affect the frequency response and once equalised out the issues go away. Lots of paper, composite or some of the stiffer poly cones all tend to exhibit this kind of resonance and whether or not the issue can be controlled depends on how severe the resonance is. If the resonance happens to produce a single peak, then you can control it quite easily, the trouble occurs when you have multiple peaks at different frequencies and of different magnitudes. These you cannot easily control, so the only real option is to crossover before you get to them.
Malignant resonances generally show up in the very stiff cones, SEAS alu and magnesium cones all exhibit this problem and so do Accuton's ceramic cones. This kind of resonance affects both the frequency response, but it also appears in the harmonic distortion plot too. You get a horrible peak in the HD at the frequency of the resonance, but you also get peaks in the individual harmonics at integer multiples below the frequency of the resonance. In other words if the cone has a resonance at 9k, there will be a peak at 3k in the third harmonic. This dictates that you cross over before 3k if you want to avoid this issue.
The FST driver, as B&W say, is in a state of controlled resonance/breakup for much of it's upper range. This much is true and the resonances are also quite benign, you get wiggles in the frequency response but the HD is completely clean. This does mean that if you can accept these resonances and their associated mild colourations due to the frequency response variations, that you can use the FST up high without any other compromises except that the off axis will suffer.
Now the FST isn't perfect, the controlled resonance/breakup that manages to extend the cones range up to around 10k isn't a problem, but the driver does exhibit one strong more bell like resonance (like a metal cone) at 3.5kHz. This resonance does produce a large peak in the harmonic distortion, luckily this doesn't plague the 3rd and 5th order harmonics at 1/3 and 1/5x the fundamental.
Now the reason why the peak at 3.5k isn't as much of an issue as it could be is that the driver is amazingly well designed. If you look at the harmonic distortion, the 4th and 5th order harmonics are at 0.001% and the third harmonic at 0.01% by the time the resonance occurs. When the resonance hits the distortion goes up by an order of magnitude, the third harmonic hitting 0.1% and the 4th and 5th hitting 0.01%. As you can appreciate this isn't a problem. Most other drivers are generally at around 0.1% or higher in the third harmonic at this time and a resonance would push the distortion above 1%, this is a problem and in any other driver the resonance at 3.5k would severely limit how high you'd want to use it, not the FST though because its motor is so incredibly linear up that high. This is one of the reasons I jumped on a pair as soon as I saw Zaph's measurements.
B&W have obviously done their homework on designing the driver, they knew they wanted to crossover high, knew that this would mean the cone must be going through a region of controlled resonance to allow it to work up that high and came up with a cone material and motor that gives them almost everything. If the driver didn't have this strong resonance at 3.5kHz then it would be perfect. I am sure B&W have tried lots of cone variations, weaves, geometries and doping materials to try and get rid of it, but it can't have been possible.
In my application with the driver I cross it over at 2.5kHz and also notch out the 3.5kHz peak so that it is rendered a complete non issue, it also crosses it over before the driver starts to go through its large region of controlled breakup so keeps the mild colourations associated with them also a non issue.
B&W and I obviously have different ideas about how we want things to be done, but this isn't anything new in the world of loudspeaker design. I would however be very interested in working with a pair of 800Ds, 801Ds or 802Ds to see what a redesign of the crossover would be like if designed to my preferences rather than B&Ws.
Contacting B&W is an idea, but the designer would probably say, we tried it with the peak notched out but preferred it with the peak left alone. I could also ask if they tried crossing over at 2.5kHz or so with steeper filters and they might just say, why would we want to do that, when we want to design a speaker with shallow filters and high crossover points. Loudspeaker design is always a game of compromises, B&W obviously know this and came up with a midrange driver that keeps these compromises as small as possible given their design preferences (you see lots of designs where the designers have these preferences also, but the drivers aren't really suitable for the task, the FST however is). Now B&W have lots of fans, I am one of them but for sightly different reasons, but these fans obviously like the way B&W loudspeakers sound. Shallow filters attached to the Kevlar cones at higher then normal xover frequencies are a major part of this B&W house sound, so perhaps they leave the peak present and crossover high because they know that sound is very popular with a lot of people. Technically speaking the loudspeaker would be more neutral if crossed over more conventionally,but then it wouldn't be what you'd expect to hear from a B&W and this does count for quite a lot.
Benign resonances generally only affect the frequency response and once equalised out the issues go away. Lots of paper, composite or some of the stiffer poly cones all tend to exhibit this kind of resonance and whether or not the issue can be controlled depends on how severe the resonance is. If the resonance happens to produce a single peak, then you can control it quite easily, the trouble occurs when you have multiple peaks at different frequencies and of different magnitudes. These you cannot easily control, so the only real option is to crossover before you get to them.
Malignant resonances generally show up in the very stiff cones, SEAS alu and magnesium cones all exhibit this problem and so do Accuton's ceramic cones. This kind of resonance affects both the frequency response, but it also appears in the harmonic distortion plot too. You get a horrible peak in the HD at the frequency of the resonance, but you also get peaks in the individual harmonics at integer multiples below the frequency of the resonance. In other words if the cone has a resonance at 9k, there will be a peak at 3k in the third harmonic. This dictates that you cross over before 3k if you want to avoid this issue.
The FST driver, as B&W say, is in a state of controlled resonance/breakup for much of it's upper range. This much is true and the resonances are also quite benign, you get wiggles in the frequency response but the HD is completely clean. This does mean that if you can accept these resonances and their associated mild colourations due to the frequency response variations, that you can use the FST up high without any other compromises except that the off axis will suffer.
Now the FST isn't perfect, the controlled resonance/breakup that manages to extend the cones range up to around 10k isn't a problem, but the driver does exhibit one strong more bell like resonance (like a metal cone) at 3.5kHz. This resonance does produce a large peak in the harmonic distortion, luckily this doesn't plague the 3rd and 5th order harmonics at 1/3 and 1/5x the fundamental.
Now the reason why the peak at 3.5k isn't as much of an issue as it could be is that the driver is amazingly well designed. If you look at the harmonic distortion, the 4th and 5th order harmonics are at 0.001% and the third harmonic at 0.01% by the time the resonance occurs. When the resonance hits the distortion goes up by an order of magnitude, the third harmonic hitting 0.1% and the 4th and 5th hitting 0.01%. As you can appreciate this isn't a problem. Most other drivers are generally at around 0.1% or higher in the third harmonic at this time and a resonance would push the distortion above 1%, this is a problem and in any other driver the resonance at 3.5k would severely limit how high you'd want to use it, not the FST though because its motor is so incredibly linear up that high. This is one of the reasons I jumped on a pair as soon as I saw Zaph's measurements.
B&W have obviously done their homework on designing the driver, they knew they wanted to crossover high, knew that this would mean the cone must be going through a region of controlled resonance to allow it to work up that high and came up with a cone material and motor that gives them almost everything. If the driver didn't have this strong resonance at 3.5kHz then it would be perfect. I am sure B&W have tried lots of cone variations, weaves, geometries and doping materials to try and get rid of it, but it can't have been possible.
In my application with the driver I cross it over at 2.5kHz and also notch out the 3.5kHz peak so that it is rendered a complete non issue, it also crosses it over before the driver starts to go through its large region of controlled breakup so keeps the mild colourations associated with them also a non issue.
B&W and I obviously have different ideas about how we want things to be done, but this isn't anything new in the world of loudspeaker design. I would however be very interested in working with a pair of 800Ds, 801Ds or 802Ds to see what a redesign of the crossover would be like if designed to my preferences rather than B&Ws.
Contacting B&W is an idea, but the designer would probably say, we tried it with the peak notched out but preferred it with the peak left alone. I could also ask if they tried crossing over at 2.5kHz or so with steeper filters and they might just say, why would we want to do that, when we want to design a speaker with shallow filters and high crossover points. Loudspeaker design is always a game of compromises, B&W obviously know this and came up with a midrange driver that keeps these compromises as small as possible given their design preferences (you see lots of designs where the designers have these preferences also, but the drivers aren't really suitable for the task, the FST however is). Now B&W have lots of fans, I am one of them but for sightly different reasons, but these fans obviously like the way B&W loudspeakers sound. Shallow filters attached to the Kevlar cones at higher then normal xover frequencies are a major part of this B&W house sound, so perhaps they leave the peak present and crossover high because they know that sound is very popular with a lot of people. Technically speaking the loudspeaker would be more neutral if crossed over more conventionally,but then it wouldn't be what you'd expect to hear from a B&W and this does count for quite a lot.
Last edited:
For a kevlar driver they are the most well behaved I have seen.
What I would like to see is sweeps at 75dB, 80dB and 90dB.
It is possible that at 75dB the peak may not be anywhere near as bad or possibly non-existant.
So while listening to music you might be experiencing 95dB peaks at 40Hz, at 3000Hz you might be only experiencing 75dB peaks, and hence the issue may not be as bad as it looks
What I would like to see is sweeps at 75dB, 80dB and 90dB.
It is possible that at 75dB the peak may not be anywhere near as bad or possibly non-existant.
So while listening to music you might be experiencing 95dB peaks at 40Hz, at 3000Hz you might be only experiencing 75dB peaks, and hence the issue may not be as bad as it looks
Last edited:
5th,
your answer is really, really interesting.
I am would be curious to see the set of measurements that B&W has taken on this speakers.
The person who was taking measurements mentioned that the frequency response B&W had taken is practically flat and it's a mistery why he couldn't match it.
Re-working the crossover maybe something I would be curious to do later perhaps, the only thing is that to get access to the network I think you have to take almost everything apart.
Anyway 5th, I really find in this last post of yours that you have very found believe on what a speaker should do and I am serious and not sarcastic, why don't you try to send to resume to all the companies in UK speaker's manufacturer? There are a lot afterall and they are all in a pretty good shape in terms of business/
Afterall you have experience and you know pretty well how things work on speakers.
Just a thought.
Anyway maybe it would be interesting to ask B&W the condition they took their measurements and the reason why it doesn't match what Stereophile has measured.
your answer is really, really interesting.
I am would be curious to see the set of measurements that B&W has taken on this speakers.
The person who was taking measurements mentioned that the frequency response B&W had taken is practically flat and it's a mistery why he couldn't match it.
Re-working the crossover maybe something I would be curious to do later perhaps, the only thing is that to get access to the network I think you have to take almost everything apart.
Anyway 5th, I really find in this last post of yours that you have very found believe on what a speaker should do and I am serious and not sarcastic, why don't you try to send to resume to all the companies in UK speaker's manufacturer? There are a lot afterall and they are all in a pretty good shape in terms of business/
Afterall you have experience and you know pretty well how things work on speakers.
Just a thought.
Anyway maybe it would be interesting to ask B&W the condition they took their measurements and the reason why it doesn't match what Stereophile has measured.
It is if you forget them completely, then nothing resonates at all!
You guys should get out more often .....
Cabling and room damping can tame what you hear as resonances and yes you can measure what you hear, also on axis response is not as important as avg ,
1 /5/15/25 watt responses should over lap when scaled and is important when looking at macro and micro dynamics ...
Measurement are an accumulation of , not any one defines it all and you can definetly measure what you hear , if you know where to look ..
Last edited:
Their measurements are anechoic , stereophiles is not , there are going to be big differences , as such many in the past would use stereophiles measuring system and technique in other to ace the test so to speak ..
Their measurements are anechoic , stereophiles is not , there are going to be big differences , as such many in the past would use stereophiles measuring system and technique in other to ace the test so to speak ..
so sorry my ignorance, but could this be the reasont for the discrepancy on the frequency response?
I couldn't see B&W publishing fake results on their flagship product...but maybe I am wrong.
I would really be interested of knowing the opinion of B&W on this, but on the other hand like 5th said they are not going, if I real flaw were present, to admit it...don't know!
However I am interested because listening to the speaker reveals no resonance on piano or any other instruments.
They are free of grain and very transparent and their sound seems to be extremely coherent on all the audible band.
I am just curious to understand.
Neither am I, but it is possible to design cables with the right parasitics to cause trouble or fix a problem.
Definitely not the way I would try and solve a problem. A simple notch filter designed with a few cheap components is much easier and effective
But the best solution is to attack the root cause of the problem, by improving the speaker cone further
Definitely not the way I would try and solve a problem. A simple notch filter designed with a few cheap components is much easier and effective
But the best solution is to attack the root cause of the problem, by improving the speaker cone further
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.