Wrong, wrong, wrong.
We're not foolish - we have perceptual capabilities that are prone to false positives, because overlooking a real positive can or could have fatal consequences.
The controversy between the golden ear and the meter reader is not about conflicting beliefs, its about reliable evidence against unreliable evidence.
The world won't disappoint me because I live in a world of universal truths.
Just prove to me that the laws of physics don't reliably operate and I just might join those who live in a world of irrational relativism and denial of obvious truths.
Just remember, you speak for yourself but not everybody.
Denial of obvious global truths is kinda interesting to witness. Why people subject themselves to this kind of torture escapes me.
Quantum mechanics show the world in a very weird light. However it seems now we totally trust these things. If you like we have reinvented faith over science. This time it seems we can actually trust that it is real. That is we know it to be true yet our everyday life seems to say it can't be true. I infered from Quantum Theory that time is an abstract that reality needs to exist. If not all substance to avoid technical terms would be in one place. Science is lucky to be PC, it gets dangerously close to being divisive.
https://www.newscientist.com/articl...roved-real-in-first-loophole-free-experiment/
https://www.newscientist.com/articl...roved-real-in-first-loophole-free-experiment/
Last edited:
Measurements help a company reliably make the same product and meet a published specification. This is very useful when transistors have to change. The science of perception is well known and to the best of my knowledge not used in the making of any hi fi product. This is mostly because hi fi like football has grown rules. Hi fi rules went like this. 1 % THD if the harmonics are in an exponential decay is at the limit of human perception. Most speakers no matter how expensive do not meet this requirement. It was thought amplifiers should have less than 0.1% THD to allow for the nasty harmonics feedback amps often had ( that is, get rid of lower ones without removing the higher one needs it to be very low overal ). Personally I would say 0.1% with exponential harmonics will not only pass the test, but reduce fatigue. We can easilly make tube or transistor amps do that. Dynamic ranges of 50 dB are useful ( whole system back to microphone ). That is the music and hi fi chain combined. Most hi fi's could do with a first class compressor. The old tube types seem ideal. A Revox serves well as one if running a tape is no problem. 1 % in exponential decay I doubt is ever met for the whole system.
People seem to say they can hear an op amp of < -100 dB distortion sounds better than a NE5534. Doubtless they can. The reasons are not easy to prove. NE5532 and MC33078 is interesting. To my ears the simplicity of MC33078 wins. The specs are very similar if not wanting to drive 600R.The 33078 sounds more transparent. The difference is not vast. The long term comfort when listening is the thing I " suspect " I hear. Capacitors sound about the same as op amps. Many cheap caps work well.
Used with care an old Quad 33/303 ESL57 meets this real science spec. Great care is needed matching it's input levels ( +/- 3dB at best ). This often will mean the system sounds quite poor. The phono input is not really good enough. Other than that it can be close to an ideal standard. The speakers are very easilly broken which is I suspect why most people would reject them. The later 405 whilst being an intelligent design it not better than the 303. This might be because it was aimed at the pro market and was made bomb proof. Sound suffered. Paralell bridged 405's in monoblocs are very OK. Very easy to do. many people here would not give the Quad system house room. And yet is meets the universal standard. Your wife might say no is totally valid. I have never heard a Martin Logan set up sound good. I know in my heart this is because the owners were rich and bough status products. The dealers seem to sell a system of the Kings New Suit of Clothes type without flinching. Often in these systems a NAD 3020 would have been better. It will never happen.
People seem to say they can hear an op amp of < -100 dB distortion sounds better than a NE5534. Doubtless they can. The reasons are not easy to prove. NE5532 and MC33078 is interesting. To my ears the simplicity of MC33078 wins. The specs are very similar if not wanting to drive 600R.The 33078 sounds more transparent. The difference is not vast. The long term comfort when listening is the thing I " suspect " I hear. Capacitors sound about the same as op amps. Many cheap caps work well.
Used with care an old Quad 33/303 ESL57 meets this real science spec. Great care is needed matching it's input levels ( +/- 3dB at best ). This often will mean the system sounds quite poor. The phono input is not really good enough. Other than that it can be close to an ideal standard. The speakers are very easilly broken which is I suspect why most people would reject them. The later 405 whilst being an intelligent design it not better than the 303. This might be because it was aimed at the pro market and was made bomb proof. Sound suffered. Paralell bridged 405's in monoblocs are very OK. Very easy to do. many people here would not give the Quad system house room. And yet is meets the universal standard. Your wife might say no is totally valid. I have never heard a Martin Logan set up sound good. I know in my heart this is because the owners were rich and bough status products. The dealers seem to sell a system of the Kings New Suit of Clothes type without flinching. Often in these systems a NAD 3020 would have been better. It will never happen.
My Godness !
Reliable ? Evidence ? Talking about the paintings (production ->records) of a non existing object (Artist's project) observed at different time from different windows (our rooms and systems) by subjective people (our ears+brain+culture) compared to measuring instruments that explore different waves lenghs ?
Hifi is a make believe game, trying to fool our senses for our pleasure, an attempt to share emotions and feelings between people.
Measurements instruments are just tools helping designers to avoid some defects in their systems conception and progress in the right direction. They will never ensure the illusion (you believe that you are listening to a real thing) is working for you or an other one.
I don't know what is a "Golden ear", because, if somebody listen differently than me, what he can say has no interest for me.
Art and science both relate to audio, but it is very helpful to not confuse them or conflate them.
We know for a fact that if we handle the audio wave that represents the art of the music with sufficient care, the art is preserved.
If that is how you need to view audio, be my guest. However other viewpoints are valid and may be more helpful.
There are in fact a small number of measurements that allow us to separate subjectively accurate from subjectively inaccurate.
That can be true in sighted evaluations, but superior alternatives to sighted evaluations do exist.
That my friend can be symptomatic of solipsism. If you want to share your perceptions, solipsism can be very self-destructive.
Havin been on the both side of the discipline (as a member of R&D department of audio equipment manufacturers, as well as technical manager of various studios, and sound engineer for the other side) I never confuse them, but consider correct scientific engineering and attention to his feelings (measurements and listenings) are two inseparable tools. Missing one, you miss all.
Bring-me an amp, i will first listen to it, and it will takes me one minute to reject-it. It is faster than any measurement.
And two days to adopt-it or not, including measurements ;-) (Not a lot succeed ;-)
When I design my own amps (i'm now retired), i usually spend 10X my time to listen at each step than measuring... Sometimes the two don't correlate, I try then to understand why.
Well, given your background your position in this matter might be fairly typical and totally self-serving.
In 40+ years of blind testing we have found that every case of "sounds bad but measures good" relates to either incomplete measurements or incomplete listening, with he former be far more common.
The myth here being that any brief listening test can be counted on to expose every audible flaw that may manifest itself in a real world operational circumstance.
If I couldn't run a sufficient set of measurements in far less time...
Try some good blind tests - they may well change your life.
So tricky that one. The possibly overlooked thing is we measure what we can. The measuring device is of the same type. It isn't us measuring beauty. All the same for sainty alone I side with you.
C
Don't take others as stupid; I organized several blind test seance, for demonstrations or publications needs (Digital VS Analog, MP3 or ATRAK VS original etc.)
Most of the things result of the choices of the material you gone use to listen, the knowledge you have of them, your culture etc...
I consider, when I am in an expectative and things are not so obvious that you are in need to practice this kind of ceremony that it is not worth the trouble.
See what I mean ? And I don't like to fool myself as well, biasing with prejudices.
About the time some need to have an opinion on amplifiers, by example, there are things like changes with "burning", temperature, various conditions of the AC lines, depending of the moments of the day, listening fatigue, that you cannot explore instant, or by a single seance of measurements.
And, I suppose it is because musical signals are VERY different from pure sinusoides, there are obvious differences in characters (specialy about micro or instant dynamic) that you cannot notice on the bench table, despite they are obvious for everybody in listening.
It is a legend that the blind hear better than others.
Don't take others as stupid; I organized several blind test seance, for demonstrations or publications needs (Digital VS Analog, MP3 or ATRAK VS original etc.)
Most of the things result of the choices of the material you gone use to listen, the knowledge you have of them, your culture etc...
I consider, when I am in an expectative and things are not so obvious that you are in need to practice this kind of ceremony that it is not worth the trouble.
See what I mean ? And I don't like to fool myself as well, biasing with prejudices.
About the time some need to have an opinion on amplifiers, by example, there are things like changes with "burning", temperature, various conditions of the AC lines, depending of the moments of the day, listening fatigue, that you cannot explore instant, or by a single seance of measurements.
And, I suppose it is because musical signals are VERY different from pure sinusoides, there are obvious differences in characters (specialy about micro or instant dynamic) that you cannot notice on the bench table, despite they are obvious for everybody in listening.
Lame attempt to deflect the discussion.
I said good blind tests. I didn't say demos for visiting firemen.
What follows appears to be yet another lame attempt to deflect the discussion:
I recently built a phono stage to get the last beans out of my system. It was designed to exactly fit the job it had to do. For once every box of good design practice could be ticked. It will not have slewing problems even if a 741. I don't want to show exactly what I did except to say it resembles a 1961 Tobey and Dinsdale circuit from Wireless World. I was wrong about that and built what I wrongly remembered. It is rather nice. Tobey and Dinsdale did the same thing, but not the same way.
I used NE5532 as I have a passion for using them for my starting point. As it has a DIL holder a swap takes 10 seconds and 1 minute to hearing the result. The MC33078 at first sounds very similar. Then it becomes clear that I have fitted an much more expensive pick up or cleaned a dirty stylus when I haven't. I don't want to say night and day, it wasn't. It just was less processed. These devices cost about the same. Looking on paper the MC33078 seems to have slightly the advantage on HF distortion. GBWP is 60% better. The chip signal path is less complicated. This seems to be to get more current out of NE5532. I suspect on a blind test I and anyone would be hard pushed to say which is the better.
Now to the important part. Just as knowing the identity of a part makes one believe it better, not knowing ruins the test. It forces us to use the wrong part of the brain to do the test. It's the same when a golf pro misses an easy shot. Fear causes the wrong part of the brain to guide the hands. I think it is important to see all of these tests are meaningless. I suppose if the choice is backed up by good science and works OK the designer should be allowed to have a preference as the reality of it. At least it nails down that part of the design. At worse it does nothing.
I went on to try TL072. It was truely awful! No way should it have been as even the noise was OK ( like a Revox or better ). It was dull, compressed, harsh ( !!!! ) and yellow sounding for want of a better word. Looking inside a TL072 it looks very nice. The output I required was 100 mV rms max, 400 K gain of 38 dB. A 22K resistor to -ve so as as to have an SE class A output. One op amp I didn't have was LM358. This circuit could work OK with it. Doubtless it would have been worse. My hunch is not as bad as we might think.
Measurements mostly tell you if something is broken. After that it is the music that says what's best. Somethings will be bad for task and that is where measurements matter. An excellent example of this was EF86 valves. Some had more noise than others. The type of noise also, the best were more pink ( not partition noise ) . Doubtless on paper they all were usable. It took less time to listen to them than get a test rig going. Often if 4 were needed and 4 you had, the best two in the input stage was how to do it. Here we are talking night and day different. My instinct is to say listening for whatever reason still is best. If no better no one looses. If better, then it is better. What often happens in this world is the proof one needs comes 50 years later. The person who steadfastly held the status quo as the bigger truth has done less than nothing worth remembering. The person who discovers a better understanding is a hero. I have no interest to be either. I would prefer not to be a member of a deaf rather than blind listening test. Deaf being mental fingers in the ears.
I used NE5532 as I have a passion for using them for my starting point. As it has a DIL holder a swap takes 10 seconds and 1 minute to hearing the result. The MC33078 at first sounds very similar. Then it becomes clear that I have fitted an much more expensive pick up or cleaned a dirty stylus when I haven't. I don't want to say night and day, it wasn't. It just was less processed. These devices cost about the same. Looking on paper the MC33078 seems to have slightly the advantage on HF distortion. GBWP is 60% better. The chip signal path is less complicated. This seems to be to get more current out of NE5532. I suspect on a blind test I and anyone would be hard pushed to say which is the better.
Now to the important part. Just as knowing the identity of a part makes one believe it better, not knowing ruins the test. It forces us to use the wrong part of the brain to do the test. It's the same when a golf pro misses an easy shot. Fear causes the wrong part of the brain to guide the hands. I think it is important to see all of these tests are meaningless. I suppose if the choice is backed up by good science and works OK the designer should be allowed to have a preference as the reality of it. At least it nails down that part of the design. At worse it does nothing.
I went on to try TL072. It was truely awful! No way should it have been as even the noise was OK ( like a Revox or better ). It was dull, compressed, harsh ( !!!! ) and yellow sounding for want of a better word. Looking inside a TL072 it looks very nice. The output I required was 100 mV rms max, 400 K gain of 38 dB. A 22K resistor to -ve so as as to have an SE class A output. One op amp I didn't have was LM358. This circuit could work OK with it. Doubtless it would have been worse. My hunch is not as bad as we might think.
Measurements mostly tell you if something is broken. After that it is the music that says what's best. Somethings will be bad for task and that is where measurements matter. An excellent example of this was EF86 valves. Some had more noise than others. The type of noise also, the best were more pink ( not partition noise ) . Doubtless on paper they all were usable. It took less time to listen to them than get a test rig going. Often if 4 were needed and 4 you had, the best two in the input stage was how to do it. Here we are talking night and day different. My instinct is to say listening for whatever reason still is best. If no better no one looses. If better, then it is better. What often happens in this world is the proof one needs comes 50 years later. The person who steadfastly held the status quo as the bigger truth has done less than nothing worth remembering. The person who discovers a better understanding is a hero. I have no interest to be either. I would prefer not to be a member of a deaf rather than blind listening test. Deaf being mental fingers in the ears.
Christophe, testing is determined by the test objective, what are you looking for, something a panel will find to be better than some other, or something that you yourself prefer. Meaning, not necessarily better in absolute terms, but it might sound better for your taste.
And sometimes, your opinion of what actually sounds better may also sound better to a listening panel. For example, the Burr-Brown phono eq/amp I posted here several times, dating back to 1990, the work of one Dr Holger Hermann of Burr-Brown Germany, it uses two linear op amps, with a passive RIAA network between them, is still the best sounding device of its type I have heard to date. It seems that quite a number of people who hear it in action agree, at the very least, that it is one the best around. No matter how you test it, it always finds its way ritgh up to the top of the list, typically one of the two best on the menu.
The point is it sits well with me and my taste, so I don't really care whether it will be the best in any group because it covers all my bases very well. True, I have heard some really good ones over the years, but most top models were not so much "better" as they were just a bit different in how they do their job.
And sometimes, your opinion of what actually sounds better may also sound better to a listening panel. For example, the Burr-Brown phono eq/amp I posted here several times, dating back to 1990, the work of one Dr Holger Hermann of Burr-Brown Germany, it uses two linear op amps, with a passive RIAA network between them, is still the best sounding device of its type I have heard to date. It seems that quite a number of people who hear it in action agree, at the very least, that it is one the best around. No matter how you test it, it always finds its way ritgh up to the top of the list, typically one of the two best on the menu.
The point is it sits well with me and my taste, so I don't really care whether it will be the best in any group because it covers all my bases very well. True, I have heard some really good ones over the years, but most top models were not so much "better" as they were just a bit different in how they do their job.
Nige, measurements are highly desirable when debeloping any circuit. They will very reliably show whether a cicruit is stable and how precisely does it do its job. But once you get it to work reliably and relatively predictably, the final step of "voicing" it is always down the designer and his hearing. Small differences in design may cause very audible differences in the whole, even if the measurements are hardly affected. For example, biasing of a differential pair, when biasing of say 1.5 mA may work very well indeed, but turning it into 3 mA may cause large differences in sound quality yet influece measurements results very little.
10 seconds is way too long of a switch over time. Your brain starts loosing its detailed memory of the sound. This makes everything sound different because of the lost details. Of course the fact that the comparison is sighted invalidates everything, too.
I know of no proper quick-switched, level-matched, DBTs where differences between good modern op amps that were properly applied led to reliably audible differences. People have tried. If you look at the measurable properties of op amps that are commonly used for audio, their noise and distortion is so low that measuring can be a challenge, let alone hearing it.
One common rookie mistake is swapping op amps and ending up with a preamp that is oscillating in the ultrasonic realm. You can't hear it oscillatoing directly, but it is adversely affecting sound quality anyway.
Reading Nigel's post (18593) gives reason to believe that there is some sanity and common sense left in this mad world of audio - in which wars of domination of personal perspective are so rife.
It is a world in which personal insult and implication should have no part. Yet it is a sad fact that there are always some half dozen or so who - despite having a good depth of knowledge and the experience required to actually assist lesser mortals - seem more intent on forcing their views down other peoples' throats, or in seeking (pyrrhic) victory, that anything of value which they post tends to be overlooked.
It is a world in which personal insult and implication should have no part. Yet it is a sad fact that there are always some half dozen or so who - despite having a good depth of knowledge and the experience required to actually assist lesser mortals - seem more intent on forcing their views down other peoples' throats, or in seeking (pyrrhic) victory, that anything of value which they post tends to be overlooked.
IOW he could understand and agree with what Nigel said, no matter how reality-based it was or was not.
It didn't seem like there was much of that.
Kinda hard to fit into the context. I can only presume that by "personal attack" and "forcing their views down people's throats" what is really meant is relates to personal understanding and agreement.
My view is 60/40 at a rough guess. That is 60% the science and 40% guessing from the science what sound I would like.
One thing I noticed from a valve amp is that a 10 % THD wave can look almost like a 0.001 % THD wave on the scope. Until the two are side by side the difference isn't as obvious as it should be. On the spectrum analyser it shows all the harmonics in a lovely suspension bridge curve. From theory the ear will do the same mechanically. Medical estimates put unprocessed ear distortion at as high as 30%. The brain is much like a digital device that can know the real sound from it's distortion ( and process external sounds into good or less good ). Medical research found a possible servo signal from the brain to ear which suggested a 2 MHz bandwidth or the speed that signal would have, I supplied special KEF T27 for that. The standard ones can do 40 kHz, these were selected to go above that. The servo is tested by swiching off the sound and noting the reaction delay. If it is a servo is not as yet known.
Next a mains voltage wave at 4 % THD. This will look horrible and one would think it 5 times the distortion of the extreme example valve amplifer.
If these two waves were played through a speaker the 10 % one will sound colourful yet mostly pure. The 4% wave will sound like a bag of nails. What fools the eye can fool the ear.
The point to make is if the measuring tools are understoood and perception likelihood considered the picture changes. Also things that seem impossible often are not. For example some 78's could offer 32 kHz ( Decca FFRR ). The actual shellac pressings good for 7 to 10 kHz if new. Now the impossible bit. The 32 kHz is still a usful part of the sound. The shape of the groove will retain some traits of the 32 kHz capability. I will go further. My baffle speakers should do 10 kHz. This was not enough. A tweeter feeding in at 6 kHz has helped greatly. It gives a nice window which is now 6 seat widths wide, it was less than 1. The weirdest thing is 1930's 78's really need the tweeter! Somehow they are doing something which really helps. It could be harmonic distortion making an extra octave. It sounds really good.
One thing I noticed from a valve amp is that a 10 % THD wave can look almost like a 0.001 % THD wave on the scope. Until the two are side by side the difference isn't as obvious as it should be. On the spectrum analyser it shows all the harmonics in a lovely suspension bridge curve. From theory the ear will do the same mechanically. Medical estimates put unprocessed ear distortion at as high as 30%. The brain is much like a digital device that can know the real sound from it's distortion ( and process external sounds into good or less good ). Medical research found a possible servo signal from the brain to ear which suggested a 2 MHz bandwidth or the speed that signal would have, I supplied special KEF T27 for that. The standard ones can do 40 kHz, these were selected to go above that. The servo is tested by swiching off the sound and noting the reaction delay. If it is a servo is not as yet known.
Next a mains voltage wave at 4 % THD. This will look horrible and one would think it 5 times the distortion of the extreme example valve amplifer.
If these two waves were played through a speaker the 10 % one will sound colourful yet mostly pure. The 4% wave will sound like a bag of nails. What fools the eye can fool the ear.
The point to make is if the measuring tools are understoood and perception likelihood considered the picture changes. Also things that seem impossible often are not. For example some 78's could offer 32 kHz ( Decca FFRR ). The actual shellac pressings good for 7 to 10 kHz if new. Now the impossible bit. The 32 kHz is still a usful part of the sound. The shape of the groove will retain some traits of the 32 kHz capability. I will go further. My baffle speakers should do 10 kHz. This was not enough. A tweeter feeding in at 6 kHz has helped greatly. It gives a nice window which is now 6 seat widths wide, it was less than 1. The weirdest thing is 1930's 78's really need the tweeter! Somehow they are doing something which really helps. It could be harmonic distortion making an extra octave. It sounds really good.
Something to add. Whilst reading up on motors it was said second harmonics rarely causes vibration as the peaks are in phase ( mostly ). When third harmonic it fights the rotation and causes vibration. A speaker is a motor more or less. That reading told me more than most hi fi reading. I wish the writter had written about exponentially reducing harmonics. My instinct is to think it would run with minimal vibration as if from a zero distortion source. If the distortion was helpful it might reduce vibration. I had set a project on this at Greenwich University which never got started. We might have tried motional feedback or negative impedance principles as in Meridian active speakers ( I think , Also Professor Korn of Belgium circa 1972. Servo Sound ) . The risk with that is making new higher order vibration at low level.
In British libraries we have a number notation for books anywhere in the country. 621 for electronics is next to marine diesels. By chance I read up on vibration in diesel engines. It is caused by one cycle being the firing stroke. Thus 3 near perfect sine wave parts and one flat top one. Again the graphs given were better than hi fi. Even allowing for 4 cylinders the vibration is very nasty.
In British libraries we have a number notation for books anywhere in the country. 621 for electronics is next to marine diesels. By chance I read up on vibration in diesel engines. It is caused by one cycle being the firing stroke. Thus 3 near perfect sine wave parts and one flat top one. Again the graphs given were better than hi fi. Even allowing for 4 cylinders the vibration is very nasty.
- Status
- Not open for further replies.