That is what you call "question of substance" ? Oh, I see.
Answer-1: I made, during 10 years, a daily ABX blind listening session with one hundred of participants each time as I did for each correction on each instrument during the studio's mixing sessions of the records you use for YOUR blind tests.
Answer-2: Most of the time, my doctor is able to diagnose your disease, just looking at you entering his office.
Last edited:
What have you got to hide? Why not share some details? I like reading about the amps that sound different. Thanks in advance.
Did you think all amps sound the same, that's what is under your question ? if not, what should be the interest for you to read comments about an amplifier older than 30 years ?
I did a few more MOS FET calculations. I was wrong with my guess at 1R loss 8R. 2R seems a more reliable estimate from amps I have built. This seems the same regardless of 8 or 4R. I never realised that . My wrong assumption was 1R for 8R load. This means the 8R load is already less than ideal. My 1R5 assumed was wrong in two ways. I had assumed it less than 1R for 8R load and then subtracted what 4 R seemed to be. Thus me saying not 1R5 was wrong. It's worse.
Now a suggestion that doesn't come from me. Peak current of 20 A real music on everyday speakers is not out of the question. My Magnepans shouldn't do that as they claim 4R resistive. The Quad 303 not known for it's muscle loves them, although gets very warm.
So to revise my estimate of as I will call it from now on I^R losses. 3 easilly bought Vgs matched device pairs should be about 0R6 loss ( better I suspect ). That means that from the same PSU as much as 270 watt 4R as compared with 150 watt. If the 20A peak is true then it will be easilly heard. I ask forgiveness in advance for what might be the romantic notion of a technical journalist. Mr Waynes 1000 watts 1R now has a place of worship.
One thing I will have to think through is this. If the bias is turned up on single pair of audio FET's ( BUZ900/905 , Exicon 10N/P20 ) at 100 mA they are said to take on typical FET rising resistance with load. I am " almost " certain I hear that. A trade between open sound and punch. Certainly at the 5 watts most are said to listen at maybe I imagine it. Problem is 5 watts isn't my typical level. I tend to have the system on and listen all over the house. The Maggies would die so I built some monsters. 20 mA when FET's will produce excellent measurements. I suspect 60 mA is about right. Audio FET's need less bias than a simple class AB 1 watt amp using BD135/136. They do not need the 5 V some think ( said so many times ). Typical is 1.2 V gate to gate. Even 0V works. This gives about 3 mA which starts to tidy the crossover region. Rod of ESP shows that and a reasonable result to 10 kHz with 0V bias is possible. That surely is food for thought. The N/P curves never truely match so higher bias should be better sounding.
One idea I had recently is simple current limiting resistors in the collectors of cheap bipolars. Reason being of all the protection methods I know of it really works. The limiters used on power amps never seem as good as one would think, if you dare test what I have just said. This isn't a hi fi application. Anyone like to say why not use this? I think that reason will be much like the FET. I built a 500 V PSU that when it choose died. Nowhere in the circuit was an obvious over current avenue. After every trick in the book I choose to loose 5V ( 100 mA ) with 47R on the drain side. I used a 0.6 watt MRS25 to see if it could cope. It did. Never had a problem again. The device was rated 21A peak. Limiting it to 11 A worked. Obviously somehow it was going > 21 amps. Apart from a 47nF 10R snubber there was nothing to say how it could happen. That 10R was the problem for a few nonoseconds. The only other loads were valves that are as gentle as you could wish. Normally the rise of the PSU stopped it being a problem. It took a few days usually to show when no resistor fitted.
BTW. On the 500V PSU I tried a 60V T03 N type Darlington protected by a zener emitter to collector. Although it made some funny noises in the amplifier ( transformer I guess ) it worked fine. The FET worked far better. This made me realise that even though on paper the Darlington was the better choice the real result favoured the FET. Reason was simple. The FET would work down to near zero gate current. This meant the voltage reference followed ( source follower ) was cleaner. The results for about $5 extra was from about -65 to - 89 dB hum ( 100 Hz , 85 dB , 50 Hz. subjectively the 100 is always louder ) . SE amps have no hum rejection ability.The FET T0220FP meaning nothing special needed to clamp it to the chassis.
Now a suggestion that doesn't come from me. Peak current of 20 A real music on everyday speakers is not out of the question. My Magnepans shouldn't do that as they claim 4R resistive. The Quad 303 not known for it's muscle loves them, although gets very warm.
So to revise my estimate of as I will call it from now on I^R losses. 3 easilly bought Vgs matched device pairs should be about 0R6 loss ( better I suspect ). That means that from the same PSU as much as 270 watt 4R as compared with 150 watt. If the 20A peak is true then it will be easilly heard. I ask forgiveness in advance for what might be the romantic notion of a technical journalist. Mr Waynes 1000 watts 1R now has a place of worship.
One thing I will have to think through is this. If the bias is turned up on single pair of audio FET's ( BUZ900/905 , Exicon 10N/P20 ) at 100 mA they are said to take on typical FET rising resistance with load. I am " almost " certain I hear that. A trade between open sound and punch. Certainly at the 5 watts most are said to listen at maybe I imagine it. Problem is 5 watts isn't my typical level. I tend to have the system on and listen all over the house. The Maggies would die so I built some monsters. 20 mA when FET's will produce excellent measurements. I suspect 60 mA is about right. Audio FET's need less bias than a simple class AB 1 watt amp using BD135/136. They do not need the 5 V some think ( said so many times ). Typical is 1.2 V gate to gate. Even 0V works. This gives about 3 mA which starts to tidy the crossover region. Rod of ESP shows that and a reasonable result to 10 kHz with 0V bias is possible. That surely is food for thought. The N/P curves never truely match so higher bias should be better sounding.
One idea I had recently is simple current limiting resistors in the collectors of cheap bipolars. Reason being of all the protection methods I know of it really works. The limiters used on power amps never seem as good as one would think, if you dare test what I have just said. This isn't a hi fi application. Anyone like to say why not use this? I think that reason will be much like the FET. I built a 500 V PSU that when it choose died. Nowhere in the circuit was an obvious over current avenue. After every trick in the book I choose to loose 5V ( 100 mA ) with 47R on the drain side. I used a 0.6 watt MRS25 to see if it could cope. It did. Never had a problem again. The device was rated 21A peak. Limiting it to 11 A worked. Obviously somehow it was going > 21 amps. Apart from a 47nF 10R snubber there was nothing to say how it could happen. That 10R was the problem for a few nonoseconds. The only other loads were valves that are as gentle as you could wish. Normally the rise of the PSU stopped it being a problem. It took a few days usually to show when no resistor fitted.
BTW. On the 500V PSU I tried a 60V T03 N type Darlington protected by a zener emitter to collector. Although it made some funny noises in the amplifier ( transformer I guess ) it worked fine. The FET worked far better. This made me realise that even though on paper the Darlington was the better choice the real result favoured the FET. Reason was simple. The FET would work down to near zero gate current. This meant the voltage reference followed ( source follower ) was cleaner. The results for about $5 extra was from about -65 to - 89 dB hum ( 100 Hz , 85 dB , 50 Hz. subjectively the 100 is always louder ) . SE amps have no hum rejection ability.The FET T0220FP meaning nothing special needed to clamp it to the chassis.
Nige, on Wayne's 1kW/1 Ohm - carfeul with it, it is in fact a can of worms.
Fot a start, when have you ever seen a real life loudspeaker rated at 1 Ohm? I have never seen it, but of course, it's quite possible they exist or can be made. The worst I have ever encountered in real life was abig Infinity Refernce speaker, which managed to dip below 2 Ohms, and had a reputation that only the Mark Levinsons and Krells of this world could drive it porperly Also AR 3a Improved and Yamaha NS-1000 had reputations for being very hard to drive. One AR owner I am in touch with, and indeed, his speakers were hard to drive. The only amp I have that did the job is H/K PA2400, nominally 170W/8 Ohms, and advertised as capabale o delivering 100A peak currents (short term, of course). It got hotter than usual, but did the job.
Now let's look at the wattage we need. The vast majority of loudspeakers are rated up to 100-150W of power, although some American speakers quote 300W and over in non repetative impulses. Anyone want to take chances with such peaks? Never mind the actual SPLs you might reach at such levels, and efficiency will play a role in it, like it or not.
Thinking back to Wayne, if you remember, I suggested a design for his speakers, and according to me, they required 8 pairs of 200W output devices per channel to make his grade. Now work out the losses on such a power stage and work out such a stage's damping factor (don't work out the size and cost of its power supply or you'll get a stroke, a bit more than "a couple of quid").
Technically, such power capability may be impressive, but reasonable (in cost and size) it is not. Now go back and work out how much current you need for some good quality German made loudspeakers, from people like Magnat, Quadral, even Canton, etc, and you start running into problems, They are mostly rated at 4 Ohms, however when you factor in something like 1...2uF in parallel coming from the crossovers, already you are talking 10-20A of CONTINUOUS current you will need..
Put it all together and you have a pretty good case not to use speakers with a nominal impedance of much less than 8 Ohms. Focal and B&W clearly state that their nominally 8 Ohm speakers have minimums of less than 3.5 Ohms. That's why one of the design criteria for my speakers was an impedance minimum of 6 Ohms with a phase shift of no more than 25 degrees. What does that make Wayne's speakers, amp killers?
Fot a start, when have you ever seen a real life loudspeaker rated at 1 Ohm? I have never seen it, but of course, it's quite possible they exist or can be made. The worst I have ever encountered in real life was abig Infinity Refernce speaker, which managed to dip below 2 Ohms, and had a reputation that only the Mark Levinsons and Krells of this world could drive it porperly Also AR 3a Improved and Yamaha NS-1000 had reputations for being very hard to drive. One AR owner I am in touch with, and indeed, his speakers were hard to drive. The only amp I have that did the job is H/K PA2400, nominally 170W/8 Ohms, and advertised as capabale o delivering 100A peak currents (short term, of course). It got hotter than usual, but did the job.
Now let's look at the wattage we need. The vast majority of loudspeakers are rated up to 100-150W of power, although some American speakers quote 300W and over in non repetative impulses. Anyone want to take chances with such peaks? Never mind the actual SPLs you might reach at such levels, and efficiency will play a role in it, like it or not.
Thinking back to Wayne, if you remember, I suggested a design for his speakers, and according to me, they required 8 pairs of 200W output devices per channel to make his grade. Now work out the losses on such a power stage and work out such a stage's damping factor (don't work out the size and cost of its power supply or you'll get a stroke, a bit more than "a couple of quid").
Technically, such power capability may be impressive, but reasonable (in cost and size) it is not. Now go back and work out how much current you need for some good quality German made loudspeakers, from people like Magnat, Quadral, even Canton, etc, and you start running into problems, They are mostly rated at 4 Ohms, however when you factor in something like 1...2uF in parallel coming from the crossovers, already you are talking 10-20A of CONTINUOUS current you will need..
Put it all together and you have a pretty good case not to use speakers with a nominal impedance of much less than 8 Ohms. Focal and B&W clearly state that their nominally 8 Ohm speakers have minimums of less than 3.5 Ohms. That's why one of the design criteria for my speakers was an impedance minimum of 6 Ohms with a phase shift of no more than 25 degrees. What does that make Wayne's speakers, amp killers?
Your both right. It's Mr Wayne remembered day today.
About blind listening test. I don't give a fig what other people think or do. I only know my own mind on this.
I had a Nakamichi Dragon to play with. Because everything can be calibrated Dolby becomes a nice thing to have. The results on cheap Maxell Metal tapes were as if a Revox or even the digital one hoped for in use.
AB testing seemed to prove the signal from tape or original were enough the same to fool nearly anyone.
For a laugh I used some TDK D as it is said to be the tape Nakamichi used for calibration when the cheaper BX 100. To my shock the D was good enough. So I recorded a lot of tapes. In the morning I listened again. They were awful. Soggy and very uninteresting. How had this happened? With heaphones on and constant reference to a better source the brain didn't have time to stop hearing the original. If I am right about this the whole concept of blind tests is a test of tests and not of music. An unfortunate example of what looks like sceince that is in fact a test of fooling the senses. I hope this is a warning to people not to accept things that look like sceinece actually are.
About blind listening test. I don't give a fig what other people think or do. I only know my own mind on this.
I had a Nakamichi Dragon to play with. Because everything can be calibrated Dolby becomes a nice thing to have. The results on cheap Maxell Metal tapes were as if a Revox or even the digital one hoped for in use.
AB testing seemed to prove the signal from tape or original were enough the same to fool nearly anyone.
For a laugh I used some TDK D as it is said to be the tape Nakamichi used for calibration when the cheaper BX 100. To my shock the D was good enough. So I recorded a lot of tapes. In the morning I listened again. They were awful. Soggy and very uninteresting. How had this happened? With heaphones on and constant reference to a better source the brain didn't have time to stop hearing the original. If I am right about this the whole concept of blind tests is a test of tests and not of music. An unfortunate example of what looks like sceince that is in fact a test of fooling the senses. I hope this is a warning to people not to accept things that look like sceinece actually are.
I was recording my sisters singing a few decades back. It was a 4000D with cheap highish imdedance dynamic mics.
I took the machine to my bedroom and was starting to review the recording on my cheap headphones, Red Devils.
Dad walked in and spoke.
I instantly turned to answer him, but he was not there ! He was on the tape !
It was so lifelike that I had turned to look out the window instead of turning to face my bedroom door. It was that convincing.
That is what Hi Fidelity is all about, being lifelike.
Much of todays hifi cannot approach that realism.
I took the machine to my bedroom and was starting to review the recording on my cheap headphones, Red Devils.
Dad walked in and spoke.
I instantly turned to answer him, but he was not there ! He was on the tape !
It was so lifelike that I had turned to look out the window instead of turning to face my bedroom door. It was that convincing.
That is what Hi Fidelity is all about, being lifelike.
Much of todays hifi cannot approach that realism.
Actually, a DBT will do a good job of informing one what is the average view of evaluation of whatever is being tested in the ears and minds of the panel. They can give a good result in say picking out the best liked amp in a group. Everything else is quasi-science. Change the speakers used and you may well arrive at completely different set of results. That average, however right or wrong, has nothing to do with my room, my hearing and my system.
We agree, Nigel.
Last edited:
I have experimented for a long time with various clamp on filters.
I have also subjected myself to blind testing of sorts.
I have loaded two tracks onto a thumb drive and allowed them to play on repeat.
I have then allowed the tracks to play with volume muted so that I have no visual clue to which recording is playing.
When volume is returned to normal listening level I have then been able to perfectly reliably identify which recording is which...ie filter applied or not.
This is with recordings that I am perfectly familiar with, and system that I am perfectly familiar with.
So, under the right conditions ABX is valid, but I contend that ABX may not be valid when the above conditions are not fully met.
Dan.
http://www.michaelgerzonphotos.org.uk/articles/Why Coincident Microphones.pdf
I think it was Stuart Lee the comdedian ( forgive me if wrong Stuart ) who printed a retraction of a statement he made about Michael Gerzon. That was to be careful because Oxford had this man who would be looking to corner you. Well as much as I thought Micael wonderful he did do that. What a contrast to how he writes. I know we vaguely knew each other and maybe he made me think the way I do. Whatever the reason within my skillls of understanding he speaks for me. Some call him Oxfords unknown Hawking. Note how a man of goodness knows what IQ can be so simple in how he transmits ideas.
Michael's most interesting paper that I read was about how seemingly identical filters sound different. He made it simple to see. These ideas are now well understood and vital.
I think it was Stuart Lee the comdedian ( forgive me if wrong Stuart ) who printed a retraction of a statement he made about Michael Gerzon. That was to be careful because Oxford had this man who would be looking to corner you. Well as much as I thought Micael wonderful he did do that. What a contrast to how he writes. I know we vaguely knew each other and maybe he made me think the way I do. Whatever the reason within my skillls of understanding he speaks for me. Some call him Oxfords unknown Hawking. Note how a man of goodness knows what IQ can be so simple in how he transmits ideas.
Michael's most interesting paper that I read was about how seemingly identical filters sound different. He made it simple to see. These ideas are now well understood and vital.
If people don't want to read Michaels paper I will take a subject. On his system that was not unlike mine he prefered 78's very often. His reason is simple. The interpaly is often more like real life. Mostly Beethoven in real life is like a musical garden coming alive. Even that lack of ultra HF is more like 78's When the same musicains then play Mahler on the same night it sounds more like LP or good CD. In this one specific case I can honestly say no hi fi I have ever heard sounds like Beethoven. And when not hi fi Beethoven must be my favourite music.
This is close.
https://www.youtube.com/watch?v=MPSSx4jYEmg
This is close.
https://www.youtube.com/watch?v=MPSSx4jYEmg
Seems reasonable to me. Hifi is not exact science, we don't have to make a rocket landing on a comet at milions of kilometers from our planet. It is for the unique pleasure of our ears. Or, sometimes, for some of us, for our job.
So choosing what we "prefer" is the only goal.
Peeeking ? Well, if the look, the Brand name or whatever can influence our jugement, is-it not a part of our pleasure ?
Now, when it is about chosing between two technical solutions, for an audio designer, what the difference between knowing or not which solution is A or B ? Did-ve want to fool ourselves ?If we have a prejudice, better to work on our personal behavior to correct this attitude.
Listening to music, or equipment to reproduce music is something to be learned. Ability grows with culture. It is easier to recognize a face in a crowd if it is your son or a friend.
Last, when we compare two gears, in order to classify them, or the difference is obvious, reproductive and we don't need to organize a graduation ceremony, or the difference is so subtle that we need many comings and goings. In this last case, I consider it is Draw. As, in the real life, the difference between the two gears will depend more of the temperature, air pressure, humidity, acoustic of your room or your mood...
Last, and to answer at previous question about studio amplifiers, the best one is the same than you: the one that require less attention to separate instruments, respect of the dynamic and micro details, and bring less fatigue.
Of course, attention has to be done to the reproduction of the basses (damping) and fluid trebles, in order to avoid mistakes when you set the levels or equalization of some instruments (Bass, kick drum, cymbals).
As no amp is perfect, we have to make a choice between pro and cons between two amps, as an example, but this is OUR problem (A sport car or a limousine ?) that has nothing to do with blind listening.
Now, if you are in the process to give a note to various gears, for publication as an example, you are in a very different situation. You cannot lie on your own preferences, so you have to organize listening sessions with enough participants to have good statistic, and, of course, blind will be a requisite.
Last edited:
Agreed.
What the "Damn the DBTs, I'll do it my way" crowd probably don't understand is that realism that is widely recognized won't ever be found without evaluating audio gear with DBTs.
I've worked in the F&B business parttime for quite a few years.
3 of 4 high dollar eaters can't tell fancy grub and fancy drinks apart from tin food and cat pee.
Over the decades, I've spoken about it on occasion with knowledgeable folks in the restaurant world in close to every nation I visited, in their own lingo, they all agreed.
Gastronomy is fashionable, globally for the last 2.5 decades**, folks with money like to think they can buy themselves into savoir de vivre.
As goes for high performance audio.
My next door neighbor is a real estate guy, likes to drive a top of the line Range Rover, and wear a fancy suit during week days.
In the weekends, he goes solid gold neck-chain with an open shirt, floats in his pool all day on loads of cheap rum, and smokes two boxes of cigarettes while he listens to his favorite songs from the Dominican Republic.
I gathered he already has serious health issues for years, and fake teeth in spite of being several years younger than me, not likely to become a longrunner.
But at least he's frank (no, not that frank), a barrel of laughs, and I'll never ever invite him over for dinner and drinks.
(** There are solid reasons which triggered it, but too lengthy a story to post. Watch e.g. the BBC's 2-part Super-Rich and Us, by Jacques Peretti)
Last edited:
Probably the best analogy on the matter that I have read.
Having been responsible for a few hundred pictures in a just published book, some my own photography, others from 3rd party sources (with no more than some manipulation) by me, I know first hand that no matter how good these illustrations appear on a high definition monitor, they look to be comparatively, without much life when printed in a book. I consider that the printed versions have limited dynamics, less focus and impact than the originals but most of all they lack vitality. That is similar to the difference between a well mixed master and an average pressing for retail outlets. The difference between the perceived original (in this case landscapes) and the best re-mixed digital screen image capture is even greater...and we know that the original 'take' can never be more than an image...not a facsimile of the original scene.
I also have quite a few analogue ‘insurance copy’ ¼” two track master tapes of reasonably well known records. These I have played a few times on a Revox and had to stop doing so as they made my normal analogue disk records sound, by comparison, extremely inferior, depite the fact that I was playing them on a very high quality TT/Arm/Cartridge combination….the rest of the chain was common to both. The differences are similar to those noted in the paragraph above….dynamics, clarity/focus contrast etc.
Like Nigel in his examples regarding paintings (above), a good few years ago I saw the major Picasso retrospective at the South Bank in London. I was already familiar with many of the exhibits from prints in books, yet the impact of viewing the originals, despite many of them having faded to an appalling degree, was totally shocking to my preconceptions gained from the printed images seen in books. Even the simple ink sketches had an impact far beyond their imagined capabilities.
Much of my spare evenings were spent listening to live music in about as broad a variety of locations, performers and genre as can be imagined. Luck has had it that I also have been lucky to have been able to borrow for extended periods some of the then best audio equipment available; many pieces being prototype equipment.
We all know that even the best (in the view of reviewers and/ EEs) can never get close to the original despite the concerted efforts of everyone in the chain of recording events applying their really finest efforts at every stage of the process.
For me the above is more than ample to accept the inevitable conclusion that recordings be they in print, on tape/digital media etc are each and all imperfect and will never be capable of re-capturing the original.
Rather than take any existing form of equipment testing as a reference to human perception of that equipment it would be far more interesting - and useful - to take a representative cross section of music lovers to three or four totally differing live performances given in a variety of locations and to then interview these same people on their perceptions, and most of all, on what elements of the live performances were most important to them.
Such a test, repeated with varying groups of listeners, would give the audio designers, and maybe the studio people, quite a different set of targets to hit than would even the best test equipment and EE (as known) skills.
Last edited:
The self-delusion, myth and even libel that is being expounded here is that audio engineers don't listen to live music and use their listening experiences for a reference in their work.
Take for example engineers that also have years of experience recording live music, such as myself...
- Status
- Not open for further replies.