When you can produce some evidence, thoriated, I will start to take you seriously.
I use higher bitrates and bitdepths for recording as it offers more headroom before mastering, but the most recent tests show that a panel of highly qualified listeners were unable to distinguish between SACD and CD in blind testing. Go have a look in RocketScientist's headphone thread, or on his website.
Now, either point me to a reference that shows different, or can we get back to the subject of digital amplifiers?
w
I use higher bitrates and bitdepths for recording as it offers more headroom before mastering, but the most recent tests show that a panel of highly qualified listeners were unable to distinguish between SACD and CD in blind testing. Go have a look in RocketScientist's headphone thread, or on his website.
Now, either point me to a reference that shows different, or can we get back to the subject of digital amplifiers?
w
I.e. "Give me right answer from multiple choices". 😀
It works in textbooks, but does not work in real life where everything is based on models obtained by empirical processes of learning. What is right on one level of abstractions is wrong on another layer of abstractions. What works to get one answer does not work for getting other answers, that's why many models even in the single discipline exist.
Funny thing is, that hunt/gather societies work 2 or 3 hours per day, enjoy a greater sense of "fulfillment", and have no ruling class.
That horse you're riding looks awful high up.
Hi,
And they way, way understate the disregard for traditional measurements real engineers** SHOULD HAVE in audio.
No available recording microphone known to me* has a noisefloor lower than 20dBA SPL Equivalent and there is not much to record in mustic that 120 - 130dB SPL.
No high end speaker or studio monitor known to me* offers a THD of less than 0.1% from 20Hz to 20KHz at 1 Watt input, at rated power distortion levels well above 1% are common place and thermal compression of 3 - 6dB is par for the course.
Knowing these simple facts it seems incredible that anyone in their right minds would even debate the point of 0.01% or 0.001% THD in an amplifier at full power.
Measurements in themselves are simply data, not information. They carry in themselves no intrinsic value and something that measures more or less on a given scale does not tell us anything.
To turn data into information we need to place it into a context that gives the data meaning. Such a context surely must take account of such simple facts as discussed above and of the way humans hear.
I do perform measurements on my gear (an extremely wide range, actually), but I perform them to confirm the gear behaves as designed, not to infer absolute quality from this.
Practically all currently used measures in attempting to quantify the performance of audio products are utterly meaningless in any use other than advertising. They offer absolutely ZIP, ZILCH, NADA information on sound quality and what is a "better" measured result in a technical sense has shown practically ZERO correlation with what "sounds good".
Sure, there are some limits in this measured performance which it would imprudent to exceed (though them being exceeded does not reliably cause "bad sound" either), but debating the goodness of 0.01% THD or 0.001% THD is like debating if the fatness of a pig is more or less green than the designated hitter rule.
Ciao T
* Engineering is the process of getting what you need from you can get. What you need is determined by the purpose of the thing being engineered.
If this purpose is to achieve the lowest measured imperfection when connected to Lab Gear one set of goals are established which have little or no congruence with those goals established if we want: "to connect the device to some speakers and maximally enjoy listening to recorded music".
** the clause "Known To Me" is simply insurance against someone jumping up and exclaiming "but this Microphone of mine has 19.9 dBA SPL self noise, you are wrong!" or someone jumping up and exclaiming this "<insert excotic driver here that is not full range anyway> has less than 0.1% THD at 3KHz, you are wrong!".
Essentially the existence of single or small quantity examples that perform better than my statements do not invalidate the point being made.
And they way, way understate the disregard for traditional measurements real engineers** SHOULD HAVE in audio.
No available recording microphone known to me* has a noisefloor lower than 20dBA SPL Equivalent and there is not much to record in mustic that 120 - 130dB SPL.
No high end speaker or studio monitor known to me* offers a THD of less than 0.1% from 20Hz to 20KHz at 1 Watt input, at rated power distortion levels well above 1% are common place and thermal compression of 3 - 6dB is par for the course.
Knowing these simple facts it seems incredible that anyone in their right minds would even debate the point of 0.01% or 0.001% THD in an amplifier at full power.
Measurements in themselves are simply data, not information. They carry in themselves no intrinsic value and something that measures more or less on a given scale does not tell us anything.
To turn data into information we need to place it into a context that gives the data meaning. Such a context surely must take account of such simple facts as discussed above and of the way humans hear.
I do perform measurements on my gear (an extremely wide range, actually), but I perform them to confirm the gear behaves as designed, not to infer absolute quality from this.
Practically all currently used measures in attempting to quantify the performance of audio products are utterly meaningless in any use other than advertising. They offer absolutely ZIP, ZILCH, NADA information on sound quality and what is a "better" measured result in a technical sense has shown practically ZERO correlation with what "sounds good".
Sure, there are some limits in this measured performance which it would imprudent to exceed (though them being exceeded does not reliably cause "bad sound" either), but debating the goodness of 0.01% THD or 0.001% THD is like debating if the fatness of a pig is more or less green than the designated hitter rule.
Ciao T
* Engineering is the process of getting what you need from you can get. What you need is determined by the purpose of the thing being engineered.
If this purpose is to achieve the lowest measured imperfection when connected to Lab Gear one set of goals are established which have little or no congruence with those goals established if we want: "to connect the device to some speakers and maximally enjoy listening to recorded music".
** the clause "Known To Me" is simply insurance against someone jumping up and exclaiming "but this Microphone of mine has 19.9 dBA SPL self noise, you are wrong!" or someone jumping up and exclaiming this "<insert excotic driver here that is not full range anyway> has less than 0.1% THD at 3KHz, you are wrong!".
Essentially the existence of single or small quantity examples that perform better than my statements do not invalidate the point being made.
Wisdom indeed but sadly most humans don't spend time in quiet contemplation.Analysis for the sake of analysis.Like a spiral within spiral...like the windmills of your mind...
Do you mean, nobody can tell whether the crowd is recorded in church, in the forest, or on a stadium, because reverberation level is way below the "noise floor" of their voices? 😉
The difference between noise floor in measurement and in hearing is, hearing recognizes sound patterns way below noise floor. Measurement equipment does not.
This is interesting,I guess if you use scientific rationale,you need more than one source or sensor to confirm it's origin like stereo microphones to have a sense of space or distance.So for a human nature provides multiple sensors for determining where or what of the environment.If say if no one tell you the recording is done at a particular location you have other ques if you there like eyesight to see where or what.Ears to hear and compare to your sonic databank.Nose to smell and touch to feel textures and of course not forgetting your sixth sense or intuition which can be developed by proper training.The last in in the realm of stargate and physic phenomena.Mechanical devices are just mechanical devices and cannot replace the faculties of a human so in essence we "interpolate" the data.
Much is still unknown about how humans operate at the "super" level.
Hi,
No, first you will find that the noisefloor of capacitor microphones (unlike most electronics) falls with a first order function as frequency rises, so it is dominated by low frequencies.
Secondly, once the reverb tail is low enough in level you cannot hear it, full stop. Try speaking very quietly in that echoing church. Even this quiet noise is subject to the same reverbration as speaking loud, yet the echo and reverb of one is audible, yet not that of the other.
That depends for the human hearing very much on the absolute SPL and the frequency. For Test equipment testing steady or repetitive patterns, using massive averaging too can "listen below the noisefloor".
The bottom line is that real recorded music has a quiet constrained dynamic range, due to technological limitation at least, more often due to heavy-handed compression, frequently due to deliberate attempts to make the music sound "loud".
Equally, maximum SPL's with low distortion are severely limited by the speakers used. Under those conditions making super low distortion amplifiers with extremely low noisefloor has no merit whatsoever.
At best (if the low[er] distortion was not gotten by trading something else (bandwidth, stability, slewrate etc.) it will not cause any problems, but will not produce additional merit.
At worst (if something that impacts on audible "good sound" was degraded in order to get low distortion) it will produce equipment that is audibly inferior.
I thought the "how many zeros after the decimal point can we get" thinking regarding THD was discredited sufficiently in the 80's, but alas not... Well, as the good man said, those who do not know their history are doomed to repeat it.
I remember my history and building some stuff that was really super low THD in the early 80's, normal meters could not even measure THD & N, we had come up with extra low noise preamp's for them, yet it sounded so bad, it was unbelievable...
At the same time a really primitive AC coupled (electrolytic cap on in and out, single rail) simple amp with just a single-ended pnp input stage and and a single npn transistor VAS with a bootstrapped resistor load and a quasi complementary class AB output sounded great and that despite only having 25W unclipped, with around 1% THD at those 25W sounded just great and excellent.
While my super duper amp was of course something silly like 200W or 300W and had nearly unmeasurable THD at that power level and used a 0.25V/uS slew rate op-amp at the input with Sziklay tripples with gain (~ 10) as powerstage using 3055/2955 equivalents and more feedback than you can shake a stick at (I was young and foolish then, now I am old and foolish) sounded so rotten, I pulled it apart and never again tried something so blatantly stupid.
Ciao T
No, first you will find that the noisefloor of capacitor microphones (unlike most electronics) falls with a first order function as frequency rises, so it is dominated by low frequencies.
Secondly, once the reverb tail is low enough in level you cannot hear it, full stop. Try speaking very quietly in that echoing church. Even this quiet noise is subject to the same reverbration as speaking loud, yet the echo and reverb of one is audible, yet not that of the other.
That depends for the human hearing very much on the absolute SPL and the frequency. For Test equipment testing steady or repetitive patterns, using massive averaging too can "listen below the noisefloor".
The bottom line is that real recorded music has a quiet constrained dynamic range, due to technological limitation at least, more often due to heavy-handed compression, frequently due to deliberate attempts to make the music sound "loud".
Equally, maximum SPL's with low distortion are severely limited by the speakers used. Under those conditions making super low distortion amplifiers with extremely low noisefloor has no merit whatsoever.
At best (if the low[er] distortion was not gotten by trading something else (bandwidth, stability, slewrate etc.) it will not cause any problems, but will not produce additional merit.
At worst (if something that impacts on audible "good sound" was degraded in order to get low distortion) it will produce equipment that is audibly inferior.
I thought the "how many zeros after the decimal point can we get" thinking regarding THD was discredited sufficiently in the 80's, but alas not... Well, as the good man said, those who do not know their history are doomed to repeat it.
I remember my history and building some stuff that was really super low THD in the early 80's, normal meters could not even measure THD & N, we had come up with extra low noise preamp's for them, yet it sounded so bad, it was unbelievable...
At the same time a really primitive AC coupled (electrolytic cap on in and out, single rail) simple amp with just a single-ended pnp input stage and and a single npn transistor VAS with a bootstrapped resistor load and a quasi complementary class AB output sounded great and that despite only having 25W unclipped, with around 1% THD at those 25W sounded just great and excellent.
While my super duper amp was of course something silly like 200W or 300W and had nearly unmeasurable THD at that power level and used a 0.25V/uS slew rate op-amp at the input with Sziklay tripples with gain (~ 10) as powerstage using 3055/2955 equivalents and more feedback than you can shake a stick at (I was young and foolish then, now I am old and foolish) sounded so rotten, I pulled it apart and never again tried something so blatantly stupid.
Ciao T
any more mouldy strawmen to wave at "idiot" engineers?
actually Bob Cordell's use of THD is at 20KHz - reveals more problems than 1 KHz THD
and it is only one test - also having smoothly decreasing distortion of all orders with reducing signal level is quite easy with high negative feedback - just use a high bias AB output stage instead of vainly hoping feedback can "clean up" under bias static/dynamic deadzone
so "modern" "ppm THD" is not the same as 70's-80's intentional gaming the THD # by only specing it at full swing
Halcro has had some market sucess
some here appear to want to treat EE and Audio engineering knowledge like the "witch":
The Witch: I'm not a witch I'm not a witch!
Sir Bedevere: But you are dressed as one
The Witch: *They* dressed me up like this!
Crowd: We didn't! We didn't...
The Witch: And this isn't my nose. It's a false one.
Sir Bedevere: [lifts up her false nose] Well?
Peasant 1: Well, we did do the nose.
Sir Bedevere: The nose?
Peasant 1: And the hat, but she is a witch!
Crowd: Yeah! Burn her! Burn her!
Sir Bedevere: Did you dress her up like this?
Peasant 1: No!
Peasant 3, Peasant 2: No!
Peasant 3: No!
Peasant 1: No!
Peasant 3, Peasant 2: No!
Peasant 1: Yes!
Peasant 2: Yes!
Peasant 1: Yeah a bit.
Peasant 3: A bit!
Peasant 1, Peasant 2: A bit!
Peasant 2: a bit
Peasant 1: But she has got a wart!
actually Bob Cordell's use of THD is at 20KHz - reveals more problems than 1 KHz THD
and it is only one test - also having smoothly decreasing distortion of all orders with reducing signal level is quite easy with high negative feedback - just use a high bias AB output stage instead of vainly hoping feedback can "clean up" under bias static/dynamic deadzone
so "modern" "ppm THD" is not the same as 70's-80's intentional gaming the THD # by only specing it at full swing
Halcro has had some market sucess
some here appear to want to treat EE and Audio engineering knowledge like the "witch":
The Witch: I'm not a witch I'm not a witch!
Sir Bedevere: But you are dressed as one
The Witch: *They* dressed me up like this!
Crowd: We didn't! We didn't...
The Witch: And this isn't my nose. It's a false one.
Sir Bedevere: [lifts up her false nose] Well?
Peasant 1: Well, we did do the nose.
Sir Bedevere: The nose?
Peasant 1: And the hat, but she is a witch!
Crowd: Yeah! Burn her! Burn her!
Sir Bedevere: Did you dress her up like this?
Peasant 1: No!
Peasant 3, Peasant 2: No!
Peasant 3: No!
Peasant 1: No!
Peasant 3, Peasant 2: No!
Peasant 1: Yes!
Peasant 2: Yes!
Peasant 1: Yeah a bit.
Peasant 3: A bit!
Peasant 1, Peasant 2: A bit!
Peasant 2: a bit
Peasant 1: But she has got a wart!
Last edited:
Hi,
I have amplifiers that have high THD at both 1KHz and 20Khz, equally high THD, to be precise. So with all respect to Robert, 20KHz THD vs. 1KHz THD in some specific context may be more revealing, but on it's own, as single number number THD20 is as meaningless as THD1.
So had Cary.
Not quite, I am just dead tired of opinion or indeed belief being presented as fact without the least proof or shred of evidence in a forcefull manner by those who will absolutely take others to task for doing the same in areas where they disagree with the position presented.
Or as Robert Nesta Marley used to say: "hypocrites and pharisees"
Now if we can agree on live and let live as in "Let a hundred flowers bloom; let a hundred schools of thought contend." we may be getting closer to "seeking truth from facts" and actually get somewhere.
Note, I am listening to the Meter Readers (actually, I trained as one and found it hard to accept that all the measurements where useless and that 0.00001% THD did not make an Amplifier a "good amplifier") and incorporate what I learn from them with all the knowledge I gather and sythesize new things from that.
Ciao T
I have amplifiers that have high THD at both 1KHz and 20Khz, equally high THD, to be precise. So with all respect to Robert, 20KHz THD vs. 1KHz THD in some specific context may be more revealing, but on it's own, as single number number THD20 is as meaningless as THD1.
So had Cary.
Not quite, I am just dead tired of opinion or indeed belief being presented as fact without the least proof or shred of evidence in a forcefull manner by those who will absolutely take others to task for doing the same in areas where they disagree with the position presented.
Or as Robert Nesta Marley used to say: "hypocrites and pharisees"
Now if we can agree on live and let live as in "Let a hundred flowers bloom; let a hundred schools of thought contend." we may be getting closer to "seeking truth from facts" and actually get somewhere.
Note, I am listening to the Meter Readers (actually, I trained as one and found it hard to accept that all the measurements where useless and that 0.00001% THD did not make an Amplifier a "good amplifier") and incorporate what I learn from them with all the knowledge I gather and sythesize new things from that.
Ciao T
This is exactly why I do measure my stuff, I want it to perform how it's supposed to be able to perform. Anything less and my implementation is flawed.
I know I most likely wont be able to hear the difference if the flaws are raising the THD from a theoretical 0.0005% to 0.005%, but that doesn't matter. If something is cited to perform to a certain level I want to get that performance out of it - I see this as the only logical way to design things.
I don't mind high THD if that's how the device is supposed to perform. What I mind is high THD when it isn't.
Hi,
I typicially design gear brutally minimalist, minimum number of stages, parts etc., massive power supplies though, as little feedback as possible (mostly zero for Class A circuitry) and as a result RELATIVELY (as in relative to for example a Chipamp or a Self Blamless) high THD is unavoidable, however this "THD" follows patters with excellent auditory masking and is still in all cases less than any speaker I know.
So I do not as such design for relatively high THD, but if the tradeoff is a bit more 2nd & 3rd HD for simplification and improved subjective sound quality I'm not loosing sleep over the increase in HD (where I do loose sleep majorly though is when noise loading, IMD and HD measurements give results that differ much).
Ciao T
I typicially design gear brutally minimalist, minimum number of stages, parts etc., massive power supplies though, as little feedback as possible (mostly zero for Class A circuitry) and as a result RELATIVELY (as in relative to for example a Chipamp or a Self Blamless) high THD is unavoidable, however this "THD" follows patters with excellent auditory masking and is still in all cases less than any speaker I know.
So I do not as such design for relatively high THD, but if the tradeoff is a bit more 2nd & 3rd HD for simplification and improved subjective sound quality I'm not loosing sleep over the increase in HD (where I do loose sleep majorly though is when noise loading, IMD and HD measurements give results that differ much).
Ciao T
Hi,
Did you miss it?
The argument is if the fatness of the Pig is more or less green than the designated hitter rule.
And if Kubelik's Mahler Lied von der erde is more or less authentic than Abbado's Mussorsky/Ravel Pictures at an exhibition.
Ciao T
Strange. I know you guys are arguing about something, but what is it?
Did you miss it?
The argument is if the fatness of the Pig is more or less green than the designated hitter rule.
And if Kubelik's Mahler Lied von der erde is more or less authentic than Abbado's Mussorsky/Ravel Pictures at an exhibition.
Ciao T
I wasn't arguing about anything. What I was trying to say is that measurements are or should be wholeheartedly encouraged as a way of making sure that your design isn't faulty anywhere.
There are some people who say, I don't need to measure anything because I like the way that it sounds. They also almost make it sound like, I don't want to measure it either because what would that tell me? Like they are worried it will show them things they don't want to see. This is what I have a problem with.
There are some people who say, I don't need to measure anything because I like the way that it sounds. They also almost make it sound like, I don't want to measure it either because what would that tell me? Like they are worried it will show them things they don't want to see. This is what I have a problem with.
Hi,
I think that this is perfectly fine as a position, it is a choice. And if they are "in the business" as long as they are candid about it I think it is fine too (even though I personally waste time to measure).
Being from (by now) diverse professional backgrounds, the "Engineer" side in me still wants to know, but if someone says "I don't measure, I like the way it sounds", it's fine with me too. I just hope they are not too upset when I listen, measure and reverse-engineer their designs... 🙂
Ciao T
I think that this is perfectly fine as a position, it is a choice. And if they are "in the business" as long as they are candid about it I think it is fine too (even though I personally waste time to measure).
Being from (by now) diverse professional backgrounds, the "Engineer" side in me still wants to know, but if someone says "I don't measure, I like the way it sounds", it's fine with me too. I just hope they are not too upset when I listen, measure and reverse-engineer their designs... 🙂
Ciao T
Engineering vs Science vs Listening
"I wasn't arguing about anything. What I was trying to say is that measurements are or should be wholeheartedly encouraged as a way of making sure that your design isn't faulty anywhere. "
Absolutely agree with the above statement in the sense that the measurements need to be within a range that one agrees the distortion products will be below the threshold of audability. But this is only the base level of good subjective performance-the Halcro amps mentioned above sound horrendous IMO.
As regards EEs, there is a tendency for many (not all) Engineers (and this applies to engineers in many fields) to be somewhat closed minded, in that they often do not see possibilities beyond the scope of their training. That is, many engineers take their training as physical fact of reality-when the truth is that their training (and medical doctors can be thought of similarly) is just a framework for practical application which generally works pretty well in most situations. Often the most successful and talented audio designers are not EEs by training, and often their background is in the theoretical sciences. In theoretical science, there are no "facts" everything is understood as frameworks for describing reality, and when new paradigms are developed, they may make for the development of a new framework. Theoretical scientists (like many audiophiles) are open minded, as they understand that we actually do not know everything about reality, and they understand that what we believe is true today, may be proven entirely wrong tomorrow.
"I wasn't arguing about anything. What I was trying to say is that measurements are or should be wholeheartedly encouraged as a way of making sure that your design isn't faulty anywhere. "
Absolutely agree with the above statement in the sense that the measurements need to be within a range that one agrees the distortion products will be below the threshold of audability. But this is only the base level of good subjective performance-the Halcro amps mentioned above sound horrendous IMO.
As regards EEs, there is a tendency for many (not all) Engineers (and this applies to engineers in many fields) to be somewhat closed minded, in that they often do not see possibilities beyond the scope of their training. That is, many engineers take their training as physical fact of reality-when the truth is that their training (and medical doctors can be thought of similarly) is just a framework for practical application which generally works pretty well in most situations. Often the most successful and talented audio designers are not EEs by training, and often their background is in the theoretical sciences. In theoretical science, there are no "facts" everything is understood as frameworks for describing reality, and when new paradigms are developed, they may make for the development of a new framework. Theoretical scientists (like many audiophiles) are open minded, as they understand that we actually do not know everything about reality, and they understand that what we believe is true today, may be proven entirely wrong tomorrow.
I must agree to this. The 'D' in class D does not stand for digital but for 'switching'. It's a logical progression from class A to B to C to D...
Bruno Putzeys, who designs class D amps for Hypex, says that class D amps are analog amps because they deal with the analog quantities of current, voltage and time. Not with digital numbers. You can't design a top-notch class D amp unless you are a top-notch analog designer; knowledge of digital doesn't help you squat.
BTW You can find class D amp both with or without global feedback.
jan didden
- Status
- Not open for further replies.