Glen, you are about 30 years behind us. WE have been making the most linear circuits possible for the last 40 years. This is NOT new, in fact, between Nelson Charles and me, you have most of the newer topologies in the last 30 years, that we first developed.
The ONLY thing that has seriously changed, is that we don't like global negative feedback as much as we used to, when we knew less about what makes circuits sound best.
The ONLY thing that has seriously changed, is that we don't like global negative feedback as much as we used to, when we knew less about what makes circuits sound best.
G.Kleinschmidt said:
Oh, my. You are so good at misinterpreting and then putting words into the mouths of others.
Actually, the design of the new circuit started with Peufeu's website that explored the "memory distortion" phenomenon described by Lavardin. The Lavardin amps have an excellent reputation, so I decided to explore this avenue myself. It turned out to be a dead-end for me, but led to some new ways to think about certain circuit topologies. I ended up with something that sounded great and that I was very happy with.
The fact that it has low THD+N is interesting, but certainly not a design goal. I only measure it for three reasons:
a) If it is unacceptably high (say over 1% or 2%), I know that it will make it more difficult to sell the product no matter how good it sounds. I'd rather p*ss downwind than up.
b) I like to know what to expect when Stereophile does measure it.
c) Running a comprehensive suite of measurements on each unit is the easiest way to ensure QC of our products. If they normally measure at 0.01% distortion and a sample measures at 0.05% distortion, then we know that there is most likely an assembly error or bad part.
So, no, it was not a "design aim", either in whole or part.
G.Kleinschmidt said:
No, not really. Below 10 watts or so, it is comparable to our design of two generations ago. But the previous model (the V-5xe) was never measured by Stereophile. The new design is somewhat better, but I wouldn't say "a lot" better.
G.Kleinschmidt said:
Well, if they get it right in the first place, why do they feel the need to apply the band-aids? Do they really think that 0.0001% distortion is going to sound better than 0.001% distortion?
Because if they did bother to listen, they would find that applying the band-aids to a good open-loop design actually makes it sound worse.
Edit: And by the way, I know of no op-amps that can achieve 0.001% open loop distortion. Which leads us back to where we started. In my experience a good discrete circuit will sound better than a good op-amp every time.
Re: Re: Re: Re: "Objectifying the Golden Ear" in latest Electronics Design ...
No need for that. If you write an interesting letter yourself, I'm sure the JA would publish in Stereophile. As I said before, he is an open-minded kind of guy.
I would, however, recommend that you accept my invitation to visit our factory before writing such a letter. Please bring up whatever op-amp based products you feel measure and/or sound good. You can compare them to our zero-feedback, fully discrete designs at your leisure. Currently we have a pair of the TAD Reference One loudspeakers, listing at $60,000 per pair and worth every penny of it. So you won't be limited by the transducers when making your evaluations.
It is much more interesting to read a letter from one who has first-hand knowledge of the subject at hand.
andy_c said:I think it would be only fair for Stereophile to give Bob Pease equal space in the mag to respond to Charles' letter.
No need for that. If you write an interesting letter yourself, I'm sure the JA would publish in Stereophile. As I said before, he is an open-minded kind of guy.
I would, however, recommend that you accept my invitation to visit our factory before writing such a letter. Please bring up whatever op-amp based products you feel measure and/or sound good. You can compare them to our zero-feedback, fully discrete designs at your leisure. Currently we have a pair of the TAD Reference One loudspeakers, listing at $60,000 per pair and worth every penny of it. So you won't be limited by the transducers when making your evaluations.
It is much more interesting to read a letter from one who has first-hand knowledge of the subject at hand.
Re: Re: Re: Re: Re: "Objectifying the Golden Ear" in latest Electronics Design ...
Hmm, the last time I checked, I didn't work for National Semiconductor. My suggestion was meant to be humorous, as Pease's contempt for audiophiles is notorious. He is also a very funny guy, so if he did write such a letter, it would be worth a Stereophile subscription just to read it.
In my view, you have a strange concept of what "knowledge" is. The experience of listening to expensive stereo equipment is not knowledge. It is merely material for endless internet ****ing contests. That may be your thing, but it is not mine.
Charles Hansen said:
Hmm, the last time I checked, I didn't work for National Semiconductor
. My suggestion was meant to be humorous, as Pease's contempt for audiophiles is notorious. He is also a very funny guy, so if he did write such a letter, it would be worth a Stereophile subscription just to read it.In my view, you have a strange concept of what "knowledge" is. The experience of listening to expensive stereo equipment is not knowledge. It is merely material for endless internet ****ing contests. That may be your thing, but it is not mine.
G.Kleinschmidt said:
Hmm. It seems you have given up altogether on making any sort of substantive replies. Instead you just make empty sarcastic comments, personal insults, and continue to put words into other people's mouths.
G.Kleinschmidt said:
Besides your sarcastic exaggerations of my previous post, yes, that is what I basically said. I was not trying to lower the "THD+N" of my previous designs. Nor do I think that the audible performance improvements were directly correlated to the improved "THD+N" measurements.
If I did have those goals, or if I did think there was a direct correlation, I would be using feedback, just like everyone else. That's not so hard to figure out now, is it?
G.Kleinschmidt said:
Yes, trying to "fix" or "improve" the design after the fact is clearly a band-aid. If your goal is low "THD+N", why not design a linear circuit in the first place instead of attempting to correct for its inherent non-linearities after the distortion has been generated? What is so "nonsensical" about calling a spade a spade?
G.Kleinschmidt said:
So the best you can come up with is personal insults? And then continue putting words into my mouth that I never said. Please point out where I said, "Bruce Candy has never listened to his amplifiers."
Let me ask you a couple of questions. Do you think that Bruce Candy listened to his amplifier open-loop? Do you think it would even function open-loop? I'd be willing to wager a large sum that the answer to both of those questions is clearly "no".
Guys, stop pushing buttons- I want you here to talk to, not in the sin bin.
Tonight I went down to the lair, put on my old sealed-to-the-head headphones to drown out ambient noise, and played some music. I also hooked the output up to the scope and looked at peak levels. Backing the attenuator down to where the music was below threshold, and then comparing the level to what I consider a comfortable mildly loud, but not hazardous level, the range was surprisingly small, about 60dB. Let's say I'm 52 and my hearing sucks. Let's also say you like to listen louder than I do- most rockers would. Let's say you also want some headroom. So add another 20-30 dB. As far as I'm concerned, anything happening at 90dB down, or more, is completely irrelevant. At that level, your ability to hear high and low frequencies is completely gone, so forget high harmonics. One can argue the accuracy of Fletcher Munson curves, but not the basic idea. IMHO, the deficiencies and differences of amplifiers, if they really exist in a specific case, are not that hard to see. It may be hard to put a name on them, and you may have to choose your tests wisely to make them obvious, but they won't require SA of -120dBV distortion residuals to see. If you can think of anything I can do to a signal that's 90dB down from that signal, that's audible, including mixing it with a brass band, tell me and I'll try it. Right now I'm completely skeptical of claims that -90dB stuff of any description is audible, detectable, or affects any perception of a performance.
edit- obviously if you crank the gain up for a higher max, everything moves up, but IMO that's not a realistic quality listening situation.
Tonight I went down to the lair, put on my old sealed-to-the-head headphones to drown out ambient noise, and played some music. I also hooked the output up to the scope and looked at peak levels. Backing the attenuator down to where the music was below threshold, and then comparing the level to what I consider a comfortable mildly loud, but not hazardous level, the range was surprisingly small, about 60dB. Let's say I'm 52 and my hearing sucks. Let's also say you like to listen louder than I do- most rockers would. Let's say you also want some headroom. So add another 20-30 dB. As far as I'm concerned, anything happening at 90dB down, or more, is completely irrelevant. At that level, your ability to hear high and low frequencies is completely gone, so forget high harmonics. One can argue the accuracy of Fletcher Munson curves, but not the basic idea. IMHO, the deficiencies and differences of amplifiers, if they really exist in a specific case, are not that hard to see. It may be hard to put a name on them, and you may have to choose your tests wisely to make them obvious, but they won't require SA of -120dBV distortion residuals to see. If you can think of anything I can do to a signal that's 90dB down from that signal, that's audible, including mixing it with a brass band, tell me and I'll try it. Right now I'm completely skeptical of claims that -90dB stuff of any description is audible, detectable, or affects any perception of a performance.
edit- obviously if you crank the gain up for a higher max, everything moves up, but IMO that's not a realistic quality listening situation.
Conrad Hoffman said:
From an intellectual standpoint, I can't argue with you. Everything you say makes sense. Yet I have had thousands of experiences otherwise, and so have hundreds of thousands of other music lovers.
So either there is some hidden flaw in your argument, or there are hundreds of thousands of music lovers all sharing the exact same illusion.
I'll let you argue with Occam about that one.
Re: Re: Re: Re: Re: Re: Re: "Objectifying the Golden Ear" in latest Electronics Desig
I would say that experience without understanding is not knowledge. Understanding takes much more than just "being there" for the experience. It takes discipline and effort.
Just to be clear, I am speaking of technical matters here, not spiritual/emotional/family matters. Of course, audio mixes these things in a way that makes them difficult to separate - namely the spiritual and emotional experience of listening to music vs. the technical understanding of sound reproduction.
Charles Hansen said:
I would say that experience without understanding is not knowledge. Understanding takes much more than just "being there" for the experience. It takes discipline and effort.
Just to be clear, I am speaking of technical matters here, not spiritual/emotional/family matters. Of course, audio mixes these things in a way that makes them difficult to separate - namely the spiritual and emotional experience of listening to music vs. the technical understanding of sound reproduction.
Charles Hansen said:
Well, since all of those hundreds of thousands of music lovers are human beings, and since we know that human beings are prone to subjectively perceiving differences even when there are no actual physical differences (vanity, ego and denial notwithstanding), the surprise would be if these people didn't share similar experiences.
se
I'm not sure there's a flaw. For example, I see differences in music signals between amps having a -3dB point of 3 Hz, vs 10 or 15 Hz. Significant differences. Everybody says there's nothing down there, but there obviously is. I believe you can hear a difference in cutoff points. Once we decide something matters, it has to be consistently treated, or progress will be difficult beyond that point. As mentioned, I'm very fond of differential measurements- if you see a difference using a musical signal, it's real. End of story, even if we can't put a nice neat name on it. If you don't see a difference, then claims of audible differences should be examined closely, and the burden of proof set very high.
Conrad, we are a little further along than this.
With the help of Scott Wurcer, Walt Jung and I differentially measured differences in caps to -100dB with relative ease. We still, can unfortunately hear the difference between Teflon, polystyrene, and polypropylend, even as power supply bypass caps, as well as coupling caps. Why? I can't measure any difference between these caps even at the -100dB level.
With the help of Scott Wurcer, Walt Jung and I differentially measured differences in caps to -100dB with relative ease. We still, can unfortunately hear the difference between Teflon, polystyrene, and polypropylend, even as power supply bypass caps, as well as coupling caps. Why? I can't measure any difference between these caps even at the -100dB level.
I read this thread too late but I still want to make a remark regarding burn-in of PCB.
Disclaimer first. however. I believe in science and would at least try to find a hypothesis to things that I cannot explain by "known" theory. I have, however, tried real snake oil and I know they have an effect, just like accupuncture also has an effect.
Some 30 years ago when I was a kid, I raced model cars. There was one make from England where the chassis was made from a folded piece of Lexan (polycarbonate). It soon became clear to me, as well as others, that a virgin new chassis felt very different to one used for a couple of races (each lasting 8 minutes). In that case new was better.
Even to date, I have no explanation for that. A chassis is determined by rigidity and damping, and there is no known explanation of change of elasticity or internal damping of a thermoplastic after some cyclic loading. But in this case, I know something has changed.
So I look at effects (like PCB burn-in) in Audio with an open mind, and accept that human beings are not intelligent enough as yet to explain every single pheonmena that exists in nature. And I try not to make judgement on other people's experiences without at least trying them myself.
Patrick
Disclaimer first. however. I believe in science and would at least try to find a hypothesis to things that I cannot explain by "known" theory. I have, however, tried real snake oil and I know they have an effect, just like accupuncture also has an effect.
Some 30 years ago when I was a kid, I raced model cars. There was one make from England where the chassis was made from a folded piece of Lexan (polycarbonate). It soon became clear to me, as well as others, that a virgin new chassis felt very different to one used for a couple of races (each lasting 8 minutes). In that case new was better.
Even to date, I have no explanation for that. A chassis is determined by rigidity and damping, and there is no known explanation of change of elasticity or internal damping of a thermoplastic after some cyclic loading. But in this case, I know something has changed.
So I look at effects (like PCB burn-in) in Audio with an open mind, and accept that human beings are not intelligent enough as yet to explain every single pheonmena that exists in nature. And I try not to make judgement on other people's experiences without at least trying them myself.
Patrick
Charles Hansen said:
Well, I'm hobbled with a fuse that isn’t particularly long and I'm not getting a lot of substantive answers, IMHO.
Amen - this is where I think that your argument clearly falls apart.
The fact is that you DID lower the THD in your most recent design. We can obviously agree on this.
You empathetically state that you were not trying to improve the THD performance – well fine.
I therefore presume, for the sake of the argument, that the design choices you made were based solely on their merits in terms of sonic performance.
Are we still in agreement? Good, I though so.
So what, may I ask, do you have to say about the fact that the design choices you made based solely on sonic performance also incurred a significant improvement in terms of measured performance, particularly with regards to THD?
Just a fluke? A Coincidence?
Would it not be a great deal more rational to at least conclude that there may in fact be some kind of relationship? If not a direct and obvious one, then at least a casual one, as I asked two posts ago?
Because there is a limit to which one can improve linearity without doing so.
Well, that is that you have implied about users of these things called "band-aids", which extend to things beyond global NFB, for one.
How many EC output stages have you built and listened too? Not nearly as many as Bruce Candy I bet. I'm sure that his amplifiers work quite fine with the EC loop open circuit too, and would have been run in such a state quite a few times during R&D.
john curl said:
It will have to be you and others on hearing the difference. One of the finest ever presentations of live sound I ever heard was broadcast over FM radio through who knows how many electrolytic capacitors. Except for gross cases I have never heard any of these differences.
People these days seem to go out of their way to make what they hear as counter-intuitive as possible. Like the folks who say a CD-R of a CD sounds "dramatically" different even after they are both ripped back onto a hard drive and played from the same computer. Needless to say also after a utility like Linux's dd or od is used to verify bit exact copies.
John, you guys are all vastly more experienced than I am, likely have in-depth EE educations, are probably smarter and might even smell better. I don't feel comfortable suggesting you're all kidding yourselves, but there are only a couple reasonable explanations. First, you might be kidding yourselves. It's extremely hard not to in these sorts of listening tests. But, let's say that's not it. Capacitors are particularly problematic. They cannot be compared to wire, as they are a reactance with complicated losses. I see differences between input and output with almost any two capacitors, regardless of type, because both the value and loss curves would have to be perfectly matched, and they never are. I go back to the bridge circuit presented a while back for making listening comparisons. It has the flaw of balancing the losses at only a single frequency, which isn't valid for a musical signal. In all honesty, I don't know how to make a truly legitimate comparison of capacitors, and I avoid them like the plague as part of the signal chain. OTOH, I've never been able to hear the difference between properly applied film types having low DF and low DA, much less how they're metalized, and I was looking for this 30 years ago when my hearing was much better than today.
I will say this- I've seen errors between the input and output of every amp I've looked at (admittedly not dozens, and not high end stuff). These are errors larger than -90dB, and occur with music as the source. These are errors that have a high probability of being audible. They would seem to be the first thing people should go after, rather than mystical buried-in-the-noise-floor and beyond-known-test-equipment-capabilities kind of stuff.
Now, if anybody can produce a couple of amplifiers that show a flat line (no band limiting, correct in amplitude and phase) in a input-to-output comparison at the -90dB level, and can reliably show that they sound different, than I'm full of it. But, I've yet to see an amplifier that can do that in a real world situation- i.e., hooked to speakers and using music as the source.
I will say this- I've seen errors between the input and output of every amp I've looked at (admittedly not dozens, and not high end stuff). These are errors larger than -90dB, and occur with music as the source. These are errors that have a high probability of being audible. They would seem to be the first thing people should go after, rather than mystical buried-in-the-noise-floor and beyond-known-test-equipment-capabilities kind of stuff.
Now, if anybody can produce a couple of amplifiers that show a flat line (no band limiting, correct in amplitude and phase) in a input-to-output comparison at the -90dB level, and can reliably show that they sound different, than I'm full of it. But, I've yet to see an amplifier that can do that in a real world situation- i.e., hooked to speakers and using music as the source.
As I understand it, stripped of bluster, attitude, and bad manners, G.Kleinschmidt's argument boils down to: Charles's current design sounds better. Charles's current design has lower distortion. Ergo, it is the lower distortion that is making it sound better.
This is the same poor reasoning used with the Halcro: The Halcro got good reviews. The Halcro has low distortion. Ergo, it is the lower distortion that makes it sound good.
Coming from people who claim to represent a rational, objective, scientific approach, this is particularly amusing. Never once do they examine the fundamental assumption underlying their hypothesis. Never once do they examine alternate hypotheses. They jump to conclusions contradicted by their own prior claims.
Those who espouse this point of view have lost the moral high ground they claim to inhabit by their very lack of scientific rigor. Ironic, yes?
I suggest that those who claim that low THD and IM are sufficient indicators pick a consensus distortion figure that represents the limit of human perception (substantiated by a published study, naturally), and desist in all further arguments that distortion figures below that number result in audible improvement. It is the necessary and inherent end point of that line of reasoning. No more can be added once you pass that point.
Grey
This is the same poor reasoning used with the Halcro: The Halcro got good reviews. The Halcro has low distortion. Ergo, it is the lower distortion that makes it sound good.
Coming from people who claim to represent a rational, objective, scientific approach, this is particularly amusing. Never once do they examine the fundamental assumption underlying their hypothesis. Never once do they examine alternate hypotheses. They jump to conclusions contradicted by their own prior claims.
Those who espouse this point of view have lost the moral high ground they claim to inhabit by their very lack of scientific rigor. Ironic, yes?
I suggest that those who claim that low THD and IM are sufficient indicators pick a consensus distortion figure that represents the limit of human perception (substantiated by a published study, naturally), and desist in all further arguments that distortion figures below that number result in audible improvement. It is the necessary and inherent end point of that line of reasoning. No more can be added once you pass that point.
Grey
- Status
- Not open for further replies.