My friend Tim has intensely studied many transistor and tube characteristic curves for decades now. He's a very technically and mathematically astute. He's also done some test building over the years, but mostly his beliefs come from studying the curves and the mathematics.
He says now that push pull BJT* amplifiers work best in Class AB, not Class A.
Of course this flies in the face of everything I've ever read, heard, or thought about, and my most prized audio machine is a Class A amplifier, because I've long been a fan of Class A amplifiers and my belief that they are best for electrostatic speakers.
He agrees it flies in the face of standard and conventional thinking, but insists that nevertheless it's true. His argument is that the "cutoff" distortion on one side precisely works with the changes on the other side to produce the lowest distortion at some particular AB bias. Class A doesn't get this benefit, and produces higher distortion.
He's been wrong before, so he could easily be wrong this time. My suspicion is that it either misses other factors, and/or
But, if he's right, and in many cases even if he's wrong, it's quite possible that Nelson Pass's Threshold amplifiers (which were BJT, he switched permanently to MOSFET after leaving Threshold) were sub-optimal. The attempt to eliminate cutoff, Tim affirms, would produce higher distortion than the correct Class AB, and would therefore produce higher distortion than necessary, and at higher operational costs.
That's certainly true of some other of Nelson's designs, such as many devices in the single-ended Aleph era, and many of the First Watt designs.
Is this a big deal? That through charisma or whatever Nelson sold higher distortion amplifiers than necessary?
Firstly, I suspect all the Threshold amplifiers all had distortion sufficiently low, and with sufficient reliability, and relative high power and current into most loads, to be considered fully High Fidelity Amplifiers, such as there is no respectable evidence that people can hear differences between. So the "loss" to the users of Threshold amplifiers is nil, by the audio objectivists very own standards.
They've been loved, held their value, etc, etc.
So, that's fun, entertainment, and playing music.
Secondarily, when the actuality doesn't really matter, the only thing that does matter is the story. The story says lower distortion, greater PRAT, or whatever. The story makes you feel good. The increased distortion is irrelevant, in most cases, even in many extreme cases where people just seem to groove on the distortion. Minimum distortion I would generally look for, myself, as much as I can afford. But if other people like other stories, other presentations ever, that's their thing, at least while it lasts. I think, in the end, as little as possible distortion, combined with the greatest flexibility (the equipment is supposed to let us Play music, not have Music imposed on us)
I believe many of the beliefs that modern distortion reducing methods create bad sound are myths, having perhaps vanishingly small examples, and none where proper engineering has been done.
But there are actually many different circuits, it's hard to say we have an "optimal" circuit as such, and if we don't have an optimal circuit, then we don't have an optimal class, and so on. Furthermore some things may create more or less distortion than other things under some circumstances, and have the reverse be true under others.
In fact, as is well known, people love the higher distortion circuits like SET's and transformers for various reasons. And since this is about fun, not Accurate Reproduction, that's fine, for them, as long as they continue to do so. And much as I might think it would happen, many people are sticking with one of many higher distortion than necessary paths.
We don't hear distortion very well anyway. It's the frequency response under load we hear mostly. People are finding or not the EQ they want. Some people fear EQ.
Many like me fear this is an expensive, and even more than that, difficult and ineffective way to get very much EQ. But if someone thinks it works for them, it works for them, fine.
I have actually experienced a "solid state" "dryness" (in a malfunctioning Citation 11) which couldn't be eliminated with (it's very own) EQ. That happens when you have higher noise and IM than necessary. You can't EQ that away. But EQ works quite well for most other things. When we have digital sources, we can have distortion and noise free EQ, which provides exactly the alteration needed or desired.
EQ lets me easily and precisely play with how things are played, and that's become very important to me.
(Tim variously prefers and not prefers MOSFET amplifiers in Class A. He has most often said MOSFET amplifiers require very high bias to be in their most linear region, and for practical reasons (would it require liquid nitrogen cooling?), nobody ever does that. But he may be softening on the bias levels needed to make MOSFET preferable to properly biased Class AB+ BJT's. I think there may be other factors and ways of looking at the situation.)
He says now that push pull BJT* amplifiers work best in Class AB, not Class A.
Of course this flies in the face of everything I've ever read, heard, or thought about, and my most prized audio machine is a Class A amplifier, because I've long been a fan of Class A amplifiers and my belief that they are best for electrostatic speakers.
He agrees it flies in the face of standard and conventional thinking, but insists that nevertheless it's true. His argument is that the "cutoff" distortion on one side precisely works with the changes on the other side to produce the lowest distortion at some particular AB bias. Class A doesn't get this benefit, and produces higher distortion.
He's been wrong before, so he could easily be wrong this time. My suspicion is that it either misses other factors, and/or
But, if he's right, and in many cases even if he's wrong, it's quite possible that Nelson Pass's Threshold amplifiers (which were BJT, he switched permanently to MOSFET after leaving Threshold) were sub-optimal. The attempt to eliminate cutoff, Tim affirms, would produce higher distortion than the correct Class AB, and would therefore produce higher distortion than necessary, and at higher operational costs.
That's certainly true of some other of Nelson's designs, such as many devices in the single-ended Aleph era, and many of the First Watt designs.
Is this a big deal? That through charisma or whatever Nelson sold higher distortion amplifiers than necessary?
Firstly, I suspect all the Threshold amplifiers all had distortion sufficiently low, and with sufficient reliability, and relative high power and current into most loads, to be considered fully High Fidelity Amplifiers, such as there is no respectable evidence that people can hear differences between. So the "loss" to the users of Threshold amplifiers is nil, by the audio objectivists very own standards.
They've been loved, held their value, etc, etc.
So, that's fun, entertainment, and playing music.
Secondarily, when the actuality doesn't really matter, the only thing that does matter is the story. The story says lower distortion, greater PRAT, or whatever. The story makes you feel good. The increased distortion is irrelevant, in most cases, even in many extreme cases where people just seem to groove on the distortion. Minimum distortion I would generally look for, myself, as much as I can afford. But if other people like other stories, other presentations ever, that's their thing, at least while it lasts. I think, in the end, as little as possible distortion, combined with the greatest flexibility (the equipment is supposed to let us Play music, not have Music imposed on us)
I believe many of the beliefs that modern distortion reducing methods create bad sound are myths, having perhaps vanishingly small examples, and none where proper engineering has been done.
But there are actually many different circuits, it's hard to say we have an "optimal" circuit as such, and if we don't have an optimal circuit, then we don't have an optimal class, and so on. Furthermore some things may create more or less distortion than other things under some circumstances, and have the reverse be true under others.
In fact, as is well known, people love the higher distortion circuits like SET's and transformers for various reasons. And since this is about fun, not Accurate Reproduction, that's fine, for them, as long as they continue to do so. And much as I might think it would happen, many people are sticking with one of many higher distortion than necessary paths.
We don't hear distortion very well anyway. It's the frequency response under load we hear mostly. People are finding or not the EQ they want. Some people fear EQ.
Many like me fear this is an expensive, and even more than that, difficult and ineffective way to get very much EQ. But if someone thinks it works for them, it works for them, fine.
I have actually experienced a "solid state" "dryness" (in a malfunctioning Citation 11) which couldn't be eliminated with (it's very own) EQ. That happens when you have higher noise and IM than necessary. You can't EQ that away. But EQ works quite well for most other things. When we have digital sources, we can have distortion and noise free EQ, which provides exactly the alteration needed or desired.
EQ lets me easily and precisely play with how things are played, and that's become very important to me.
(Tim variously prefers and not prefers MOSFET amplifiers in Class A. He has most often said MOSFET amplifiers require very high bias to be in their most linear region, and for practical reasons (would it require liquid nitrogen cooling?), nobody ever does that. But he may be softening on the bias levels needed to make MOSFET preferable to properly biased Class AB+ BJT's. I think there may be other factors and ways of looking at the situation.)
No comments:
Post a Comment