Trust me..
Trust me, I know the difference between different modes of operation. Like I said earlier, I have worked in audio and RF for over 30 years. I know what biasing is (class A, AB, B, C and D).
For the unitiated, Class A is when current flows in the device for the entire waveform. Best quality, but also gives off the most heat (and stresses the circuit thermally) Class AB is when the device is biased more then 50% of the time. This way, the crossover distortion that happens when the waveform swings from positive to negative is eliminated (because both devices are on at the same time). This also allows for some class A operation (in the area where both are on) Despite John Curl's implication that class AB operation is bad , there is nothing inherently wrong with a properly designed class AB circuit. 99% of all amplifiers out there ARE class AB, including most of the Stereophile class A power amps.
Class B is when the devices are biased to an exact 50% duty cycle. mOst older op amps were designed this way for thermal reasons. Usually, there is a nasty "glitch" where the waveform crosses over from positive to negative; however negative feedback can reduce it (but not eliminate it). There is a work around for this by pulling current out of the opamp. Most times this is done be a single resistor from the output to the negative supply, though some use an actual current source. Care must be exercised when doing this however, because one can thermally overstress the IC.
Class C is when the device is cut off less than 50% of the time. It's primarily used for RF amplifers, where the flywheel effect of the output tank circuit can complete the waveform reasonably, and the output low pass filter can remove the harmonics (remember, we're only talking about operating on one frequency here).
Class D is when the device is turned on and off very rapidly (in essence, the waveform is sampled). Nyquist theory sates that if this is done at over twice the audio frequency, the audio can be reconstructed perfectly. The problem is that you need a reconstruction filter that has perfect amplitude response, phase response, impulse response and group delay right up to of the highest frequency desired. This is difficult to do with a low sampling index (below 4) The T amps actually sample in the mHz range (X20 sampling), so the reconstruction filter is a simple one with good characteristics. In many ways, a T amp more resembles a CD player then a classic audio amp.
I can not understand your refusal to embrace proven technology. Though class D amps are fairly new compared to their older class A and AB cousins, there's nothing wrong with their underlying technology. They CAN sound good! Is it that you believe that NOTHING that's not 100% analog is capable of sounding decent?
If so, then you are the one that's living in the past, not me!
Trust me, I know the difference between different modes of operation. Like I said earlier, I have worked in audio and RF for over 30 years. I know what biasing is (class A, AB, B, C and D).
For the unitiated, Class A is when current flows in the device for the entire waveform. Best quality, but also gives off the most heat (and stresses the circuit thermally) Class AB is when the device is biased more then 50% of the time. This way, the crossover distortion that happens when the waveform swings from positive to negative is eliminated (because both devices are on at the same time). This also allows for some class A operation (in the area where both are on) Despite John Curl's implication that class AB operation is bad , there is nothing inherently wrong with a properly designed class AB circuit. 99% of all amplifiers out there ARE class AB, including most of the Stereophile class A power amps.
Class B is when the devices are biased to an exact 50% duty cycle. mOst older op amps were designed this way for thermal reasons. Usually, there is a nasty "glitch" where the waveform crosses over from positive to negative; however negative feedback can reduce it (but not eliminate it). There is a work around for this by pulling current out of the opamp. Most times this is done be a single resistor from the output to the negative supply, though some use an actual current source. Care must be exercised when doing this however, because one can thermally overstress the IC.
Class C is when the device is cut off less than 50% of the time. It's primarily used for RF amplifers, where the flywheel effect of the output tank circuit can complete the waveform reasonably, and the output low pass filter can remove the harmonics (remember, we're only talking about operating on one frequency here).
Class D is when the device is turned on and off very rapidly (in essence, the waveform is sampled). Nyquist theory sates that if this is done at over twice the audio frequency, the audio can be reconstructed perfectly. The problem is that you need a reconstruction filter that has perfect amplitude response, phase response, impulse response and group delay right up to of the highest frequency desired. This is difficult to do with a low sampling index (below 4) The T amps actually sample in the mHz range (X20 sampling), so the reconstruction filter is a simple one with good characteristics. In many ways, a T amp more resembles a CD player then a classic audio amp.
I can not understand your refusal to embrace proven technology. Though class D amps are fairly new compared to their older class A and AB cousins, there's nothing wrong with their underlying technology. They CAN sound good! Is it that you believe that NOTHING that's not 100% analog is capable of sounding decent?
If so, then you are the one that's living in the past, not me!
It's not a question of "proven" technology (you're asking the wrong questions again, so naturally you're going to get the wrong answers), it's a question of sound quality. All it takes is a pair of ears.
Speaking of which...another exercise for ears, one requiring a bit more energy on the part of the listener:
--Find a cinder block wall somewhere where you can get a bit of distance from the wall. The side of a building facing a parking lot is about right. Close your eyes. Clap your hands. Your subconscious knows where the wall is; if you threw a ball, you could probably hit the wall between waist and head height fairly accurately. The problem is to educate your conscious mind. Estimate your distance from the wall in feet (or meters, if you prefer) and open your eyes to check. Close and walk closer. Repeat. Farther away. Repeat. This one takes effort. Most people assume that it will be easy--and it is, if you use your eyes. To use your ears to estimate distance is another matter entirely.
--Variation: Try a wall made out of something other than cinder blocks.
--Variation (much harder): Get in fairly close, stand diagonally to the wall, clap your hands. Listen to the reflection as it faaaaaades away from you down the length of the wall. This is going to be a very short duration event, so you're going to have to concentrate. If you get really good at it, you can estimate roughly how long the wall is, because the sound quits reflecting back to you once it runs out of wall and heads off into the trees behind the building. (For that matter, spend a few minutes clapping and listening to the sound reflected from the tree trunks.)
--Now go inside and clap your hands in a fairly reverberant room. Estimate how big it is, just from the sound. (Note that we're working our way towards concert halls...sneaky me.)
If you're not paying to attention to details such as the ones I've outlined, you're not going to hear why class D amps aren't ready for prime time. And trust me, from the comments I've seen so far, nobody's paying attention to the details. If you master some of these hints I've been giving, you won't need anyone to tell you digital sucks...you'll know why.
Grey
Speaking of which...another exercise for ears, one requiring a bit more energy on the part of the listener:
--Find a cinder block wall somewhere where you can get a bit of distance from the wall. The side of a building facing a parking lot is about right. Close your eyes. Clap your hands. Your subconscious knows where the wall is; if you threw a ball, you could probably hit the wall between waist and head height fairly accurately. The problem is to educate your conscious mind. Estimate your distance from the wall in feet (or meters, if you prefer) and open your eyes to check. Close and walk closer. Repeat. Farther away. Repeat. This one takes effort. Most people assume that it will be easy--and it is, if you use your eyes. To use your ears to estimate distance is another matter entirely.
--Variation: Try a wall made out of something other than cinder blocks.
--Variation (much harder): Get in fairly close, stand diagonally to the wall, clap your hands. Listen to the reflection as it faaaaaades away from you down the length of the wall. This is going to be a very short duration event, so you're going to have to concentrate. If you get really good at it, you can estimate roughly how long the wall is, because the sound quits reflecting back to you once it runs out of wall and heads off into the trees behind the building. (For that matter, spend a few minutes clapping and listening to the sound reflected from the tree trunks.)
--Now go inside and clap your hands in a fairly reverberant room. Estimate how big it is, just from the sound. (Note that we're working our way towards concert halls...sneaky me.)
If you're not paying to attention to details such as the ones I've outlined, you're not going to hear why class D amps aren't ready for prime time. And trust me, from the comments I've seen so far, nobody's paying attention to the details. If you master some of these hints I've been giving, you won't need anyone to tell you digital sucks...you'll know why.
Grey
GRollins said:
So for everyone other than the Special Forces of HiFi/HiEnd class D amps are fine. Perhaps you are right. 😀
The exercises you list are genuinely interesting btw (and tbh I have tried some of the things you mention about live performances), but I prefer to spend my time listening to some music instead. It's far more enjoyable and I've had enough hardcore training during my military service.
I did my military service hardcore training stationed at the Curacao naval base in the carribean, with back-up from US female tourist special forces.
Obviously, it's all about interpretation.
Obviously, it's all about interpretation.
I've been going through this thread and really see a lot of bunk about op-amps and discrete vs integrated wrt overload, noise, distortion, and the rest of it.
For starters, lets look at phono-stage bandwidth and overload. The contention from our experts is that op-amps cannot do a good job here because they can't handle the bandwidth and don't have the headroom. Id say if a phono amp has these problems its to do with the topology more than anything else. I see designs that use passive equalization after a front end amplification stage. Well, if you get a scratch or pop with this type of design you are likely to have problems. Look at the recent review of the Moon phono amp in Stereophile - tested by John Atkins - where he complains about t he 3dB (!) overload capability of their product at 20Khz. Well, it does not matter whether its an op-amp or a discrete design, this is a basic design flaw (BTW, I don't know if its discrete or IC - I jus t use it as an example).
Doug Self mentions in his book tha t he tried many times to make a lower noise gain stage than the old 5532/4 but never got it right (leaving out any discussion at this point about sound signature etc). This is a renowned designer (remember, a large part of his career was in studio equipement design - think of all those damn op-amps in some of those signal chains!)
So, I would not defend an op-amp based design for th e sake of it, but niether would I defend a discrete only position by the same token. A few people on this thread have noted that you can get device matching, thermal tracking and noise performence out of IC's that are just not possible with discretes. Go Figure.
As for the dirty sand stuff . . . . well, where do I start. (BTW, I work in the semi industry - both discretes and IC's so I am well qualified to comment on this)
For starters, lets look at phono-stage bandwidth and overload. The contention from our experts is that op-amps cannot do a good job here because they can't handle the bandwidth and don't have the headroom. Id say if a phono amp has these problems its to do with the topology more than anything else. I see designs that use passive equalization after a front end amplification stage. Well, if you get a scratch or pop with this type of design you are likely to have problems. Look at the recent review of the Moon phono amp in Stereophile - tested by John Atkins - where he complains about t he 3dB (!) overload capability of their product at 20Khz. Well, it does not matter whether its an op-amp or a discrete design, this is a basic design flaw (BTW, I don't know if its discrete or IC - I jus t use it as an example).
Doug Self mentions in his book tha t he tried many times to make a lower noise gain stage than the old 5532/4 but never got it right (leaving out any discussion at this point about sound signature etc). This is a renowned designer (remember, a large part of his career was in studio equipement design - think of all those damn op-amps in some of those signal chains!)
So, I would not defend an op-amp based design for th e sake of it, but niether would I defend a discrete only position by the same token. A few people on this thread have noted that you can get device matching, thermal tracking and noise performence out of IC's that are just not possible with discretes. Go Figure.
As for the dirty sand stuff . . . . well, where do I start. (BTW, I work in the semi industry - both discretes and IC's so I am well qualified to comment on this)
Grey is quite right in this. High quality audio is hard to get, and fairly rare in reality. Just like a 5 star restaurant or a first class bottle of wine.
If any of you think that mid fi is good enough, then go for it, but it is not what we are working at to achieve.
If any of you think that mid fi is good enough, then go for it, but it is not what we are working at to achieve.
high quality fi
yupp... even in live venues, what with sub par mixing boards, tone def sound men, lousy performers, wind noises (if outside), sirens, audience coughing, etc.
OTOH, I saw david Sanborn live at the blue note about a decade ago and was utterly floored by how real the music sounded!!!
Go figure...
here's an exercise... stand in the shower, sing your favorite song while lathering up, turn the water on as hot as you can stand it, and see if you have an epiphany....
John L.
john curl said:
yupp... even in live venues, what with sub par mixing boards, tone def sound men, lousy performers, wind noises (if outside), sirens, audience coughing, etc.
OTOH, I saw david Sanborn live at the blue note about a decade ago and was utterly floored by how real the music sounded!!!
Go figure...
here's an exercise... stand in the shower, sing your favorite song while lathering up, turn the water on as hot as you can stand it, and see if you have an epiphany....
John L.
GRollins said:
Very interesting, I have my own assortment of things I listen for and have learned to understand many things by being aware and testing reality.
The world shrinks a little, Mike.
Re: high quality fi
On a "where's it all going too" note: my wife and I went to see the Richmond Symphony a few weeks ago... To enhance the experience they mic'd it and most of what we heard came through the house sound system. Everyone enjoyed it completely and a good time was had by all...
Sadly Mike.
auplater said:
On a "where's it all going too" note: my wife and I went to see the Richmond Symphony a few weeks ago... To enhance the experience they mic'd it and most of what we heard came through the house sound system. Everyone enjoyed it completely and a good time was had by all...
Sadly Mike.
GRollins said:
+1 Here.
I've tried some of the differing Class D and T amps out there, stared closely at their methods of operation, made massive changes and modifications to fix the inherent flaws, as much as can be done when working with a finished design. I've listened closely, and then surmised that the technolgy has along way to go..and it may never get there. The only place I'll accept digital generation is in a PS rail for an amp, and even then I'm very skittish about it.
To give you and idea, even the factory floor manager who has no experience in audio, realized quite early on..that the Class T amps, falsify and accentuate high frequencies on simple singals, thus leading to a false sense of 'information and detail' but then fall completely apart on complex driven passages. With no expereince in High-end audio at all, this is the conclusion he came to, all on his own.
This (conclusion of his) was done on the 6th version of a T-amp I had under my knife, a completely modded out (Super) T-amp that is heavily modded an has some proprietary mods on it that some T-amp manufacturer's would just love to get their mitts on.
So sorry guys, no-where near a mature technology. Get thee to a drawing board. Not so bad for listening to re-runs of The Love Boat on the idiot box, which is where it comes from. It was never meant for this duty, this High End audio task, and it shows.
It's a case of the Dancing bear all over again. The Bear does not dance well, it is merely amazing that it dances at all.
As for temporal issues:
I tell people...that.. ON MY OWN DESIGNS..I can hear the difference between.. or the separate contribution of the side panels, the back wall, the driver basket, the internal fill, the differing capacitors, the cone, the box face, everything, down to the last iota. All components. All parts.
Simply because I've trained myself to do so. Via single cause analysis. For 20+ years.
Walking through places I've never been before with eyes closed is scary..but doable.
I wish class D was better. Class A (electronic, not Stereophile) and heavily biased AB are about the only way I can stand solid state over the long haul, as it's the closest thing I can get to the sound of tubes, which I have pretty much forsaken due to cost and availability problems. The problem with class A being that it's hot, heavy as hell, and expensive. None of which appeal to me. Just the sound quality. If class D were better sounding--or even as good--I'd be there in a skinny instant. Cheaper, lighter, and cooler are all to the good. There's just this thing about the sound...which is, unfortunately, everything to me.
Egad, I can't even stand to play bass through a class D amp. Went into a guitar shop. Played. Frowned. Looked at what I was playing through (you know how the store critter plugs you in and turns the amp on). Top of the line Crest. Nasty. Horrible. Intolerable. Once I'd figured out what was going on, I went on a masochistic binge and tried every class D in the place, just to see if it was all as bad as the Crest. No, it wasn't. It was worse.
And to think that people are powering PA systems with this junk, whereupon people go home and think that their stereo systems are supposed to sound that way; after all that's the way it sounded when they heard the band live. (Let's not get into the PA speakers.)
....Oh, it's just too horrible to contemplate...
No wonder people accept ipods.
And people make yak-yak about how high end companies are swaying people with advertising? Say what? Look at the ads blaring "Digital Quality Sound" and say that with a straight face. Jeez, they even swoop the word DIGITAL at you on movie screens as though it's something to brag about.
And people chastise Charles? Call a letter to the editor advertizing?
Compare the amount spent on digital advertising to Charles's entire operating budget and tell me who's being influenced by advertising.
Well, a quick run through this thread would certainly yield a dozen names. But, oh my, you can't pull the wool over their eyes...they saw right through Charles's blatant attempt to get free advertising.
Yep...them's some smart customers, ain't they?
Grey
Egad, I can't even stand to play bass through a class D amp. Went into a guitar shop. Played. Frowned. Looked at what I was playing through (you know how the store critter plugs you in and turns the amp on). Top of the line Crest. Nasty. Horrible. Intolerable. Once I'd figured out what was going on, I went on a masochistic binge and tried every class D in the place, just to see if it was all as bad as the Crest. No, it wasn't. It was worse.
And to think that people are powering PA systems with this junk, whereupon people go home and think that their stereo systems are supposed to sound that way; after all that's the way it sounded when they heard the band live. (Let's not get into the PA speakers.)
....Oh, it's just too horrible to contemplate...
No wonder people accept ipods.
And people make yak-yak about how high end companies are swaying people with advertising? Say what? Look at the ads blaring "Digital Quality Sound" and say that with a straight face. Jeez, they even swoop the word DIGITAL at you on movie screens as though it's something to brag about.
And people chastise Charles? Call a letter to the editor advertizing?
Compare the amount spent on digital advertising to Charles's entire operating budget and tell me who's being influenced by advertising.
Well, a quick run through this thread would certainly yield a dozen names. But, oh my, you can't pull the wool over their eyes...they saw right through Charles's blatant attempt to get free advertising.
Yep...them's some smart customers, ain't they?
Grey
I'm surprised no one mentioned the obvious - unprocessed voice. It's incredibly hard to get close and impossible to make deceivingly real. Yet it's something everyone hears live daily. Part of the problem is we're so wired to focus on meaning, and so accustomed literally from birth to hearing it reproduced badly, that we're unconsciously willing to accept a wide range of error.
Try decoupling yourself from meaning when listening to a conversation and focus on the nature of sibilance, its variance by gender and age, the power of plosives and the subtle burr of fricatives. Now try your audio system. For me it was very difficult to do. Fortunately as I've mentioned elsewhere I have the perfect regular opportunity in the form of management meetings and sales reports.
Try decoupling yourself from meaning when listening to a conversation and focus on the nature of sibilance, its variance by gender and age, the power of plosives and the subtle burr of fricatives. Now try your audio system. For me it was very difficult to do. Fortunately as I've mentioned elsewhere I have the perfect regular opportunity in the form of management meetings and sales reports.
The dumbing down of ears...
Well, there is a hypothesis that our ears are getting 'dumber', thanks to all the perceptual coding that we hear every day. Virtually every TV and radio transmission uses it, as do iPods, MP3s, etc. Even phone calls today are bit rate reduced. The quesion is: does all this coding make our ears "lazier"? Does having less information make our brain lazy? Interesting thoughts, eh?
rdf said:
Well, there is a hypothesis that our ears are getting 'dumber', thanks to all the perceptual coding that we hear every day. Virtually every TV and radio transmission uses it, as do iPods, MP3s, etc. Even phone calls today are bit rate reduced. The quesion is: does all this coding make our ears "lazier"? Does having less information make our brain lazy? Interesting thoughts, eh?
john curl said:
So, John, any amp that is not on par with the Stereophile class A list (or whatever your benchmark is) is considered mid-fi?
I have not heard a Class D amp sound as good as a Class A (electronic that is) amp, and Class A is still my favorite (with all its flaws that have been mentioned already), but, in life, its not just about the extremes. It's not just "excellent" and "rubbish"... There's a lot going on in between.
Not to mention certain cases where people need to drive hard loads... I know people who have always dreamed of getting Magnepans or other speakers that need serious power to drive them... Good luck finding an amp working in class A to drive those without having to sell you house... Then you could get a pair of ICEpower 1000 amps for $1200 and really give them all the power they need. If you are not into DIY it'll be hard to find another amp within that budget to perform as good when driving power-hungry speakers.
Each case is unique and generalizations do not help.
About the "fake high freq's" of Tripath amps, I can only say we have compared a Trends 10.1 with an Electrocompaniet AW250 and 3 out of 3 people found the highs on the Tripath to be better and more natural. If they can fake highs that good, well, way to go Dr. Tripathi...
Moreover, it was almost as good as Audio Note 300B amps driving Avantgarde Duo speakers (though horns and stuff are far from being my favourites...).
Then, no, it wasn't nearly as good as an all Krell Evolution setup... But we are talking about a $100 amp here...
And that's before even going to the discussion of "how do you know what the highs recorded on this particular disc sound like?"... Because, as you probably know, the term Fidelity on Hi-Fi refers to the reproduction of the information recorded on the media and not to reproduction of the live band we heard last night at the Pub....
There's a curious logic at work here. I've seen it often enough that I think it's simply a characteristic of human nature. It goes something like this:
--Consumer decides to buy something
--Consumer sees that item A is best
--Consumer can't afford A
--Consumer looks further, finds item B, which he can afford
--Consumer then decides that B is "as good as" or "nearly as good as" A, no matter how wide the gulf between their performance
It's not just in audio. I've seen the same thing in cookware, cars, cameras, and guitars.
This is another pretty clear case of the consumer hearing what he wants to hear, in this case influenced by his wallet (albeit perhaps with a little help from advertising).
If you give the man on the street a graphic equalizer, he will almost inevitably boost the lows and highs and tell you that it sounds better. The resulting shape of the sliders on the graphic equalizer gave rise to the name for this: The Smile...low in the middle, raised at both ends.
If someone hasn't educated their ears, anything that emphasizes either (or both) frequency extremes will be perceived as "better" whether it actually is or not. Thus you have people listening to high feedback amp circuits (which tend to have distortion products piling up in the upper midrange and treble) and calling the resulting sound "detail." It isn't. A little time listening to real music and employing one of the lessons I suggested above (the one where you imagine real music to be a stereo and ask yourself whether you're wanting to turn up the treble) will show you that the circuit isn't really more detailed.
The same holds true for the unnatural highs in digital amps. If your ears aren't tuned up, you're likely to think that the exaggerated highs are "better" than another amp's highs. Ain't so.
Get thee to a concert hall, lad.
Grey
--Consumer decides to buy something
--Consumer sees that item A is best
--Consumer can't afford A
--Consumer looks further, finds item B, which he can afford
--Consumer then decides that B is "as good as" or "nearly as good as" A, no matter how wide the gulf between their performance
It's not just in audio. I've seen the same thing in cookware, cars, cameras, and guitars.
This is another pretty clear case of the consumer hearing what he wants to hear, in this case influenced by his wallet (albeit perhaps with a little help from advertising).
If you give the man on the street a graphic equalizer, he will almost inevitably boost the lows and highs and tell you that it sounds better. The resulting shape of the sliders on the graphic equalizer gave rise to the name for this: The Smile...low in the middle, raised at both ends.
If someone hasn't educated their ears, anything that emphasizes either (or both) frequency extremes will be perceived as "better" whether it actually is or not. Thus you have people listening to high feedback amp circuits (which tend to have distortion products piling up in the upper midrange and treble) and calling the resulting sound "detail." It isn't. A little time listening to real music and employing one of the lessons I suggested above (the one where you imagine real music to be a stereo and ask yourself whether you're wanting to turn up the treble) will show you that the circuit isn't really more detailed.
The same holds true for the unnatural highs in digital amps. If your ears aren't tuned up, you're likely to think that the exaggerated highs are "better" than another amp's highs. Ain't so.
Get thee to a concert hall, lad.
Grey
Well said, once again, Grey.
For the record, I design hi end and high quality mid fi components. I know, and hear the difference between amps and preamps. Most of us can, IF we are exposed to good sounding stuff and can use that as a reference.
When it comes to class D, I have consulted with Tripath, on two occasions, for Parasound and technically evaluated their designs. I even helped them with an important problem in one case. They are interesting designs, but they are not A quality hi fi, in my opinion.
For the record, I design hi end and high quality mid fi components. I know, and hear the difference between amps and preamps. Most of us can, IF we are exposed to good sounding stuff and can use that as a reference.
When it comes to class D, I have consulted with Tripath, on two occasions, for Parasound and technically evaluated their designs. I even helped them with an important problem in one case. They are interesting designs, but they are not A quality hi fi, in my opinion.
- Status
- Not open for further replies.