 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
76.90.100.230
In Reply to: RE: And that's why they call it AB but it is not truly A posted by AbeCollins on February 08, 2025 at 10:35:38
Yes, I think it is just easier to call the power an AB amplifier makes, while not reaching cutoff, Class A vs. trying to explain that it only acts like Class A is one narrow way but it's not real Class A. I believe that is the most likely reason designers (who for sure know better) say what they say.
I wish the wouldn't do that. They are confusing people. :-(Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 02/08/25Follow Ups:
One thing is for sure, the distortion behavior is not the same as a Class A amp. Also, if you look at measurements, you can see that not all Class AB amps are sufficiently high bias to remove the zero crossing cutoff distortion.
As an aside, newish brand Infigo claim to operate always in class A using a novel sliding bias scheme but JA's measurements showed it rife with cross-over distortion though THD was still low and the reviewer loved the sound. I've no idea how much bias is required for a 'smooth' handover between push & pull and vice versa but many years ago Douglas Self wrote a series of articles about audio power amp design and claimed, IIRC, that optimally biased class B actually produced the lowest distortion. I didn't pay much attention at the time so don't know his reasoning but think it was that the variation of composite transconductance of the push-pull pair could be made much less through the cross-over region than for an AB pair. This was set in the context of applying as much loop gain as possible so, perhaps, with the right class B bias and a lot of feedback Topping has gone in this direction and try maximize efficiency to go against class D designs.
I found that amp review on Stereophile (see above). I don't take too much stock in what the reviewer liked and didn't like. You can always find someone to like your product. Having had amps early in my audiophile career that were both Class A SS and what I would call close to Class B (meaning later measurements revealed zero crossing distortion), I can tell you the one with zero crossing distortion sounded nasty and congested and was not very long in my system.
The Naim measurement I found also has zero crossing distortion and I don't know about you, but I find all Naim amps to sound dry (as in no harmonics and poor low level resolution needed for ambience retrieval) and flat. Horrible amps that traded on the selling point of having PRAT...because they had nothing else that sounded like music.
To the best of my understanding, you cannot remove distortion caused by zero crossing with negative feedback. I would need to read the reference I learned that from again to know the exact reasons why. Might have something to do with the fact that it is not an inherent transistor property but related to the stitching together two transistors to make a whole waveform and the glitch created by the imperfect handoff.
Distortion is created if the transfer function is anything other than a straight line (with slope = gain). For a push-pull output stage, if there is a little wiggle in the line around the zero crossing that creates cross-over distortion. The larger the deviation, the higher the distortion and sharper wiggles create higher frequency harmonics. Feedback can only suppress any open loop distortion by the amount of loop gain at the frequency of interest. So, higher frequency harmonics are suppressed less because amplifier loop gain is always rolling off as frequency increases, and this applies to all distortions not just cross-over.
One danger, I think, is that biasing too hard into class B could create a 'dead-zone' at the cross over point where there is a small region with no transconductance and that will give low, or no, loop gain in that region and, hence, no suppression. It would be interesting to study but I don't get much spare time and I'd rather listen!
Conversely, class AB output stage should have twice the slope in the region where both devices are conducting (class A region) and a signal transitioning between A & B regions will be distorted by that slope change and there will also be a halving of loop gain in the class B region. So open loop distortion is created by that transition and the amount of suppression is modated. This is the so-called 'gm doubling' effect that is in the literature. Most recently it was tackled by Jason Stoddard in his blog, he came up with something he calls a 'continuity bias'to get around that problem. But, IIRC, that biasing burns significant power that is still lower than staying class A all the time but much more than a traditional AB.
Where both finals are in cutoff at idl, it's class "C" not "B" . . . Not good for audio. Almost all audio amps claiming "B" operation are really "AB" with idle bias current set low but high enough to prevent both finals from being cutoff simultaneously under real-world conditions of manufacturing and operation. It could be argued that true class B exists only at a single operating point where "AB" and "C" have an infinite number of points.
:)
Interesting how this amplifier class discussion arises every few years and involves some of the same persons each time. The basic amp classes were defined nearly a century ago . . .
However, that means the performance around zero crossing is very non-linear and will create a lot of distortion that is of a very obnoxious nature.
Look at the stereophile measurements I shared above. You can see the sharp discontinuity they have around the zero crossing. As one can imagine, this is essentially only high order dissonant distortion components. Even very low levels of these harmonics will audibly damage the sound.
As Tre was pointing out, even AB is not putting the bias in a very linear part of the transistors transfer function, which will also result in increased higher order harmonics.
Yes, sliding bias is a whole different thing. I mentioned that early on in this thread.
"I think there are some designs that operate in both Class A and (when needed) Class AB but that requires "sliding bias"." See link.
"with the right class B bias and a lot of feedback"
I wasn't aware that there was different types of class B bias. I thought class B was biased to where the devices are just off and ready to come on at the first sign of a positive signal voltage? I thought all Class B amps were biased at the same spot?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
I think bias is a spectrum so you could under bias heavily into class B such that both devices are off for a region and make a dead zone around the zero crossing (which would be bad) or over bias until you end up in class A. I suppose an optimum class B is something that gives the best trade off between distortion and power dissipation. I don't have direct experience of this but, I think, Self delves into it and his work is online and he justifies his choice of output stage biasing.
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]