Thanks Bigun,
Its a very interesting design indeed. I've had another look and to me its a non switching design as the opposite device is held "on" @ 70ma as opposed to conducting the signal. Not taking away from the design at all...just an observation. 🙂
Still not having read Douglas Self's article and with deep respect for the man, we will forever argue about a non-argument if one deviates from the classic definition. In that an absolute class-B is theoretical as the transition need be exactly where one device takes over from the other - otherwise it becomes class-AB in one direction or somewhat class-C in the other.
Those definitions were formed in the days of tubes, when sharp cut-offs were not nearly such a reality as with BJTs. What are we supposed to call a say 80W power stage delivering a peak of 4,47A into 8 ohm (resistive - one thing at a time, please!)? In my design case the optimum bias current was about 80mA.
That gives 1,8% of the peak current - surely low enough to be called class-B! But decrease that stand-by to below 50mA and harmonics start rearing their ugly peaks (in my case). At some 20mA the high-order distortion components are severe enough to cause definite listener fatigue at low listening level. Also distortion as a percentage of the maximum signal level can be misleading. If the absolute levels are comfortably below the audible hearing threshold they do not matter.
So the matter cannot be accurately expressed in simple terms. One needs to also include spectral analysis and absolute signal level. Class distinction then becomes relative to one's definition.
Those definitions were formed in the days of tubes, when sharp cut-offs were not nearly such a reality as with BJTs. What are we supposed to call a say 80W power stage delivering a peak of 4,47A into 8 ohm (resistive - one thing at a time, please!)? In my design case the optimum bias current was about 80mA.
That gives 1,8% of the peak current - surely low enough to be called class-B! But decrease that stand-by to below 50mA and harmonics start rearing their ugly peaks (in my case). At some 20mA the high-order distortion components are severe enough to cause definite listener fatigue at low listening level. Also distortion as a percentage of the maximum signal level can be misleading. If the absolute levels are comfortably below the audible hearing threshold they do not matter.
So the matter cannot be accurately expressed in simple terms. One needs to also include spectral analysis and absolute signal level. Class distinction then becomes relative to one's definition.
Have you seen this patent? Link: Active bias circuit for operating push-pull amplifiers in class A mode - Threshold Corporation
Thanks.
I was aware that NP had an active bias patent but never really worked out if it was a non switch etc..I also believe it was dropped due to thermal issues or possibly misbehaviour into reactive loads.
Hi Guys
Johan, if you recall from the RCA "Class-B Design" routine, the last step is to set the idle current for one-third of the tube's rated plate dissipation. Not exactly class-B as anyone regards it to be!
The only true class-B audio amp that does not sound bad is the screen-drive type for tetrodes/pentodes. The tubes idle at zero current and only turn on when signal is present. This requires drive from a small power amp and with common power tubes provides a power boost of 6x. I show practical circuits in TUT5.
Personally I would say that Nelson's earlier work represents much "straighter" engineering. The later stuff is an interesting exercise into simplicity at the expense of best performance.
Have fun
Kevin O'Connor
Johan, if you recall from the RCA "Class-B Design" routine, the last step is to set the idle current for one-third of the tube's rated plate dissipation. Not exactly class-B as anyone regards it to be!
The only true class-B audio amp that does not sound bad is the screen-drive type for tetrodes/pentodes. The tubes idle at zero current and only turn on when signal is present. This requires drive from a small power amp and with common power tubes provides a power boost of 6x. I show practical circuits in TUT5.
Personally I would say that Nelson's earlier work represents much "straighter" engineering. The later stuff is an interesting exercise into simplicity at the expense of best performance.
Have fun
Kevin O'Connor
That low scale high-order harmonics creating bad sound and fatiguing is not true as directly, but it indeed has corellation of cause. Open loop classB doesn't have much fatiguing and many classA with still with fatiguing sound.
Do not accept that <0.01%, <3dB NFB is the same, it will be most different between bjts and mosfets. ClassA amps usually using low or very low negative feedback or even open loop and it has no problem with distorted sound, but class B need feedback. Using classB will be more difficult getting the best sounding one because the things is happened and they don't really know.
For example, there will be impossible for hexfet output that have no fatiguing sound with only common used classAB topology no matter if it tweaked for 30years, but it could be better than bjt in SE classA with only simple gate resistor tweak.
Yamaha B4 has class operating switch from classAB to classA but this switch is really do unimportant job.
Meanwhile on the jlh-10-watt-class-amplifier thread there are hundreds of fans of this design who apparently appreciate it more than you do.
This is one of the few power amplifiers around that does not require stabilisation capacitors.
Michael J
Hi Guys
Sloth is inherently stable.
I thought the JLH was fine when I built it. It worked okay. I built a lot of different power versions of it and actually the higher power ones sounded better at normal listening levels than the original. My previous amp was an even worse SE tube thingy, so for a while the JLH-10 was an improvement. Then I built a far more linear amp (that also did not require comp caps - who cares? ) and parked the '10.
JLH is fine if you want a solid-state amp that sounds sort of tubey - bad tubey. It's okay as a guitar amp, too. If you read the original articles carefully, you see quite clearly that economy was a large factor of the design, to the detriment of performance as far as many have seen and measured.
It's fine if you or others like it. I like other things. Want some boards?
Have fun
Kevin O'Connor
Sloth is inherently stable.
I thought the JLH was fine when I built it. It worked okay. I built a lot of different power versions of it and actually the higher power ones sounded better at normal listening levels than the original. My previous amp was an even worse SE tube thingy, so for a while the JLH-10 was an improvement. Then I built a far more linear amp (that also did not require comp caps - who cares? ) and parked the '10.
JLH is fine if you want a solid-state amp that sounds sort of tubey - bad tubey. It's okay as a guitar amp, too. If you read the original articles carefully, you see quite clearly that economy was a large factor of the design, to the detriment of performance as far as many have seen and measured.
It's fine if you or others like it. I like other things. Want some boards?
Have fun
Kevin O'Connor
There are always group of fans there.
-Voltage driving amp vs current driving amp
voltage driving has more fellow
-ClassA vs ClassAB
classA with higher fans
-Distorted vs Undistorted
-VFB vs CFB
-Listening vs Measuring
being at the middle may got the best amp.
-driving loudspeaker with good impedance
-ClassB with better than classA sound
-Accept some distortions because it is the nature.
-Perfect feedback (canonic)
-Do listening, measuring, and calculating/predicting with help of simulation.
-Voltage driving amp vs current driving amp
voltage driving has more fellow
-ClassA vs ClassAB
classA with higher fans
-Distorted vs Undistorted
-VFB vs CFB
-Listening vs Measuring
being at the middle may got the best amp.
-driving loudspeaker with good impedance
-ClassB with better than classA sound
-Accept some distortions because it is the nature.
-Perfect feedback (canonic)
-Do listening, measuring, and calculating/predicting with help of simulation.
Kevin,
I was not aware of that stipulation - where do I find that? I was referring to basic class-B as generally taught and also stated in the RCA Tube Manual. Sure, what you say is true; not exactly class-B. That is what I would call class-AB. (It seems that there are different definitions of class-B. I am confused as to why that should be. Once we use the terms class-A, -AB and -B, unless we use classic definitions as a starting point there will be confusion. We can go forward from there with any variations for whatever reason.)
Ontoaba,
The influence of high-order harmonic artifacts has first been shown by Peter Walker if I recall, mainly from an initial difference in hearing experiences when first 'discovered' with a comparison of his QUAD II and first transistor amplifier, followed by thorough analyses. Thereafter it was supported by several test results from other researchers. In fact Norman Crowhurst analysed and predicted this increasing influence the higher the order of harmonics back in 1957 (not sure of the year), long before it was experienced in practice with the advent of semiconductors.
I can only go by the tests done by others; such tests are time-consuming and need a number of experienced listeners apart from spectrum analyses. I do not have access to these and can only quote said results, but it sounds logical to me. One only needs to estimate the dozens/hundreds of generated harmonics possible in an orchestral performance and then consider the possibilities of hundreds more - intermodulation products of those that exist also above the audible band, where many amplifiers are no longer linear and have loop phase shift. I once read such an exposé giving startling results. I can unfortunately not recall which; it was a few decades ago.
Thus respectfully, I have difficulty to agree with you about the effect of higher order harmonics. It has been shown that some combinations of tones apparently cause such stridency, even if almost below the limit of human audibility (and depending on which frequencies) that the listener experienced listener fatigue when they were present (again I apologise for not giving a reference because of the time elapsed). Such an effect is not really measurable because of the constant change of signal components while the music is proceeding. I have not Googled at this time; I am sure you will find references to this effect there.
It has been shown generally what distortion is audible, irrespective of what kind of amplifier generates it. Any generated harmonic products above some 0,7% will be audible from whatever class of operation. The 'unpleasantness factor' will depend on the nature of the music at any instant.
This is a whole subject on its own which I will not raise now - we will soon go quite off-topic
PS: I have not nearly read all about this. If there are any tests refuring the above I would like to read about them. (Such must naturally be on the same level as the tests supporting the fatigue-confirming tests. Observations by a few individuals under different circumstances will not cut it.)
.
Last edited:
How about a little context around the "SE tube thingy" that was so bad and the "far more linear amp (that also did not require comp caps - who cares" which is so good.
Michael J
crossover distortion below 0.7% is usually very audible.
Crossover distortion is harmonic in that it repeats exactly in time with the main AC signal and it contains harmonics at very high multiples.
The big difference is that crossover distortion is only audible during the very short "period" around the crossover region. It is the very short "period" that arithmetically makes crossover distortion appear to be very low, whereas in level it is actually very high during the period that it is evident.
I would guess that crossover would need to be at least 10 times lower in level to be inaudible. Would that put the target @ < 0.07%?
Last edited:
Oh entirely so, Andrew.
I was making a general basic statement in reply to the notion that amplifiers operating in certain classes can even be hi-fi without NFB. Meaning, any topology I know of will have audible distortion if operated without NFB. (But as someone said before, again this must be qualified. If no NFB means 'no global NFB' but there is ample local/nested NFB then of course distortion may be inaudible, etc. Perhaps one should not make general/unqualified statements on this subject!)
Also there is the ambiguity of stating distortion as a percentage. Percentage of what? Cross-over distortion as percentage of maximum output certainly has little relevance to its 'stridence factor'. The absolute level compared to threshold-of-hearing (t.o.h) is then more relevant - with an ample safety factor, because as said earlier it was shown that certain combinations of harmonic frequencies can be so disturbing to hearing (meaning actually the brain) that they have an effect (meaning listener fatigue) even when on or slightly below t.o.h.
Apology repetition; just to make my stance clear. (I must try get the above mentioned tests back; need to try Google - they were published in the 70/80s.)
I was making a general basic statement in reply to the notion that amplifiers operating in certain classes can even be hi-fi without NFB. Meaning, any topology I know of will have audible distortion if operated without NFB. (But as someone said before, again this must be qualified. If no NFB means 'no global NFB' but there is ample local/nested NFB then of course distortion may be inaudible, etc. Perhaps one should not make general/unqualified statements on this subject!)
Also there is the ambiguity of stating distortion as a percentage. Percentage of what? Cross-over distortion as percentage of maximum output certainly has little relevance to its 'stridence factor'. The absolute level compared to threshold-of-hearing (t.o.h) is then more relevant - with an ample safety factor, because as said earlier it was shown that certain combinations of harmonic frequencies can be so disturbing to hearing (meaning actually the brain) that they have an effect (meaning listener fatigue) even when on or slightly below t.o.h.
Apology repetition; just to make my stance clear. (I must try get the above mentioned tests back; need to try Google - they were published in the 70/80s.)
Hi, that is fine. Many people in many decades has disagreement with high order harmonics(1st Peter Walker) and phase shift(Matti otala).
It is my little experiece while built my projects.
The listening fatigue effect is indeed present, most in mosfet and tubes with feedback and less in BJT.
I ever doing harmonic combination test with long and carefull listen, and it has no corellation with fatiguing experience, end up with 0.01% listenable differences at high order plus multi order, and 0.02% at H2.
I am not able to completely explain the listening fatigue (and sound stage imaging effect), but most corelated with this is damping characteristics. To reveal it, may it need some knowledge about hearing mechanism and its brain based autolevel.
Hi Guys
Johan, that RCA ref is the Transmitting Tube manual (TT-5) for "Class-B Triodes". The design for tetrodes and pentodes is identical except for adding the screen supply and figuring out power for that.
Michael, the amp that was better than the JLH was actually a true Lin design - an app note from TI pushing their complimentary Darlington packages. The Lin circuit has just a VAS and output buffer and is inverting overall. Because of its simplicity, there was no comp cap. Someone previously in this thread had referred to an amp that did not need compensation as something to strive for, which in reality it is not. That's why I said "who cares?".
Overall, there are good and bad class-A and class-AB amps. Self shows the modified Lin and how to refine it in the traditional way, assuming the "rules of thumb" established by economics rather than real electronic limitations to be true. To me, Cordell hit his high point design-wise with his reference mosfet amp. Other designers have their version of "what works" and some is verified with measurement and listening while the rest can only be subjectively appreciated.
Most DIYers lack knowledge and fancy test equipment. Simulators are only so good. You ultimately have to build a real circuit with real parts and test it for basic functionality, then listen to it. Apart from the novelty of getting something to work, your ears will eventually tell you if the circuit is flawed compared to what you like, which is not the same as telling you that the performance is accurate or not.
The JLH-10 is a simple circuit anyone can build and get working - although the others in the high-school class we built them in took longer than I would ever have thought.... For everyone else, it should be a confidence building experience to move on to better performing circuits.
Have fun
Kevin O'Connor
Johan, that RCA ref is the Transmitting Tube manual (TT-5) for "Class-B Triodes". The design for tetrodes and pentodes is identical except for adding the screen supply and figuring out power for that.
Michael, the amp that was better than the JLH was actually a true Lin design - an app note from TI pushing their complimentary Darlington packages. The Lin circuit has just a VAS and output buffer and is inverting overall. Because of its simplicity, there was no comp cap. Someone previously in this thread had referred to an amp that did not need compensation as something to strive for, which in reality it is not. That's why I said "who cares?".
Overall, there are good and bad class-A and class-AB amps. Self shows the modified Lin and how to refine it in the traditional way, assuming the "rules of thumb" established by economics rather than real electronic limitations to be true. To me, Cordell hit his high point design-wise with his reference mosfet amp. Other designers have their version of "what works" and some is verified with measurement and listening while the rest can only be subjectively appreciated.
Most DIYers lack knowledge and fancy test equipment. Simulators are only so good. You ultimately have to build a real circuit with real parts and test it for basic functionality, then listen to it. Apart from the novelty of getting something to work, your ears will eventually tell you if the circuit is flawed compared to what you like, which is not the same as telling you that the performance is accurate or not.
The JLH-10 is a simple circuit anyone can build and get working - although the others in the high-school class we built them in took longer than I would ever have thought.... For everyone else, it should be a confidence building experience to move on to better performing circuits.
Have fun
Kevin O'Connor
OK Kevin,
Sensing that we might reach a "boredom point" here, I was merely indicating that unless one starts at and agrees on some basic definitions, confusion will set in. One cannot throw about terms like class-B and class-AB unless there is at least agreement on what each basically means, whether we use them like that or not. They are mutually exclusive - otherwise why the need for "AB" at all? (If that was the RCA Transmitting Tube Manual's description of class-B, as you say, it was indeed far from correct. It might have been "Designing with class-B tubes" etc. .... )
Simply that.
Agreed! .... but not without some degree of sadness. Again apology for sounding academic, but my dictionary stipulates "fidelity" as "as close as possible to the original". The days of high fidelity are actually gone; these days "what your ears tell you" has become the norm. (Not therewith to open another can of worms, but whose ears? One already has amplifiers like the "on-Gaku" or such and many S.E.T.s having very audible distortion, but preferred by some; the reasons are basic to hearing. I will presume members are aware of the vagaries of hearing - the internet abounds with research results.)
I am the last one to say that is not an individual's right. (I would not have used the word "flawed" the way you did though, rather the word "different".) I can only ask: As a designer, for whose ears do I design? The only clear way open for me is to strive for accuracy, as in "blameless" etc. Otherwise, as Self said, one must add a niceness control.
Bottom line: Design the way you prefer, just don't call it the wrong thing; rather then refrain from calling it 'class-whatever'.
Sensing that we might reach a "boredom point" here, I was merely indicating that unless one starts at and agrees on some basic definitions, confusion will set in. One cannot throw about terms like class-B and class-AB unless there is at least agreement on what each basically means, whether we use them like that or not. They are mutually exclusive - otherwise why the need for "AB" at all? (If that was the RCA Transmitting Tube Manual's description of class-B, as you say, it was indeed far from correct. It might have been "Designing with class-B tubes" etc. .... )
Simply that.
Agreed! .... but not without some degree of sadness. Again apology for sounding academic, but my dictionary stipulates "fidelity" as "as close as possible to the original". The days of high fidelity are actually gone; these days "what your ears tell you" has become the norm. (Not therewith to open another can of worms, but whose ears? One already has amplifiers like the "on-Gaku" or such and many S.E.T.s having very audible distortion, but preferred by some; the reasons are basic to hearing. I will presume members are aware of the vagaries of hearing - the internet abounds with research results.)
I am the last one to say that is not an individual's right. (I would not have used the word "flawed" the way you did though, rather the word "different".) I can only ask: As a designer, for whose ears do I design? The only clear way open for me is to strive for accuracy, as in "blameless" etc. Otherwise, as Self said, one must add a niceness control.
Bottom line: Design the way you prefer, just don't call it the wrong thing; rather then refrain from calling it 'class-whatever'.
- Status
- Not open for further replies.