Member
Joined 2009
Paid Member
You might be interested to read this quote from JLH on publication of his Class AB amplifier noting that it sounds as good as his Class A amplifier:
"The transient response of the 10-watt class A design (as originally published(1), without the modifications(2), suggested in October 1969 to reduce the h.f. bandwidth) is superior to that of the present circuit in the range 50kHz-2Mhz under load conditions of fairly low capacitive reactance. Under more adverse load conditions the present design will be (technically) better. However, the most careful comparative listening trials, with several of the author's long-suffering friends, have failed to uncover any audible difference between these two designs, both of which will almost certainly surpass in performance the best available valve-operated, transformer-coupled units." The Class-A Amplifier Site - JLH Class-AB Amplifier
The circuit you are looking for is here: http://www.diyaudio.com/forums/solid-state/189599-my-little-cheap-circlophone.html
Last edited:
Tinituso I detect the voice of youth? Wait until you are my age,you will want to sit around!
Still4given:Horses for courses.Where high sound levels are required,a good Class AB fits the bill,but at low levels(and when sitting around(see above!) ),the ear tends to be far more critical of what it hears,hence Class A.
Again...when was the last time you listened to a Symasym?
For the cost of an average class A you can build an excellent
class AB amplifier so i really dont see the point to trade quality
for heat and inherent lower reliability since an electronic apparatus
lifecycle is more or less temperature dependent , it works this way
whatever the amplification class , the more components are molested
the less the life expectancy.
May be class D is bright led. It still need much time to be a better light.
Offcourse, some people enjoying its real fire color in a bulb with 50Hz flicker
, but it just a fire color, not the "sun-like" light, while new generation of philips with ultra small tube just closer to real daylight.I spotted some people also working with this incoming classAB, but some of their obstacles are still unsolved.
It really different with those classAB. Most of them are caring its THD and its other measuring results, but this new classAB are more focused at its output impedance characteristics and active damping loop, most at HF because it is very fragile with complex interactions, even in class A design, most expert just failed here and limit the frequency. Here in HF, this new classAB are better because of its less limitations compared to classA.
Sounds interesting. Certainly a known problem with conventional feedback amps. Are there any published papers on this?
Funny that the first post was talking about the inefficiency of class A designs and now we are lost in the woods of personal preference, class generalisations about sound quality, making sweeping statements that 'this' is better than 'that' full stop & writing off entire classes of amp design.
The thread's 'bias' seems to be running a bit wonky - bias.....geddit?? 
Me? I run 5w class A (which is enough to kill me), AB & D. Each has its place & job
The thread's 'bias' seems to be running a bit wonky - bias.....geddit??
Me? I run 5w class A (which is enough to kill me), AB & D. Each has its place & job
I have found no paper, may be there is some somewhere, but better try yourself start playing with damper, and take long time researcing complex things
. Or wait until this new class AB amp available on market.Just like negative feedback affecting sound quality, people said it is just a myth, but it is true. The feedback has the major rule in delayed amp system, because loudspeaker is working with this feedback based damper.
Most classA desingner are really do trial and error, doing tweaking for very long time without knowing what really happened. I too, in the past, also do that trial and error
.
May be this one is more similar than the others, very old and poor document, actually not that simple and not only in bass frequency. The variation on triode amps are easily noticed but on modern amps variations only spotted as fatiguing and less fatiguing sound of lag damper that feel deep at speed-forced slow amps.
http://www.aikenamps.com/dampfac.pdf
http://www.aikenamps.com/dampfac.pdf
Hi Guys
Johan had the correct interpretation back in post-72. It's worth reading again.
There is conflicting information about how good or bad class-A is. Two engineers I respect have opposite views here. On one side, Doug Self has his class-A amp in the form of the Trimodal, A,AB,B amp. Lowest THD is in class-A, next lowest is B and worst is AB.
For those not familiar with the amp, it is his basic Blameless design - a modified and refined Lin - with two bias schemes. A standard bias reg allows setting of the optimal "class-B" bias, which is defined as the lowest THD idle current, and this happens to be around 15mA for CFP or 115mA for an EF output according to other data in his book.
The class-A current is set in the usual way for a split-rail push-pull design, but uses direct current monitoring in the bias control circuit. Personally I think such a circuit should be used for class-B as well, but on with the tome...
Class-AB occurs when the signal pushes out of the class-A range. The evidence that arises is that the part of the circuit turning 'off' has the negative impact on THD. In class-B, turn-off of one side or the other occurs at a very low level that is swamped by other crossover artifacts, many of which can be tamed. As an aside, one fix that Soulution uses, is to make the turn-off so quick that the detriments of that happening are much reduced.
Doug favours class-B for most things, but if you don't have to worry about the electricity bill says class-A is better if done correctly.
Doing class-A correctly has more to do with the overall circuit design than simply setting the output idle current way high.
Contradicting this is the word of Chris Russell, designer of the fantastic Bryston amplifiers and preamps. Early on he found that raising idle current and thus heating the BJTs had an adverse affect on THD, to the point of making distortion an order of magnitude higher. The device manufacturer's data supported this.
For my own listening room, I need very little power. A class-A amp allowing 1-2W per channel would move the AB transition to a listening SPL higher than I want to bear anyway, and designing for such low power keeps total dissipation low enough to tolerate in the summer. On the other hand, if the amp could be biased at optimal class-B then there is a certain peace of mind in that coolness. The realisation of the latter was in adapting Bryston's output stage as a follower, taking advantage of many of the benefits of the unique Quad Complimentary output stage Chris developed back in the late '70s or so once decent complimentary pairs were available. Thanks Chris!
From most expert writings, the state of design seems to be stuck with some amount of idle current resulting in some small or large class-A region. My experience with the JLH 10 is that it is utter crap. If you push it you get mud. Doug Self analysed it in an EW article finding no idle condition where the circuit halves worked well together or where THD could be minimised. This fully relates to the sound of the circuit. If JLH's friends could not tell this one from his other creations, to me that says little about those other circuits.
JLH is a great guy but many of his ideas were incorrect, based on rules of thumb that themselves evolved for economic reasons rather than purely technical ones. Doug perpetuates some of those ideas, as his career spans back to 1970, as well.
Overall, it is possible to design good performing and good sounding class-A and class-B amps - as well as bad ones of each.
Have fun
Kevin O'Connor
Johan had the correct interpretation back in post-72. It's worth reading again.
There is conflicting information about how good or bad class-A is. Two engineers I respect have opposite views here. On one side, Doug Self has his class-A amp in the form of the Trimodal, A,AB,B amp. Lowest THD is in class-A, next lowest is B and worst is AB.
For those not familiar with the amp, it is his basic Blameless design - a modified and refined Lin - with two bias schemes. A standard bias reg allows setting of the optimal "class-B" bias, which is defined as the lowest THD idle current, and this happens to be around 15mA for CFP or 115mA for an EF output according to other data in his book.
The class-A current is set in the usual way for a split-rail push-pull design, but uses direct current monitoring in the bias control circuit. Personally I think such a circuit should be used for class-B as well, but on with the tome...
Class-AB occurs when the signal pushes out of the class-A range. The evidence that arises is that the part of the circuit turning 'off' has the negative impact on THD. In class-B, turn-off of one side or the other occurs at a very low level that is swamped by other crossover artifacts, many of which can be tamed. As an aside, one fix that Soulution uses, is to make the turn-off so quick that the detriments of that happening are much reduced.
Doug favours class-B for most things, but if you don't have to worry about the electricity bill says class-A is better if done correctly.
Doing class-A correctly has more to do with the overall circuit design than simply setting the output idle current way high.
Contradicting this is the word of Chris Russell, designer of the fantastic Bryston amplifiers and preamps. Early on he found that raising idle current and thus heating the BJTs had an adverse affect on THD, to the point of making distortion an order of magnitude higher. The device manufacturer's data supported this.
For my own listening room, I need very little power. A class-A amp allowing 1-2W per channel would move the AB transition to a listening SPL higher than I want to bear anyway, and designing for such low power keeps total dissipation low enough to tolerate in the summer. On the other hand, if the amp could be biased at optimal class-B then there is a certain peace of mind in that coolness. The realisation of the latter was in adapting Bryston's output stage as a follower, taking advantage of many of the benefits of the unique Quad Complimentary output stage Chris developed back in the late '70s or so once decent complimentary pairs were available. Thanks Chris!
From most expert writings, the state of design seems to be stuck with some amount of idle current resulting in some small or large class-A region. My experience with the JLH 10 is that it is utter crap. If you push it you get mud. Doug Self analysed it in an EW article finding no idle condition where the circuit halves worked well together or where THD could be minimised. This fully relates to the sound of the circuit. If JLH's friends could not tell this one from his other creations, to me that says little about those other circuits.
JLH is a great guy but many of his ideas were incorrect, based on rules of thumb that themselves evolved for economic reasons rather than purely technical ones. Doug perpetuates some of those ideas, as his career spans back to 1970, as well.
Overall, it is possible to design good performing and good sounding class-A and class-B amps - as well as bad ones of each.
Have fun
Kevin O'Connor
Last edited:
Funny that the first post was talking about the inefficiency of class A designs and now we are lost in the woods of personal preference, class generalisations about sound quality, making sweeping statements that 'this' is better than 'that' full stop & writing off entire classes of amp design.
The thread's 'bias' seems to be running a bit wonky - bias.....geddit??
Me? I run 5w class A (which is enough to kill me), AB & D. Each has its place & job
I agree with JKRO.... most of this is just words of preference.
How you doing mate?, Long time no hear from - will send you an email drekly, interesting things to discuss etc
Bryston is not contradicting D.Self.
Bryston's statement is correct. An overbiased, relative to optimally biased, output stage gives more distortion.
D.Self says EXACTLY the same.
Overbiasing an optimally biased output stage gives more distortion.
The two designers are in agreement.
Is it possible that D.Self's adoption of "his" definition of ClassB is quite different from everyone one else that I have read and this is misleading you?
No, this is not true Wahab. Class A when done properly, sound excellent, and performance is also good.
As for running hot, modern semis are VERY good in this regard. An enormous amount of effort has gone into solving and obviating the die- header and bonding mechanical interface that was in the old days a major cause of failure.
Hi Guys
Bryston and Self do not contradict on the class-B definition or performance.
The contradiction I was suggesting was that Doug's statements that class-A is superior in performance contradicts Bryston's views and the results with their products.
So actually... maybe it is Self who contradicts Self about class-A vs class-B?
Have fun
Kevin O'Connor
Bryston and Self do not contradict on the class-B definition or performance.
The contradiction I was suggesting was that Doug's statements that class-A is superior in performance contradicts Bryston's views and the results with their products.
So actually... maybe it is Self who contradicts Self about class-A vs class-B?
Have fun
Kevin O'Connor
Oh Dear. Where are we now........
I read with trepidation what the formidable Douglas Self found. I will have to find that and study ....
For now: I guess one will have to make peace with the "pure class-A superiority" cliché. As Ken Kessler once said: "Repeat anything often enough and everybody will believe it." In practice in both class-A and class-AB there are poor and excellent designs. One will have to be careful to characterise a topology simply by its execution in a particular commercial amplifier or circuit.
I have a problem, current-wise, with this is optimal and that is optimal per se. Having played a lot with Spice, I found there is a particuler 'optimum' for each set of BJTs and secondary, circuit configuration.
Then, THD (somewhat OT). There have been many publications indicating that that figure can be rather misleading as a measure of eventual music enjoyment. It is probably known by now that it is the composition of that figure that reveals the true propensity for enjoyment/fatigue. One is speaking of high order harmonic distortion. (Revision: 2nd and 3rd products can be tolerated to some 1%, but when orders of 7th - 11th are present to any degree even as low as the threshold of hearing, plus their intermodulation, mucho fatigue is inevitable.) Thus again there is more to the final 'acceptability' than simple class distinction.
For me, I could simply find no reason for distiction between class-A and -AB over many dozens of analyses - and I am talking of distortion of <0,01% at very low up to full output (80W), using moderate NFB factors of around 30dB (that will be global; there is obviously nested feedback as e.g. in most output configurations).
[How did they sound? Sorry gents, at my age my hearing is no good any more at >6kHz apart from the right ear being down by 8 dB. But others are quite satisfied. I make an 80W stereo job much appreciated by all sorts of ears. So I am probably doing something right ....]
I read with trepidation what the formidable Douglas Self found. I will have to find that and study ....
For now: I guess one will have to make peace with the "pure class-A superiority" cliché. As Ken Kessler once said: "Repeat anything often enough and everybody will believe it." In practice in both class-A and class-AB there are poor and excellent designs. One will have to be careful to characterise a topology simply by its execution in a particular commercial amplifier or circuit.
I have a problem, current-wise, with this is optimal and that is optimal per se. Having played a lot with Spice, I found there is a particuler 'optimum' for each set of BJTs and secondary, circuit configuration.
Then, THD (somewhat OT). There have been many publications indicating that that figure can be rather misleading as a measure of eventual music enjoyment. It is probably known by now that it is the composition of that figure that reveals the true propensity for enjoyment/fatigue. One is speaking of high order harmonic distortion. (Revision: 2nd and 3rd products can be tolerated to some 1%, but when orders of 7th - 11th are present to any degree even as low as the threshold of hearing, plus their intermodulation, mucho fatigue is inevitable.) Thus again there is more to the final 'acceptability' than simple class distinction.
For me, I could simply find no reason for distiction between class-A and -AB over many dozens of analyses - and I am talking of distortion of <0,01% at very low up to full output (80W), using moderate NFB factors of around 30dB (that will be global; there is obviously nested feedback as e.g. in most output configurations).
[How did they sound? Sorry gents, at my age my hearing is no good any more at >6kHz apart from the right ear being down by 8 dB. But others are quite satisfied. I make an 80W stereo job much appreciated by all sorts of ears. So I am probably doing something right ....]
Hi Guys
AndrewT, the numbers quoted from Bryston were from an investigation Chris did decades ago to see if there would be any benefit to class-A. He found none, and rather disturbingly found detriment to biasing so high. His amps are biased at the optimal class-B point as Doug describes for his class-B designs. (Note that I never stated how high the bias was raised too, so assuming it was just to high class-AB is your own interpretation)
The issue is class-A vs class-B - per the thread suggestion. Chris says 'no' to class-A. Doug says 'yes'.
Everyone has generally considered the "optimal class-B" biasing to be class-AB until Doug presented his definition. Truly, that is what confuses many about the A,B,AB debate.
If "optimal class-B" is considered as AB, then the debate is between A and AB, more exactly per the thread name. Confusing this again is Doug's evidence of THD results for AB occurring at a very low signal level being "good" versus occurring at a very high level being "bad". Good and bad are just with respect to the THD level itself, where bad is higher.
I think the practical implication of Doug's results are counterintuitive. One would expect THD at low signal levels to be more annoying that that which occurs above most music levels. The latter is masked by the excessive loudness and aural compression, whereas the signal always traverses the low-level region and for the most part lives there. The THD referenced to this low power level would have to be very small indeed to be musically unimportant. "Buried in the noise floor" might be low enough but it also might not be?
I like clear, transparent sound. I like cool amplifier operation. I like knowing that the performance of the gear I'm using is as good as it can be. It seems that there are multiple paths to all of these goals.
Have fun
Kevin O'Connor
AndrewT, the numbers quoted from Bryston were from an investigation Chris did decades ago to see if there would be any benefit to class-A. He found none, and rather disturbingly found detriment to biasing so high. His amps are biased at the optimal class-B point as Doug describes for his class-B designs. (Note that I never stated how high the bias was raised too, so assuming it was just to high class-AB is your own interpretation)
The issue is class-A vs class-B - per the thread suggestion. Chris says 'no' to class-A. Doug says 'yes'.
Everyone has generally considered the "optimal class-B" biasing to be class-AB until Doug presented his definition. Truly, that is what confuses many about the A,B,AB debate.
If "optimal class-B" is considered as AB, then the debate is between A and AB, more exactly per the thread name. Confusing this again is Doug's evidence of THD results for AB occurring at a very low signal level being "good" versus occurring at a very high level being "bad". Good and bad are just with respect to the THD level itself, where bad is higher.
I think the practical implication of Doug's results are counterintuitive. One would expect THD at low signal levels to be more annoying that that which occurs above most music levels. The latter is masked by the excessive loudness and aural compression, whereas the signal always traverses the low-level region and for the most part lives there. The THD referenced to this low power level would have to be very small indeed to be musically unimportant. "Buried in the noise floor" might be low enough but it also might not be?
I like clear, transparent sound. I like cool amplifier operation. I like knowing that the performance of the gear I'm using is as good as it can be. It seems that there are multiple paths to all of these goals.
Have fun
Kevin O'Connor
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.