To me, what matters for hobby circuits is how much fun I have from them. So if you consider it fun to build your own amplifier, just go ahead with it. Whether you actually believe it to sound better than a commercial amplifier is irrelevant. (I consider circuits that are somewhat unconventional more fun than Self's blameless designs, but that is just my personal abberation.)
There are even hobbyists who like to build their own 32-line electromechanical black and orange television sets. None of them believe that their television sets give a better picture quality than a modern ultra-HD colour television, but they have much fun building them anyway.
Last edited:
Good argument. In my early days I built amps from just that reason. Usually circuits that other had constructed.
Nowadays I think it's a good sport to try to surpass myself when it comes to subjective sound quality. Every amp learns me something new.
But if I really believed that all this was futile, that all well amps sounds the same, yes then I wouldn't bother anymore.
And we perhaps must distinguish between audiophiles and DIY builders. The latter perhaps have lower ambitions. Or - which is quite usual - are more susceptible to raw performance and nice figures, etc..
The audiophile part of this forum usually resides in the Passlabs part. I probably belong there, but since I build my own minimalistic amps, I feel like an intruder.
Nowadays I think it's a good sport to try to surpass myself when it comes to subjective sound quality. Every amp learns me something new.
But if I really believed that all this was futile, that all well amps sounds the same, yes then I wouldn't bother anymore.
And we perhaps must distinguish between audiophiles and DIY builders. The latter perhaps have lower ambitions. Or - which is quite usual - are more susceptible to raw performance and nice figures, etc..
The audiophile part of this forum usually resides in the Passlabs part. I probably belong there, but since I build my own minimalistic amps, I feel like an intruder.
A few thoughts on "subjectivists vs objectivists"
Subjectivists = to feel it arouse the senses, stir the emotion.
Objectivists = measure and discover.
But I wonder who makes the blueprint?
Bruce de Palma puts it this way...
"The choice is in the eye of fhe beholder and in the ear of the listener"
Don't you think these two discipline should work in harmony?
Subjectivists = to feel it arouse the senses, stir the emotion.
Objectivists = measure and discover.
But I wonder who makes the blueprint?
Bruce de Palma puts it this way...
"The choice is in the eye of fhe beholder and in the ear of the listener"
Don't you think these two discipline should work in harmony?
Ha, I actually think it might be so that the reason for subjectivists and objectivists not usually are compatible is that they in their depths of their soul are a bit uncertain.
I am a subjectivist and sometimes I think a bit like this:
"What if I'm deluding myself completely? Perhaps I'm lost in a pleasant placebo fog "
The objectivists on the other hand may be afraid of losing grip. They want a nice world of figures and measurements, and suddenly some guy urges me to feel and go in to my own personal impressions. Nasty. I want clarity.
But these are extremes, but sadly they aren't very uncommon. Usually we make our choices early in life.
But is this off topic? Rather psychology.
I am a subjectivist and sometimes I think a bit like this:
"What if I'm deluding myself completely? Perhaps I'm lost in a pleasant placebo fog "
The objectivists on the other hand may be afraid of losing grip. They want a nice world of figures and measurements, and suddenly some guy urges me to feel and go in to my own personal impressions. Nasty. I want clarity.
But these are extremes, but sadly they aren't very uncommon. Usually we make our choices early in life.
But is this off topic? Rather psychology.
The first time I moded an amp, I am soldering the components under the board avoiding not to do damage on the copper traces by repeated soldering. After few probing here and there, juzt making sure no abnormalities was happening. [the safety protocol 
] I fired it up and listened a lot and I exclaimed nahhh ...and the cycle of adjusting values and soldering repeats again.
To date, I learned CAD and Spice and I spend long hours in simulation and evaluating results....and I find myself looking for the better figures.
Looking back, it makes me think what has become of me now? A subjectivists objective?
Technology is very challenging, it awakens your inquisitve mind because you find yourself pushing the wheels ...and now I'm off topic. Sorry SaSi.
] I fired it up and listened a lot and I exclaimed nahhh ...and the cycle of adjusting values and soldering repeats again.To date, I learned CAD and Spice and I spend long hours in simulation and evaluating results....and I find myself looking for the better figures.
Looking back, it makes me think what has become of me now? A subjectivists objective?
Technology is very challenging, it awakens your inquisitve mind because you find yourself pushing the wheels ...and now I'm off topic. Sorry SaSi.
I think posts 3, 6 and 10 have quite completely answered the thread starter's question, after that the whole thread went off topic.
The strict definition of Class B is when the angle of flow in an active device is 180 degrees or a half cycle in a power amplifier output stage. Class A is 360 degrees and Class AB anywhere between these boundaries. Most Hi-Fi amplifiers run a modest standing current through the output stage under no signal conditions.
That puts them across the line into AB territory but only just - so they might be regarded as defacto Class B by some due to the close proximity with the boundary.
There are problems with correct adjustment of bias which is heat related and finicky whatever the operating mode - forcing operation more deeply into Class AB will increase the impact of heat.
Also the crossover point will move from pretty much a signal level to an intermediate power level where the disparity in output halves will be more prominent.
As I see it Self stronly warns against going down that road.
Anyone arguing for either voltage would simply be showing that he does not understand bipolar junction transistors. There is no 'switch-on voltage' or 'minimum bias', but an exponential rise which just happens to give significant current somewhere around the 0.5-0.8V region. 0.6V is a convenient approximation for rough calculations, and a convenient simplification for teaching newbies.SaSi said:
I think the problem is that people misunderstand the oversimplified explanations given in the elementary textbooks. Hence they think that Class B means zero bias, when even the simple textbook model actually means zero quiescent current - which could mean positive or negative bias, depending on the device used. It is unfortunate that the simple textbooks often assume a device where zero bias is correct, so people then think that this (and not zero current) is the definition of Class B.sgrossklass said:
So I think that 'common usage' is wrong and not what the textbooks actually mean, while Self is right.
Music is about feelings. Hi-fi is about reproducing music - a purely engineering function. Tension in blind testing will only occur if what is being tested is the listener rather than the audio system. This occurs when the listener has previously made claims, either publicly or within his own private thoughts, about his hearing ability which now need to be backed up by evidence.Svitjod said:
Whether you like the stuff or not has nothing to do with hi-fi. Most people prefer something different from hi-fi: they want modified frequency response and/or some added low order distortion.
Some of us enjoy the technical challenge of thinking about, designing and building audio stuff. We don't kid ourselves that it actually has better fidelity than any competent stuff made by someone else.
Perhaps it is a not technical cause. Perhaps your ears just prefer the little bit of more distortion given by class AB. It is not a mystery that "good sounding" often refers to more distortion and not to other measurable (or un-measurable) features.
I agree with the first comment in post 30.
Returning to the AB theme and argument against this - it is easier to switch a transistor on or off in the turn on region since the base charging and discharging currents are less and the current flow between emitter and collector will be low.
The base charge is a fraction of the charges between emitter and collector and when the signal stimulus on the base is reversed the larger current will have the effect of holding some charge in place within the base region after the forward conduction stimulus is terminated.
The effect is inconsequentially benign when the emitter and collector current is low as it is in the turn on region however it is exponential with increasing current. When this is extreme it is described as stored charge effect which can promote mutual conduction. There is some interesting reading on Bob Cordell's website about output stage speed.
Outside the turn on region the angle of transistor IC versus Vbe slope increases steeply - driver transistors contribute to overall steepness of slope.
Consider IC current on the vertical axis and the distance between vertical points drawn on the slope in a graph to intersect the Vbe horizontal axis and compare this with the turn on area.
Returning to the AB theme and argument against this - it is easier to switch a transistor on or off in the turn on region since the base charging and discharging currents are less and the current flow between emitter and collector will be low.
The base charge is a fraction of the charges between emitter and collector and when the signal stimulus on the base is reversed the larger current will have the effect of holding some charge in place within the base region after the forward conduction stimulus is terminated.
The effect is inconsequentially benign when the emitter and collector current is low as it is in the turn on region however it is exponential with increasing current. When this is extreme it is described as stored charge effect which can promote mutual conduction. There is some interesting reading on Bob Cordell's website about output stage speed.
Outside the turn on region the angle of transistor IC versus Vbe slope increases steeply - driver transistors contribute to overall steepness of slope.
Consider IC current on the vertical axis and the distance between vertical points drawn on the slope in a graph to intersect the Vbe horizontal axis and compare this with the turn on area.
The A-B-C lesson is just so a junior EE student can tell a cow from a kangaroo across the field.
Real devices don't fit the hard distinction.
"Class B" is a real mis-begotten child.
The idea, in audio, seems to be that it can sit at zero current, yet leap into action on the smallest signal. But the gain (transconductance, Gm) of all devices drops with current, to zero at zero current.
A field that many here have never explored, but has some to tell us, is radio amplifiers. In AM and SSB radio, the shape of the 1MHz wave is not important. It will be rounded-out by the next tuned circuit. But the amplitude of the 50Hz-5KHz audio modulation must be preserved very well.
You can actually cut HALF the 1MHz wave away and still preserve the audio waveform.
So an amplifier that worked as a precision rectifier would be fine. This could be a single device.
There is no simple way to do that. But does it have to be "exactly half the wave"?
No. You can find a low-current point where the Gm is "about half" of the high-current Gm. If you swing both ways from this point, the modulation amplitude is preserved (for some assumptions of how smoothly the Gm changes with current).
And for plate-loaded vacuum-tubes, the "linear" point is very nearly the "cutoff" point predicted by extending the curve for the highest currents. In my diagram, straight red line is an ideal half-wave rectifier, bent blue curve and dot is where a "linear B" amplifier would be biased.
This is small benefit for AM transmitters. (However AM IF strips often run this way for large signals.) AM modulation has large carrier even when audio is silent. Power saving is small. It became a great benefit when SSB came in, because SSB degenerates to a no-carrier condition when the talker shuts-up, and the amplifier power consumption goes small (not zero).
I do not know. But I wonder if Doug Self understands radio B linear operation better than most. Understands that the "extended cutoff" operating point ensures good linearity over a wide range of currents. While we still need two devices to cover the whole audio wave, we can pick the "cutoff" so they merge sweetly. Using the same concepts used for "Class B linear" radio amplifiers.
Biasing an emitter-loaded BJT to the equivalent point is a little more tricky. NFB from both load and emitter resistors makes it "easy" to be clean over 99% of the swing but complicates that last 1%. Perfectly possible, with insight and patience.
Real devices don't fit the hard distinction.
"Class B" is a real mis-begotten child.
The idea, in audio, seems to be that it can sit at zero current, yet leap into action on the smallest signal. But the gain (transconductance, Gm) of all devices drops with current, to zero at zero current.
A field that many here have never explored, but has some to tell us, is radio amplifiers. In AM and SSB radio, the shape of the 1MHz wave is not important. It will be rounded-out by the next tuned circuit. But the amplitude of the 50Hz-5KHz audio modulation must be preserved very well.
You can actually cut HALF the 1MHz wave away and still preserve the audio waveform.
So an amplifier that worked as a precision rectifier would be fine. This could be a single device.
There is no simple way to do that. But does it have to be "exactly half the wave"?
No. You can find a low-current point where the Gm is "about half" of the high-current Gm. If you swing both ways from this point, the modulation amplitude is preserved (for some assumptions of how smoothly the Gm changes with current).
And for plate-loaded vacuum-tubes, the "linear" point is very nearly the "cutoff" point predicted by extending the curve for the highest currents. In my diagram, straight red line is an ideal half-wave rectifier, bent blue curve and dot is where a "linear B" amplifier would be biased.
This is small benefit for AM transmitters. (However AM IF strips often run this way for large signals.) AM modulation has large carrier even when audio is silent. Power saving is small. It became a great benefit when SSB came in, because SSB degenerates to a no-carrier condition when the talker shuts-up, and the amplifier power consumption goes small (not zero).
I do not know. But I wonder if Doug Self understands radio B linear operation better than most. Understands that the "extended cutoff" operating point ensures good linearity over a wide range of currents. While we still need two devices to cover the whole audio wave, we can pick the "cutoff" so they merge sweetly. Using the same concepts used for "Class B linear" radio amplifiers.
Biasing an emitter-loaded BJT to the equivalent point is a little more tricky. NFB from both load and emitter resistors makes it "easy" to be clean over 99% of the swing but complicates that last 1%. Perfectly possible, with insight and patience.
Attachments
Forr:
Yes, it's quite well known that some people seem to like devices that distorts in a benign way.
Also, some not so very experienced wannabee audiophiles seems to be stimulated by amps that gives a broad spectrum of THD. I think the THD ( which is more pronounced at higher frequencies due to the usual NFB that is set by a capacitor ) boosts the highs and gives an impression of detailing.
But when it comes to biasing, my personal impression is that the usual low bias gives a more "soft" or "foggy" sound. When the bias is tuned up, the sound becomes more focused. Not everyone likes that. So for most people, the low bias situation is preferred.
Perhaps you have heard of the NOS DAC's. The ones that has no output filter or oversampling. I suspect that the heavy distortion that comes from that will stimulate some people. Perhaps.
But as I said earlier, an amp biased to 250mA will work almost entirely in the class A region under normal circumstances ( quite high volume )and with normal sensitivity of the speakers.
To increase the bias from 30mA to 60 is probably of no use. It has to be hoisted quite a bit.
PRR:
That was really advanced, I didn't really get the point.
Yes, it's quite well known that some people seem to like devices that distorts in a benign way.
Also, some not so very experienced wannabee audiophiles seems to be stimulated by amps that gives a broad spectrum of THD. I think the THD ( which is more pronounced at higher frequencies due to the usual NFB that is set by a capacitor ) boosts the highs and gives an impression of detailing.
But when it comes to biasing, my personal impression is that the usual low bias gives a more "soft" or "foggy" sound. When the bias is tuned up, the sound becomes more focused. Not everyone likes that. So for most people, the low bias situation is preferred.
Perhaps you have heard of the NOS DAC's. The ones that has no output filter or oversampling. I suspect that the heavy distortion that comes from that will stimulate some people. Perhaps.
But as I said earlier, an amp biased to 250mA will work almost entirely in the class A region under normal circumstances ( quite high volume )and with normal sensitivity of the speakers.
To increase the bias from 30mA to 60 is probably of no use. It has to be hoisted quite a bit.
PRR:
That was really advanced, I didn't really get the point.
There was a reliance on Class B in the strict form in the early days of transistor amplifiers which led to complaints about the thin nature of the tone and listener fatigue. By rule of thumb Gm is roughly 40 per m.A of collector current. A Gm value of 20 equates to .5 mA.
One has to allow for changes in season climate and worldwide for the life of the equipment.
The manufacturers recommendations on setting standing current are higher of necessity to conform with the thd specifications in their advertising material. In that regard a dead band is not admissible.
--
We ALL like distortion and clean sound
We like THEM BOTH So we need an amp with a pot for balancing clean<>distorted xD
Like a crossfader between two input stages: one with negative feedback, another without.
Last edited:
--
We ALL like distortion and clean sound
So we need an amp with a pot for balancing clean<>distorted xD
Like a crossfader between two input stages: one with negative feedback, another without.
That may work with the glass bottle amplifiers. I have heard an example that could be switched in this way. The demonstration took place in a class room of a tertiary learning institution. It was a two man lift to get it into the building. I have information on the project behind the build. I am away from home so looking at this will have to wait for a week.
As I understand Self's position on crossover distortion, He wants to avoid where the output impedance of both transistors are driving hence lower Z. Really, if you adjust the bias for a minimum distortion then you have done that. I think his point is that the minimum distortion is not the maximum (~AB) bias.
Years ago when I repaired amps for professional sound, I usually under-biased amps to favor durability over sonic purity, in order to reduce the odds of repeated failure. Self promotes that I have since come to appreciate more, is cross coupling since it turns out the cause of bi-polar amp failure is often shoot through from driving high frequencies. Cross coupling adds active off-drive, which drastically reduces shoot-through current. Cross coupling also allows you to reduce driver bias and dissipation, another cause of failure, and by making cross-over softer, put the crossover distortion down within the control bandwidth of the feedback.
Years ago when I repaired amps for professional sound, I usually under-biased amps to favor durability over sonic purity, in order to reduce the odds of repeated failure. Self promotes that I have since come to appreciate more, is cross coupling since it turns out the cause of bi-polar amp failure is often shoot through from driving high frequencies. Cross coupling adds active off-drive, which drastically reduces shoot-through current. Cross coupling also allows you to reduce driver bias and dissipation, another cause of failure, and by making cross-over softer, put the crossover distortion down within the control bandwidth of the feedback.
Good analogy! The trouble is that when they get too close to a female kangaroo they notice that it has nipples inside the pouch which look rather like a miniature version of a cow's udders so then they ask whether a kangaroo is really just a jumping cow. Arguments then break out between those who want to concentrate on the nipples/udders and those who want to measure how high a cow can jump.PRR said:
No distortion, but lots of images. As high frequencies do not carry pitch it may be that the images somehow sound like the source material which has been removed by the anti-aliasing filter. NOS DACs are a good example of how avoiding high fidelity sound reproduction can sound better to some people.Svitjod said:
He was explaining how to do Class B properly. I can't remember whether Self has any RF experience, but he may have. RF knowledge certainly helps make better audio.
I would not call that a rule of thumb; it is more a statement of physical fact (around room temperature).mjona said:
Depends on the output stage topology. If you have a thermally-stable CFP output then low current is needed for low THD.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.