Hello Lumba Ogir and SandyK,
I have built Dougs designs, his blameless Class B being one of them. It was built on the official PCB,s and built EXACTLY to the construction details supplied, no component substitutions at all.
The wiring followed Dougs recommendations also included with the PCB's. It was used with a passive pre amp.
I still have the amp and dragged it out about 6 months ago to have another listen as my listening room had been redesigned and I now had much better speakers B&W703's.
I have to agree with your comments-- I found it "boring", there was no "soundstage" it was all two dimensional, instruments were "cold" no warmth or vibrancy. Going back to my MOSFET design was truly a revelation.
I would further add that in my experience designs like Dougs, and there are many that use similar topologies, do have this "same sound" about them.
Without wishing to detract from Dougs work-- and his knowledge of the subject is second to none I believe-- my conclusion is that technical perfection is not the whole answer to that "Magic" factor that some designs seem to have.
Regards Karl
I have built Dougs designs, his blameless Class B being one of them. It was built on the official PCB,s and built EXACTLY to the construction details supplied, no component substitutions at all.
The wiring followed Dougs recommendations also included with the PCB's. It was used with a passive pre amp.
I still have the amp and dragged it out about 6 months ago to have another listen as my listening room had been redesigned and I now had much better speakers B&W703's.
I have to agree with your comments-- I found it "boring", there was no "soundstage" it was all two dimensional, instruments were "cold" no warmth or vibrancy. Going back to my MOSFET design was truly a revelation.
I would further add that in my experience designs like Dougs, and there are many that use similar topologies, do have this "same sound" about them.
Without wishing to detract from Dougs work-- and his knowledge of the subject is second to none I believe-- my conclusion is that technical perfection is not the whole answer to that "Magic" factor that some designs seem to have.
Regards Karl
Whyt does a Class A distortion increase with frequency ?
Mooly
Do you have a schematic of the actual amplifier you constructed, or does it appear in Issue 2 of the D.S. book ? If so, where ?
My email address is alexkethel@optusnet.com.au
Several DiyAudio members from Sydney have been able to make kits based on the D.S. topology really sing, with large 3D soundstages and a natural sounding warmth. Especially the Class A variants, however, Class AB versions can also be made to sound markedly better.
SandyK
Mooly
Do you have a schematic of the actual amplifier you constructed, or does it appear in Issue 2 of the D.S. book ? If so, where ?
My email address is alexkethel@optusnet.com.au
Several DiyAudio members from Sydney have been able to make kits based on the D.S. topology really sing, with large 3D soundstages and a natural sounding warmth. Especially the Class A variants, however, Class AB versions can also be made to sound markedly better.
SandyK
Accepted. We all know Self's approach is primarily driven by quantitative measurements.Nelson Pass said:
I don't know about the "no bias at all" bit. I meant a Class B amp with bias carefully adjusted to minimise THD at a few Watts output. This means there's some bias current flowing through the OPS devices, right? Is that what you mean by "no bias at all"? If yes, then that's what I meant too.
Found the PDF, am reading it. Thanks a lot.
I'll read your article and then return to this discussion. Thanks a lot. I'm pretty sure there's stuff in there I'll find very interesting.
Whats the alternative conclusion --- that's easy --- Listen to an amp that you really enjoy. One that makes the music "come alive"
The conclusion is that many people don't seem to like "the piece of wire with gain" in much the same way that musicians prefer one instrument over another. An old Stradivarius or Testore wasn't designed with computer simulation, yet a cheap modern instrument probably was, and is probably technically superior. But which would you rather play ??
The conclusion is that many people don't seem to like "the piece of wire with gain" in much the same way that musicians prefer one instrument over another. An old Stradivarius or Testore wasn't designed with computer simulation, yet a cheap modern instrument probably was, and is probably technically superior. But which would you rather play ??
Read the article. It was an eye-opener. I didn't know Class A OPS had such low open-loop distortion. I think I'll go away and think some more about this entire Class A HF distortion thing.Nelson Pass said:
What measured closed-loop distortion would you get with a high-bias Class A, if you applied a lot of GNFB? Do you have any distortion graphs for such cases handy? I know you don't prefer such amps, but I'm just curious to know what the "numbers" look like.
Mooly,
No, no. Keep your wits about you. You are drifting into AKSA territory now.
You listened to Self's design and found it lacks magic. You assumed Self is a leading expert at making "technically" accurate circuits. Therefore, you've concluded, that technically accurate circuits don't sound great.
Are you sure?
No, no. Keep your wits about you. You are drifting into AKSA territory now.
You listened to Self's design and found it lacks magic. You assumed Self is a leading expert at making "technically" accurate circuits. Therefore, you've concluded, that technically accurate circuits don't sound great.
Are you sure?
Wow. That's impressive measured performance at 20KHz. Found the article. Will read more details at leisure. Thanks for the pointer. From some comments in some forums, it seems it's basically Self's Blameless, biased into Class A the way he'd demonstrated it in his book. Is this the case?sandyK said:
Hi,
there is no thing as high bias ClassA. You mean high bias ClassAB where the bias is set above optimum ClassAB.
The bias for ClassA is set to provide the desired output current into your chosen load.
D.Self defined "his ClassB" and he makes it clear that Self ClassB = optimum bias ClassAB that almost everyone else uses.
there is no thing as high bias ClassA. You mean high bias ClassAB where the bias is set above optimum ClassAB.
The bias for ClassA is set to provide the desired output current into your chosen load.
D.Self defined "his ClassB" and he makes it clear that Self ClassB = optimum bias ClassAB that almost everyone else uses.
traderbam said:
I am not sure no All I do know for SURE is that the more "Technically Perfect" amps I have owned have had the least enjoyable sound. Commercial amps included in this as well.
My current design --- and it's certainly not perfect--- far from it technically, is the amp that has given me by far the most listening pleasure of any. Not just some of the time, with a few chosen discs, but consistently year after year with 90% or so of my music collection.
Hi,
in the beginning there were the great triodes of the thirties, the most magnificent amplification devices ever made producing an exceptionally favourable distortion spectrum. Triode amplifiers may serve as dignifying references for any design application. The THD level of the blameless amplifiers may be thousand times lower than in triode amplifiers indicating theoretical irregularities. The price of that is appalling.
When the more powerful but less linear penthods came the new revolutionary invention, GNF was utilized (firstly meant for multiple cascaded stages in telephone lines). Some people noticed pretty soon that while the distortion was much lower, the sound quality much worse. After six decades of heavy (ab)use we are still facing the same agonizing dilemma.
in the beginning there were the great triodes of the thirties, the most magnificent amplification devices ever made producing an exceptionally favourable distortion spectrum. Triode amplifiers may serve as dignifying references for any design application. The THD level of the blameless amplifiers may be thousand times lower than in triode amplifiers indicating theoretical irregularities. The price of that is appalling.
When the more powerful but less linear penthods came the new revolutionary invention, GNF was utilized (firstly meant for multiple cascaded stages in telephone lines). Some people noticed pretty soon that while the distortion was much lower, the sound quality much worse. After six decades of heavy (ab)use we are still facing the same agonizing dilemma.
Lumba Ogir
It seems likely to my feeble brain, that for global feedback to work as intended, that we need bandwidth far in excess of the accepted norm , for feedback to do it's job properly without smearing of the waveform that it is trying to correct.
But I have been wrong many times before ! Probably showing my ignorance, now, too !
SandyK
It seems likely to my feeble brain, that for global feedback to work as intended, that we need bandwidth far in excess of the accepted norm , for feedback to do it's job properly without smearing of the waveform that it is trying to correct.
But I have been wrong many times before ! Probably showing my ignorance, now, too !
SandyK
Hi Andrew, I presume this comment was in response to mine.AndrewT said:
Since we're in terminology territory, I'll have to tread carefully so that my conceptual weaknesses/confusion doesn't trip me up.
In that case, I agree with you about Self's Class B being what most people call Class AB. But about "high bias Class A" versus "lower bias Class A", I've borrowed the terms from Nelson's article. There he basically discusses how increasing the bias current reduces OPS open-loop distortion. Why don't you take a look at his note? It's very interesting and as always very well written.
This is very different from high-bias Class AB, which some designers prefer over optimum-bias Class B. In high-bias Class AB, both OPS devices conduct for more than 180-deg of the cycle, but neither conducts for the full 360-deg. What Nelson's article was talking about was pure Class A, where both OPS devices conduct for the full 360-deg. In that situation, he has demonstrated that higher current is delivering lower THD.
From Nelson's article, I didn't understand what Nelson means by "Class B, i.e. no bias". Does he mean actually zero bias current, or does he mean the minimum bias current needed for exactly 180-deg conduction? Maybe I need to re-read the note. But in any case, his note focused on Class A rather than Class B issues, so perhaps it's peripheral to the focus of his note.
Your ears are the BEST measurement device for interpreting sound that you have. ALWAYS trust your ears.Mooly said:
All I do know for SURE is that the more "Technically Perfect" amps I have owned have had the least enjoyable sound. Commercial amps included in this as well.
My current design --- and it's certainly not perfect--- far from it technically, is the amp that has given me by far the most listening pleasure of any. Not just some of the time, with a few chosen discs, but consistently year after year with 90% or so of my music collection.
There are two things going on that cause confusion.
1) One is that standard "techincal" measures are incomplete. So it is possible for a standard measure to suggest the distortion is inaudible but the actual distortion, when listening to music, will be significant.
2) Our brains have different sensitivity to different types and combinations of distortion. This is why your amp will play music very well inspite of modest distortion, and do so much better than another design that has lower, measured distortion.
It is not correct to conclude, as some do, that a "straight wire with gain" would not be the ultimate amplifier. This conclusion is based on an ignorance of distortion sources in "technically" excellent designs.
Eg: Suppose I cannot measure any distortion in an amp but when I listen to it, it sounds imperfect. I can make two conclusions:
1) Perfect amps will not produce the ultimate sound. There is a need in my brain for added distortion to make music sound like music.
2) There is distortion in the amp that I have not measured but that my brain can detect.
Obviously, it is easier for a designer to draw the first conclusion rather than research better measurement methods. But that doesn't make the first conclusion correct.
My advice, ALWAYS trust your ears. Your ears are fantastically excellent at detecting "unnatural" distortions. Way, way better than standard electrical measurements. It is these distortions that suck the life out of the music.
AndrewT said:
Nelson is referring to the conventional view of ClassB where one hands over to the other and takes no further part in passing low current into the load, i.e. not ClassAB.tcpip said:
Note my repeated quote. It is down to the designer/builder to choose the bias of the ClassA amplifier. It is still ClassA in that the devices never turn off AND always play their part in controlling the output current.
BTW,
I did read the link to Nelson's article. I agree with all the comments. It's so nice that Nelson is prepared to keep passing on his experience.
PS,
is this method of active biasing similar to what D.Self is doing with the Cambridge amps?
Your ignorant brain is much smarter than most. It is important to really try to understand what feedback is. Textbooks teach it as if it were a component that one introduces to a circuit to fix the circuit. As if it has its own functionality. As with most things in engineering, the greatest insights are to be had from examining a design at the most basic possible level.sandyK said:
Feedback is a means to cause a circuit to use its own energy to modify its own behaviour. It causes the circuit's output to settle at a value that is different from what it would have been in the absence of the feedback. Feedback systems create equilibrium points where a balance of input signal and feedback signal(s) is achieved. The output of the circuit settles at the equilibrium point. The circuit contains reactive elements that store energy (just like masses and springs in a mechanical system) and these cause the settling to take time. The magnitude and phase of the feedback affects the settling time; in the extreme the output never settles at all.
This settling over time towards an equilibrium output value is true of all feedback systems.
So, as you say, to use feedback as a means to control a circuit into a more accurate output requires the system to be able to settle within a certain time and to a desired accuracy. The circuit has to use its own energy to correct itself, and it cannot correct itself faster than it can react.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.