thanks everyone,
i think i will try this: http://www.plitron.com/PDF/PB/Article/Atcl_4.pdf
any comment?
i think i will try this: http://www.plitron.com/PDF/PB/Article/Atcl_4.pdf
any comment?
Robertc,
Just to confirm certain things and also perhaps explain why there is an often perceived difference in the way different folks like their different amplifiers.
As others have mentioned (and perhaps putting it a bit blunt but respectful), NFB is only bad where too much has been applied in an effort to improve a design that was mediocre in the first place. Excellent circuits exist with as much as 30 dB global NFB, though as much as that is not required for inaudible distortion. (Ergo, apply it correctly!)
Feedbackless amplifiers will have audible distortion, but not all distortion is objectionable! Musically 2nd harmonic is everything an octave higher; never strident, and most music will sound somewhat "warmer" with some. (The ear is quite tolerant of 2nd harmonic products.) Third harmonic products are an octave and a fifth higher, what to me is a rather dull combination but again not strident. Some of this is reported to give an "edge" to strings and brass instruments. This then would explain why some people prefer SET and feedbackless amplifiers - it is their money, their preference and no objection from me. But let us just understand the how and why of it all.
There are also no such things as "linear triodes". They all work to a log characteristic, and the term is a non-scientific one attached to a good sounding amplifier (for the above reasons) because the term would seem to be synonymous with good sound. (Some triodes are of course more linear than others.)
There was a remark about ultra-linear. Such a topology has the advantage of providing almost pentode output and efficiency (sometimes a little higher), but with 80% or more of the distortion, output impedance and load tolerant characteristics of triode operation. That is the reason for its popularity.
Regards.
Just to confirm certain things and also perhaps explain why there is an often perceived difference in the way different folks like their different amplifiers.
As others have mentioned (and perhaps putting it a bit blunt but respectful), NFB is only bad where too much has been applied in an effort to improve a design that was mediocre in the first place. Excellent circuits exist with as much as 30 dB global NFB, though as much as that is not required for inaudible distortion. (Ergo, apply it correctly!)
Feedbackless amplifiers will have audible distortion, but not all distortion is objectionable! Musically 2nd harmonic is everything an octave higher; never strident, and most music will sound somewhat "warmer" with some. (The ear is quite tolerant of 2nd harmonic products.) Third harmonic products are an octave and a fifth higher, what to me is a rather dull combination but again not strident. Some of this is reported to give an "edge" to strings and brass instruments. This then would explain why some people prefer SET and feedbackless amplifiers - it is their money, their preference and no objection from me. But let us just understand the how and why of it all.
There are also no such things as "linear triodes". They all work to a log characteristic, and the term is a non-scientific one attached to a good sounding amplifier (for the above reasons) because the term would seem to be synonymous with good sound. (Some triodes are of course more linear than others.)
There was a remark about ultra-linear. Such a topology has the advantage of providing almost pentode output and efficiency (sometimes a little higher), but with 80% or more of the distortion, output impedance and load tolerant characteristics of triode operation. That is the reason for its popularity.
Regards.
Robertc,
By all means. My remarks were not intended to criticise your choice, just to put matters straight, that are often touted as gospel by some with more emotion than data. (Not referring to this thread.)
Regards and success! Let us know.
By all means. My remarks were not intended to criticise your choice, just to put matters straight, that are often touted as gospel by some with more emotion than data. (Not referring to this thread.)
Regards and success! Let us know.
A good triode or ultralinear push-pull amp should require no extra feedback at all. By good, I mean a straightfoward schematic with high-quality tubes and output transformer with high-quality core and construction. A push-pull amp won't have any even-order (so-called good sounding) harmonics anyway (when measured with a sine wave). Most of the problems with a push-pull amp arise from the collapsing field of the OPT core, so the higher quality the core, the better. Does anyone make a Permalloy push-pull output transformer?
John
Jlsem,
They will certainly sound quite good. But it has been shown by many controlled tests that human hearing is capable of detecting down to 0,1% distortion, and distortion in the best tube designs I have seen over the years go as low as 1% before NFB. But this is at peak output, which is encountered only infrequently. Thus you will be listening to a factor of below 0,2% most of the time.
On the other hand, the "fatiguing factor" of amplifier distortion is more complex than just total harmonic distortion (thd). (This is proved very dramatically by some semiconductor designs with high order products.) This is mercifully not much of a problem with good tube designs, especially triodes, but a UL output stage can have some 5th harmonics. When you say that 2nd harmonics are cancelled py p.p., that is theoretically true. When output tubes go out of balance 1 - 3% can occur. Also so with the popular triode input stage driving e.g. EL34s from a concertina phase splitter; this input stage alone can give some 2% distortion at 38V output.
But after all this, why not NFB? I have not mentioned frequency response, phase shift and damping. I have read of many listeners commenting that NFB makes an amplifier "dull", and as many that lauds it - ergo, it is subjective. When I design, for whose experience (taste) must I design? When such an instrument sounds dull, have tests been done to check why? In Stereophile tests one is often able to correllate the measurements (graphs) with the commentary. I am still respectfully asking for sustainable evidence that well applied NFB has a negative audible effect. In a half century I have yet to see that.
I have not seen output transformers with a Permalloy core, but my commercial experience is not that much. I use C-cores (M5), which can yield down to 0,6% distortion at maximum output, 50 Hz.
Regards.
They will certainly sound quite good. But it has been shown by many controlled tests that human hearing is capable of detecting down to 0,1% distortion, and distortion in the best tube designs I have seen over the years go as low as 1% before NFB. But this is at peak output, which is encountered only infrequently. Thus you will be listening to a factor of below 0,2% most of the time.
On the other hand, the "fatiguing factor" of amplifier distortion is more complex than just total harmonic distortion (thd). (This is proved very dramatically by some semiconductor designs with high order products.) This is mercifully not much of a problem with good tube designs, especially triodes, but a UL output stage can have some 5th harmonics. When you say that 2nd harmonics are cancelled py p.p., that is theoretically true. When output tubes go out of balance 1 - 3% can occur. Also so with the popular triode input stage driving e.g. EL34s from a concertina phase splitter; this input stage alone can give some 2% distortion at 38V output.
But after all this, why not NFB? I have not mentioned frequency response, phase shift and damping. I have read of many listeners commenting that NFB makes an amplifier "dull", and as many that lauds it - ergo, it is subjective. When I design, for whose experience (taste) must I design? When such an instrument sounds dull, have tests been done to check why? In Stereophile tests one is often able to correllate the measurements (graphs) with the commentary. I am still respectfully asking for sustainable evidence that well applied NFB has a negative audible effect. In a half century I have yet to see that.
I have not seen output transformers with a Permalloy core, but my commercial experience is not that much. I use C-cores (M5), which can yield down to 0,6% distortion at maximum output, 50 Hz.
Regards.
robertc said:
Robert;
In general it is not true that an amplifier without NFB sounds better.
NFB makes the amp better by substracting part of output signal from input one, so distortions are partially cancelled. The resulting gain is less, but it is more stable and depends less on used tubes and their age.
Hi Johan;
our perception is tricky; it does not hear significant distortions when they similar to "natural" distortions of our perception system (ears+brain), but hears well "unnatural" part of distortions. Common method of measurements is not natural so it does not show how clean is amplifier for human ears. The more of "natural" distortions we hear, the louder seems to be the sound subjectively. Good amp with "natural" distortions sounds subjectively quieter, more relaxed, less boring with NFB that minimizes distortions leaving their spectrum natural. If NFB is connected to the wrong point altering character of distortions the result will be not as expected.
our perception is tricky; it does not hear significant distortions when they similar to "natural" distortions of our perception system (ears+brain), but hears well "unnatural" part of distortions. Common method of measurements is not natural so it does not show how clean is amplifier for human ears. The more of "natural" distortions we hear, the louder seems to be the sound subjectively. Good amp with "natural" distortions sounds subjectively quieter, more relaxed, less boring with NFB that minimizes distortions leaving their spectrum natural. If NFB is connected to the wrong point altering character of distortions the result will be not as expected.
Johan Potgieter said:
I have a commercial Ultralinear EL 34 amp that is purposely unbalanced. That is, the bias (fixed) is purposely set up about 20% different one side to the other. When I measured the distortion with RMAA, it shows a descending stairstep of second, third, and fourth, with almost no higher order components. I think the second was around 70dB down. I played with the bias settings, and the bias recommendations of the manufacturer gave the lowest overall distortion and also the best profile (HD decreasing with order). Frequency response is dead flat out from 20Hz to 20kHz.
I got these amps before I got into making stuff and once my system is all DIY, I plan to sell them so I haven't taken them apart to explore. But I wonder if maybe the OPT is gapped a bit, and if this doesn't avoid zero crossing distortion in the OPT somewhat. They were the best sounding of all the amps I listened to at the time I picked them up - very clean and strong top to bottom, with very good imaging. I think this is an example of "tuning" the circuit to achieve a specific end. I haven't seen much discussion of this technique. Most PP amps seem to stress perfect balance, but maybe?
Sheldon
If an amp performs well enough in all of those regards, then WHY use NFB? If someone (like myself) likes the way the amp sounds without it, then he ought to have it without a bunch of scope jockeys with their signal generator-source measurements telling him they have proof he's wrong.
John
jlsem said:
Thank you! I agree completely. The poster asked for schematics and got bombarded with posts telling him he shouldn't want what he asked for.
In any case, I haven't built Tossie's PP 2A3, but a few others (with ears that I respect) have built it and report that it sounds very good. It offers optional NFB so you can try it both ways and decide for yourself.
Tossie's PP 2A3
-- Dave
jlsem and DaveCigna,
This is the kind of response that saddens me. I will ignore the almost derogatory and biligerent tone; let us just look at the accusations:
Dave,
Those last three words look to me like a question - English is only my second language, but I suspect that I am correct. Am I (and others) not allowed to reply?
Then, can you quote the posts where anyone told Robertc that "he shouldn't want what he asked for"? I am not going to clutter up the screen with posted quotes (I hope we are not in court), but the only relevant portions I found is in Tubelab's post #9, where he said that if Robert don't like what it does he can leave it.
In my post #22 I stated that I have NO objection to folks' choice, and in #24 I said I am NOT criticising ....
So excuse me and others for simply stating relevant basic electronic principles in reply to a question. And if either of you disagree with those principles, don't criticise us; we are only the messengers. With respect, go and consult your textbooks and disprove science - or at least motivate your choices. But let us just agree that "I have tied a knot in my speaker cable and it sounded better" does not constitute proof. It has been shown in this very thread that for every person's subjective experience, someone else experienced the opposite. To repeat myself (with apology), for whose taste must I now design my amplifiers?
I do not know in what different terms to state that I respect people's choices and that there is no intrinsic right or wrong there - just a few posts ago I granted you that, Jlsm, regarding your feedbackless amplifier; I pointed out why it also will sound good. But let us not confuse that with electronic principles; our subjective tastes do not give us the right to brush those aside as if engineers are fools, just because. I think (hope!) nobody will disagree that a sorry state of affairs will indeed have arrived when what one says and how one says it amounts to: I have made up my mind; don't confuse me with facts.
And after all that, I (and Robertc) still had no reply as to why NFB should not be used. If I need to repeat it's advantage: It will make all good amplifiers better in several aspects.
Unless we can have constructive and respectful dialogue on this, I am bowing out of this particular line of argument.
Regards.
This is the kind of response that saddens me. I will ignore the almost derogatory and biligerent tone; let us just look at the accusations:
robertc said:
Dave,
Those last three words look to me like a question - English is only my second language, but I suspect that I am correct. Am I (and others) not allowed to reply?
Then, can you quote the posts where anyone told Robertc that "he shouldn't want what he asked for"? I am not going to clutter up the screen with posted quotes (I hope we are not in court), but the only relevant portions I found is in Tubelab's post #9, where he said that if Robert don't like what it does he can leave it.
In my post #22 I stated that I have NO objection to folks' choice, and in #24 I said I am NOT criticising ....
So excuse me and others for simply stating relevant basic electronic principles in reply to a question. And if either of you disagree with those principles, don't criticise us; we are only the messengers. With respect, go and consult your textbooks and disprove science - or at least motivate your choices. But let us just agree that "I have tied a knot in my speaker cable and it sounded better" does not constitute proof. It has been shown in this very thread that for every person's subjective experience, someone else experienced the opposite. To repeat myself (with apology), for whose taste must I now design my amplifiers?
I do not know in what different terms to state that I respect people's choices and that there is no intrinsic right or wrong there - just a few posts ago I granted you that, Jlsm, regarding your feedbackless amplifier; I pointed out why it also will sound good. But let us not confuse that with electronic principles; our subjective tastes do not give us the right to brush those aside as if engineers are fools, just because. I think (hope!) nobody will disagree that a sorry state of affairs will indeed have arrived when what one says and how one says it amounts to: I have made up my mind; don't confuse me with facts.
And after all that, I (and Robertc) still had no reply as to why NFB should not be used. If I need to repeat it's advantage: It will make all good amplifiers better in several aspects.
Unless we can have constructive and respectful dialogue on this, I am bowing out of this particular line of argument.
Regards.
Wavebourn said:
Wavebourn,
A very interesting point - but firstly we will have to clarify terms. I am not quite with you unless you mean by distortions, harmonics, etc.?
Unfortunately I also sense that we might go a little off thread here - Robertc just asked about an amplifer here and I do not want to hijack his thread. (As you saw I am already flacked!)Hopefully a moderator is reading here and can advise on whether this should become a separate thread, if not already discussed in the past. (I have joined only 16 months ago.)
For now, I will just briefly say that in my terms any amplifier generated distortion (by-products) of whatever nature is something added to the original, and thus by definition should be avoided in a quest towards a "perfect" amplifier.
But perhaps we could continue this here or somewhere else hopefully for the benefit of others too. I am waiting for guidance.
Regards.
At a dinner table with a group of intelligent and passionate people, the conversation will tend to jump around a bit. Don't worry about it.
By the way... the amp shown in post #31 is using NFB at the cathodes of the output stage tubes...
🙄
🙄
Hi Johan! I don't 100% agree with what you've written but it's good to see you back in the swing.
Johan Potgieter said:
No need to hijack his thread, we continue discussion about schematics of amplifiers without feedback VS with feedback to better understand what the OP want. Suppose, the question was of a skeleton of a bold man; we discuss what is the difference between bold man and curly man, and how it is reflected on skeleton, and is it really reflected? Take any amp, remove feedback loop, and you have a feedback-less amp. Or not?
My point was, that human perception of quality of sound is subjective, and the character of distortions that the amp has without feedbeck and with one may be different.
SY,
No-no, was not looking for protection! (You should have heard us at the CSIR lunch table!) There should have been three spaces before that remark instead of none.
I only wondered at what stage one is getting a little wide off-thread, regarding the (very) interesting point that Wavebourn raised; also whether or not this aspect was perhaps discussed to death before I joined. (My mother taught me to obey rules. I taught myself not to be caught out when I don't.) I take the post by Wavebourn as guidence and will try to put my side succinctly.
Regards.
No-no, was not looking for protection! (You should have heard us at the CSIR lunch table!) There should have been three spaces before that remark instead of none.
I only wondered at what stage one is getting a little wide off-thread, regarding the (very) interesting point that Wavebourn raised; also whether or not this aspect was perhaps discussed to death before I joined. (My mother taught me to obey rules. I taught myself not to be caught out when I don't.) I take the post by Wavebourn as guidence and will try to put my side succinctly.
Regards.
- Status
- Not open for further replies.