I just want to emphasize that an academic degree will not cut the "donkey ears" to anyone, just in case...
Another interesting article outlining Lynn Olsons mistake with regard to triodes;
KISS 104 by Andre Jute
The point here is that Triodes employ innate local negative feedback, which as has been discussed, is more desirable than gNFB.
Reasons why the same designer eschews gNFB;
KISS 123 by Andre Jute
As I have said before, most people who employ high amounts of gNFB design high power Williamson type pentode amps that need gNFB to sound acceptable. Very few people out there have ever even heard a zero gNFB PP using class A1 triodes and no coupling caps.
Shoog
KISS 104 by Andre Jute
The point here is that Triodes employ innate local negative feedback, which as has been discussed, is more desirable than gNFB.
Reasons why the same designer eschews gNFB;
KISS 123 by Andre Jute
As I have said before, most people who employ high amounts of gNFB design high power Williamson type pentode amps that need gNFB to sound acceptable. Very few people out there have ever even heard a zero gNFB PP using class A1 triodes and no coupling caps.
Shoog
It isn't intended. I understand it is easy to through things out into the wind. Perhaps some background is in order
I had a pair of tube amplifiers that were given to me. These were Antiques Sound Lab AV-8, the same as numerous friends, and I decided to modify these and add transistor elements. Personally I don't care what I use, the more parts that exist the more creativity and perhaps understanding presents itself. Anyway, in the course of the design I ended up creating an internal high frequency low level oscillator as not affected by the overall feedback loop. It must have had some modest non-linearity in the internal loop gain and as a result it was fairly sinusoidal. My issue was that in removing the oscillation the quality of sound degraded significantly, it lost a natural quality,spatial information and extension.
This is to suggest that my overall feedback loop was accidentally interfered with by an internal high frequency oscillatory system. If the theoretical circuit discussed is now given a frequency and amplitude, which is to say the loop has to pass through an oscillator it would be expected that the mechanism in the loop would be dominated as a function of a shift from the quadrature containing the signal frequencies, more of an FM signal than AM, and whereupon signals other than the oscillation frequency were attenuated.
Bear in mind this is just speculation to try and rationalize my experiences. This is not to suggest that the outcome would in any way change as harmonics, phase or whatever, rather that the nature of the inclusion of an oscillator in the overall feedback path can change the mechanism, perhaps for the better.
I've found that the difference between "marketing document" and "technical treatise" is not always so clear. This "white paper", Beam Power Tubes by O. H. Schade reads very much like a marketing document. Well, what else would you expect? Schade worked for RCA, and, naturally, wanted as many as possible to be using RCA products, and to have good experiences with same. Therefore, the recommendation that this new type he's discussing, the 6L6, have local NFB applied. Though Schade doesn't admit it, the 6L6 (and the 6L6-oids) sound hideous without lNFB. Much worse than types that have much higher nominal THD figures, due to the lower overall THD having higher order frequencies.
Of course, if Schade had come right out with it: "The 6L6 produces high order harmonics and will sound horrible unless you do the following...", he would have been told in no uncertain terms to remove that (may be he was) or find himself on an unemployment line.
Same with Lynn Olson:
Of course it's nonsense, and a bit of prevarication: what's the definition of "large amounts" here? Like RCA, Olson $ells $tuff. What's the biggest difference between a triode and a pentode anyway? With triodes, Vgk and Vpk are pulling Ip in opposite directions: NFB by definition. No getting around that. What Olson is alluding to is the difference between local and global NFB. What are you gonna do? Until a largely atechnological public at large becomes educated, they will continue to believe nonsense like: NFB is the Spawn of Satan.
I've seen some positively hideous designs. One of the worst was a production amp: PP 6V6s working into OPTs with tiny cores. The PS consisted of a 5Y3 with a 100uF capacitor connected across the load, and no bleed down resistor. What did they care? How many end users would realize that having to replace the 5Y3 (and how many even knew what the 5Y3 did?) every few months wasn't normal? I repeat: this was a production amp from the Great Age of HiFi made by one of the big corporations, and not some half fast design found on an amateur forum.
Nothing has changed since 1938: you have to separate wheat from chaff, sense from nonsense, marketing dept BS from technical golden nuggets, and read between the lines to discover what's really being said.
Last edited:
When reading Schade's article I understood it's message like, "People like triodes and hate 6L6, but if to kook it properly, with local feedback, it is very usable tube". Despite it advocated production of his employer, it was a real technical article, with proper calculations, explanations, tables and graphs of measurements.
It came from an RCA mono record player I picked up on "big discard day" held every spring. I got the PTX and the 6V6s and a pretty nice whole range speek from it, but the OPT was borked. No loss: it wasn't any good anyway, having a core that wasn't even an inch high stack. The thingy was obviously made on the cheap.
To consider all the aspects that emerged from the many consideration-worthy posts above, begins to take on the scope of a post-graduate study, particularly for infrequent visitors like myself!
Two worrying things wriggle on the river bed: (1)The many adjectives used, involuntarily leading to dangers of a meta-physical kind (postulate something however honestly believed, and then progressing with that as if it were established science), and (2) glossing over less-emphasized contrary statements made along the way as if they were simply irrelevant to the main flow of narrative.
All respected, only there is the danger of some important aspects shifting ever closer to the end of the table until they drop off the edge unnoticed.
My repeated contribution at this time:
I have not seen any zero-nfb amplifier which does not have audible distortion. (That according to general acceptance based on a wealth of experimental proof of what is audible).
The much lauded 'linearity' of triodes of whatever kind: Not according to my data books or tests. Yes, mainly 2nd harmonic distortion, to be cancelled in p.p. blah blah blah .... Perhaps the expensive analysers used in the prestigious labs where I worked lacked calibration. Shure, some better than others, requirement of proper circuit design etc. - been there, done that.
My main contribution at this time by measurements and tests (if it is possible for such to still carry weight in the haystack of pronouncements by the celebrities ....): The matter of harmonic multiplication so glibly and simplistically offered as condemnation of gnfb at all. Sure - true enough; but figures please! In any properly designed amplifier this does not occur before the application of heavy amounts of gnfb, far above the level required for reduction of all harmonic products to below any audible level. In this (misuse) is included the low thd figure level chasers hovering in the 0,001% regions.
The following are graphs of decrease of harmonic levels up to 11th, for an ordinary if well-designed 80W transistor amplifier operating at about 40W into a resistive load, vs. amount of gnfb. Sure, the reduced decreasing tendency is evident, particularly for the 11th harmonic. But at the design figure of 28dB of nfb, levels are already close to or buried in noise - no need for more gnfb. If there are golden ears still capable of hearing these - my sympathy for your disadvantage. Mercifully most of us fall short of that.
(Apology for a somewhat smudged picture - figures pasted on afterwards.)
Two worrying things wriggle on the river bed: (1)The many adjectives used, involuntarily leading to dangers of a meta-physical kind (postulate something however honestly believed, and then progressing with that as if it were established science), and (2) glossing over less-emphasized contrary statements made along the way as if they were simply irrelevant to the main flow of narrative.
All respected, only there is the danger of some important aspects shifting ever closer to the end of the table until they drop off the edge unnoticed.
My repeated contribution at this time:
I have not seen any zero-nfb amplifier which does not have audible distortion. (That according to general acceptance based on a wealth of experimental proof of what is audible).
The much lauded 'linearity' of triodes of whatever kind: Not according to my data books or tests. Yes, mainly 2nd harmonic distortion, to be cancelled in p.p. blah blah blah .... Perhaps the expensive analysers used in the prestigious labs where I worked lacked calibration. Shure, some better than others, requirement of proper circuit design etc. - been there, done that.
My main contribution at this time by measurements and tests (if it is possible for such to still carry weight in the haystack of pronouncements by the celebrities ....): The matter of harmonic multiplication so glibly and simplistically offered as condemnation of gnfb at all. Sure - true enough; but figures please! In any properly designed amplifier this does not occur before the application of heavy amounts of gnfb, far above the level required for reduction of all harmonic products to below any audible level. In this (misuse) is included the low thd figure level chasers hovering in the 0,001% regions.
The following are graphs of decrease of harmonic levels up to 11th, for an ordinary if well-designed 80W transistor amplifier operating at about 40W into a resistive load, vs. amount of gnfb. Sure, the reduced decreasing tendency is evident, particularly for the 11th harmonic. But at the design figure of 28dB of nfb, levels are already close to or buried in noise - no need for more gnfb. If there are golden ears still capable of hearing these - my sympathy for your disadvantage. Mercifully most of us fall short of that.
(Apology for a somewhat smudged picture - figures pasted on afterwards.)
Attachments
If one looks at 2nd and 3rd harmonics as appearing on a spectrum analyzer it seems common to attribute other presentations as being necessarily clean or true, namely the DC line, the fundamental frequency and the base line. Under conditions whereupon a 3rd harmonic exists it also contains a fundamental frequency as equally variant in ratio and of variant exponent the 3rd is changed to the major fundamental frequency.
The same can be said for the 2nd, however its artifacts attach to the DC line under conditions of a fixed amplitude fundamental. In a DC circuit with single sided non-linearity the DC quiescent point shifts in relation to the variant amplitude of the fundamental frequencies. As you modulate the amplitude, as again happens in applying audio signals, the modulation frequency component comes off the DC line.
In other words as higher amplitudes of non-linearity are observed in the second harmonic spectrum it also predicts more substance in the DC component. This could result in the psycho acoustic effect of having more energy in the lower frequencies or that an envelope exists around the audio signals being modulated. Perhaps such modulations are viewed as accurately part of the reproduction.
What I am suggesting is that the observable spectrum lines of second and third predicts the existence and amplitudes of the hidden artifacts by some fixed ratio, hence making them unnecessary to include, however the consideration of these artifacts may be critical to the psycho acoustic understanding.
The same can be said for the 2nd, however its artifacts attach to the DC line under conditions of a fixed amplitude fundamental. In a DC circuit with single sided non-linearity the DC quiescent point shifts in relation to the variant amplitude of the fundamental frequencies. As you modulate the amplitude, as again happens in applying audio signals, the modulation frequency component comes off the DC line.
In other words as higher amplitudes of non-linearity are observed in the second harmonic spectrum it also predicts more substance in the DC component. This could result in the psycho acoustic effect of having more energy in the lower frequencies or that an envelope exists around the audio signals being modulated. Perhaps such modulations are viewed as accurately part of the reproduction.
What I am suggesting is that the observable spectrum lines of second and third predicts the existence and amplitudes of the hidden artifacts by some fixed ratio, hence making them unnecessary to include, however the consideration of these artifacts may be critical to the psycho acoustic understanding.
I experimented with discovering the nature of the distortion component waveform some time ago, mainly because a pure 3rd harmonic didn't make any sense in the face of a transfer function creating the distortion. Every point of a fundamental frequency input waveform must be translated to some point on the transfer function.
In the experimentation it was found that the distortion component shape was a constant as observable by taking every point on the fundamental frequency and taking this to the cubic power and maintaining the polarity to the fundamental. This shape contains the 3rd harmonic and a fundamental component, both also changing at the cubic rate as the amplitude of the fundamental is varied.
Lets assume I am mistaken. From the opposing perspective how can one rationalize obtaining only a pure 3rd harmonic as the only form of the non-linearity when you impose every point of a fundamental frequency to the amplifiers transfer function. In other words, what mechanism is in play that could cause that result?
It should also be considered that a third harmonic has one and a half cycles in each half of the fundamental, hence must contains both polarities. Ultimately, how can one explain the dual polarities of the third harmonic appearing on each half of a transfer function having a single sided non-linearity, this if only the third harmonic exists? Notwithstanding my delusions the third harmonic cannot exist alone.
In the experimentation it was found that the distortion component shape was a constant as observable by taking every point on the fundamental frequency and taking this to the cubic power and maintaining the polarity to the fundamental. This shape contains the 3rd harmonic and a fundamental component, both also changing at the cubic rate as the amplitude of the fundamental is varied.
Lets assume I am mistaken. From the opposing perspective how can one rationalize obtaining only a pure 3rd harmonic as the only form of the non-linearity when you impose every point of a fundamental frequency to the amplifiers transfer function. In other words, what mechanism is in play that could cause that result?
It should also be considered that a third harmonic has one and a half cycles in each half of the fundamental, hence must contains both polarities. Ultimately, how can one explain the dual polarities of the third harmonic appearing on each half of a transfer function having a single sided non-linearity, this if only the third harmonic exists? Notwithstanding my delusions the third harmonic cannot exist alone.
An in-out function which is just linear and cubic will produce from a single input sine an output spectrum consisting of just fundamental and third harmonic. The level of the third harmonic will vary like input^3. The level of the fundamental will vary like input plus/minus an input^3 component, because the third-order distortion produces both 3rd and fundamental.
Yes, the third cannot exist alone. It can exist with just the fundamental, though. This is precisely what you get at the output of an ideal P-P stage.
Yes, the third cannot exist alone. It can exist with just the fundamental, though. This is precisely what you get at the output of an ideal P-P stage.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.