Member
Joined 2009
Paid Member
You are throwing generalizations without in depth knowledge.
It can be shown for example that an unbalance of 1% of the saturation current ( which is a coefficient and not a real current) generates two order of magnitude larger second order distortion than third order in a differential amplifier if we consider the second and third order intercept, no such thing as perfect balance of all the parameters except in spice.
The same for mismaches in the threshold voltage and load resistors for mos differential amplifiers.
Production of identical devices generates random variations of parameters.
I wonder if you have listen to equipments where the balanced pairs were perfectly mached.
JPV
Unfortunately is not so simple, even with a DSP. Many "niceness" related issues are very probably tied to some distortion with memory, which require more heavy mathematical processing for their rebuild. Are not so (relatively) straightforward as static distortions usually are, that need only to multiply the signal by some chain of polynomial terms. In order to craft an effective "niceness" control that not reduce itself just as a reproposal of somewhat modified electro-musical circuits, need that someone initiate a systematic and extensive study on what distorsive syndromes really are for whatever kind of existing audio amplifiers.
'Til now researchers were been oriented in a totally different way: that is how to minimize distortions everywhere. Rightly, a search of some "optimal" (or at best "euphonic") distorsive syndrome to be breed instead of being suppressed, is, at best of my knowledge, quite out of target of many audio engineers. Instead something of this kind of thoughts are quite usual among RF engineers where, because of technical and physical issues, NFB is unapplicable and problems related to devices nonlinearites should be solved EVEN acting some "choice" of those nonlinearity that are lesser hearmful than others, better reducible by practical filter circuits and so on.
Hi
Piercarlo
PS - In the past a problem similat to that of "choosing the best nonlinearity" apted for audio purpose, was in the design of companding circuits. I don't know how these things stand now, but I think that these, with correlated headaching issues, are now blissfully transferred in digital domain where almost everything can be solved by some efficient, CLEAN alghorithm instead of rely on log-amps, OTA, precision fast rectifiers and assorted delviries!
.
quite ambitious, surely very good transients..
i m somewhat more modest, with a pair of
classical two ways passives, plus a bass
speaker for frequencies less than 55 hz...
I consider the Carver challenge result to be conjecture. Partly because I am not familiar with the experimental conditions of it and mostly because I have my own observations. You seem to be arguing that tube amps sound better because of added distortion. I don't buy that except in a very superficial sense...perhaps a warmness of the sound or mellow bass due to modest damping. Personally, I don't really care for these "flavourings" if you will, but I am very interested in the vivid portrayal of music that some tube amps achieve where most solid state amps fail miserably. I do not believe this is due to added distortion at all; quite the opposite. I think our brains are sophisticated enough to appreciate the baby in the presence of some bathwater. So rather than wash my hands of the whole thing I want to know why tubes do so well that which they do so well, so that I can determine how to achieve it with a solid state design.
He would say that.
I don't have much time for D. Self, myself, as what I have read of his work indicates he is more interested in the witness of his oscilloscope than his or anyone else's ears. His categorisation of distortion types is not comprehensive and the presumptuous name "Blameless Amp" grates. Perfectly good, basic circuit tips, sure.
I couldn't agree more about the source quality variation. Having said that, the situation gets better as the playback system gets better but the best get better still. Some of my recent CDs are pretty good and so are some very old recordings. The ones in between are are variable. I took a CD with a 1955 recording of "Everytime We Say Goodbye" by June Christy to a local hifi dealer and played it through my amps on his best stuff and he could not believe it was a 1955 recording. He was convinced it was a modern recording. Modern recordings can be outstanding too, such as the last Johnny Cash album which converted me to JC - before I heard this I actually quite disliked his music. I suppose I still do, in general, but the sheer intimacy, resolution, space <insert standard audiophile adjectives here> of this recording is captivating.A lot happens before the music ever reaches the power amp. I've found a HUGE difference between different CDs in my collection when played on my YBA source. If I listen critically, there are very few CDs that can even kneel at the feet of my Chesky disks. There seems to be a critical dependency here on recording quality. The 'niceness' control is something I've thought much about and it seems to me that we would first need very good recordings, preferably unmixed tracks so we can apply our own mixing - but this sounds like another hobby altogether
I'm sure that is right. For example, do you think one can pre-distort an image so that projecting the pre-distorted image through a lens out of focus, the final image will be sharp?
Have you elaborate some hypothesis of what mav be the cause of this state of things? In many years I have reasoned on many clues of that it should be, everything containing a bit of truth but nothing holding the entire truth. Surely it's not impossible build solid state amplifiers that sound very closely as tube sound but as much surely the path for achieving this is not written in stone somewhere. My only certainty about this is that a sound "tube sounding" may really obtained only by carefully designed amplifiers, no matter if equipped with tubes or with solid state devices. In other word only state-of-the-art design may lead to appreciate the true character of a kind of amplifier (and its limit of course).
I'm interested in knowing your thoughts about this theme.
Hi
Piercarlo
indeed, a solid state amp must be very "carefully" designed
to produce as much distorsion as a tube amp...
it s the only way to sound "tube" like, doesn it?..
I've read something about many years ago and, indeed, things are not so straightforward!
. I don't remember the exact details but, although the whole thing is possible (if I'm not wrong similar tecniques have been used for years in correcting some early flaws of Hubble Space Telescope), is not a simple task to do (I remember that the whole thing appear to mimic more a sort of "optical feedforward" than a predistorsive path - but pheraps, I'm not sure, are just two side of the same coin).However, back to topic, i think these things as possible but only if well thoughed and not flawed by ingenuity as may be putting on signal path some tubes in order to obtain a "tubish sound" (as done, in fact, in many "tubized" CD players). I think about in the same fashion I'm thinking about cooking: good ingredients are worth but not enough for obtain good dishes. And, if misused, also good ingredients turn to be bad.
I bit confused, i know... ;-)
Hi
Piercarlo
Or at least, for saying the same thing with other words: Tube designers are CONSTRAINED by tube limits itself (mainly but not only poor transconductance) to excel in design problem solving where usually, in the same conditions, solid state designers "bury" many discomforting problems below tons of NFB for the simple reason that is cheap to obtain it (with bipolars is easy obtaining huge voltage gain yet by a single device when gain of same entity require, with tubes, at least a chain of two very sensitive triodes or even pentodes).
My opinion about is simple: FIRST design circuits, not matter which active device are used, AS IF NFB IS NOT AVAILABLE and only at one further stage call for a moderate and judicious use of NFB. A third point: NFB in an amplifier should be use ONLY for flatten out its proper non linearity. Other issues (mainly SVRR and self intermodulation) should be solved by their own count AGAIN AS IF NFB IS NOT AVAILABLE. In other words, NFB must be just considered as a sort of "lottery win" what may aid to ease our livings... but not a substitute of a "regular job" (i.e well done design) regularly renumerated (i.e. with a solidly established "native" good performances).
I understand that this modus operandi lead engineers to sweat more than as they retain right to do but, if the target is the excellence then also our work should be excellent. And if we need to sweat more we shall do! Take more towels from home!
Hi
Piercarlo
Last edited:
What do you mean by "conjecture"?
Here's the full Stereophile article, in case that helps: Stereophile: The Carver Challenge
Member
Joined 2009
Paid Member