Output Transformer symmetry
Dona Greene
ohms on one side of the primary center tap, and 72 ohms on the other
side. Is this disparity enough to cause sonic degradation? The thing
sounds okay, but not as dynamic as other Vibroluxes of the same era
(model AA270, early silverface). I've replaced virtually everything
in the signal path, incl. filter caps, and reverted it back to the
AA964 model except for the transformers.
In a related note, Dan Torres had some articles in Vintage Guitar
about matching components in the splitter section of an amp to get
better, fuller tone. Yeah, I know Dan ain't always the cat's meow
about this stuff. So I'm wondering, if OP transformers are often
unmatched in their primary impedances, what good does it do to match
components in the splitter section? Shouldn't we vary components
(perhaps the 1 Meg resistors in the splitter) to create a perfectly
symmetrical waveform at the speaker?
Any help would be appreciated. If others are curious and don't have
answers, I'll report on my experiments with the Vibrolux. "Look,
Igor, it's alive! Don't pull the plug, switch it to standby!!"
-Brad Bolton
Mark Garvin
>I noticed that my Vibrolux guitar amp's output transformer reads 80
>ohms on one side of the primary center tap, and 72 ohms on the other
>side. Is this disparity enough to cause sonic degradation? The thing
Some texts would imply that balanced tubes and exactly equal signals
are the holy grail of all amps. Not necessarily. Keep in mind
that second harmonic distortion sounds great for guitar. Probably
the best argument for matched tubes in a guitar amp is keeping the
dc currents thru the output equal--to avoid core saturation.
Single ended amps are always working asymmetrically, after all. And
they sound great.
A quick experiment: try pulling one of your output tubes and playing
at low level. If your bias is set hot enough, you'll probably be
surprised that the sound will not be that bad. (Don't do this on
a permanent basis).
>In a related note, Dan Torres had some articles in Vintage Guitar
^^^^^^^^^^ Eeeeek Eeeeek! Mmmeps! Mmmmeps! <g>
>about matching components in the splitter section of an amp to get
>better, fuller tone. Yeah, I know Dan ain't always the cat's meow
I'll try to keep an open mind. Tell us what he said about matching.
>about this stuff. So I'm wondering, if OP transformers are often
>unmatched in their primary impedances, what good does it do to match
>components in the splitter section? Shouldn't we vary components
>(perhaps the 1 Meg resistors in the splitter) to create a perfectly
>symmetrical waveform at the speaker?
The 1 meg resistor is not the place to balance the signal. Best to
trim the plate load resistors. That's why they are unequal in the
first place. And then you'll have to find 12at7's with both sides
balanced, or your signal will be unbalanced again. And as the tubes
age, they generally go back out of balance.
Regards,
Mark Garvin
James Mcshane
measurement made on the OPT is not a measurement of its impedance. The
impedance of the OPT is probably 3000-5000 ohms, and the small dc
difference you measured is inconsequential.
If you apply a small known AC voltage across the primary and measure the
AC current through and voltage across each half you can calculate the
impedance using ohms law. But even that calculation is flawed, since your
results will change with the frequency of the test signal you apply.
Mr. Garvin's tips about balance are good advice. Keep in mind you'll
never get perfect balance, but the closer the better! Incidentally, I
think one of the big advantages of single ended outputs is the lack of
imbalance in the phase splitter circuitry. no phase splitter=no imbalance.
Enjoyed your post!
-
JAMES MCSHANE MLJ...@prodigy.com
Bob Bruhns
length difference caused by one half of the primary being wound nearest
to the core, and the other half being wound on top of that. The diameter
keeps increasing as the windings pile on, so the lengths are unequal and
so are the ohmic resistances.
Especially with class-B outputs, you may encounter a different output
transformer asymmetry problem - the "leakage inductance" of the two
halves of the primary winding may be different enough to make a
difference. When this happens, one half of the waveform is slightly
delayed with respect to the other half of the waveform, and an
unsymmetrical crossover distortion results. This problem is the reason
some output transformers are wound in a more complex fashion, to reduce
inequalities of leakage inductance.
I have an inexpensive modulation transformer which exhibits unequal
winding resistance and leakage inductance. The resistance difference is
probably trivial, but the effect of the unequal leakage inductance is
clearly visible above 4 KHz. To compensate, I added a small amount of
capacitance to ground (about 100 pF) in the output circuit of the phase
splitter driving the output tube on the less delayed side. Not perfect,
but much better. A small inductance in the plate circuit of the less
delayed output tube would probably be better, but much more cumbersome
and harder to adjust.
Check the output with an oscilloscope and see if the two halves of the
waveform stay aligned as you sweep from below 1000 Hz up to 10 or 20,000 Hz.
If not, you probably have unequal leakage inductance in the two halves of
the output transformer primary.
James Mcshane
When you model the equivalent circuit of an OPT the resulting model is
very complex in terms of its AC behavior. Now imagine the goal is to
produce identical circuitry for both halves of a push-pull OPT!
Add to that the mechanical difficulties encountered when assembling and
winding the transformer. Add to that the effect that potting, mounting,
etc. will have. Add to that the unbalance inherent in every phase
splitter known (to me, at least)....
No wonder single-ended designs are gaining in acceptance.
-
JAMES MCSHANE MLJ...@prodigy.com
Don Borowski
: The previous post is the reason top notch output transformers cost $$$.
: very complex in terms of its AC behavior. Now imagine the goal is to
: produce identical circuitry for both halves of a push-pull OPT!
:
: Add to that the mechanical difficulties encountered when assembling and
: winding the transformer. Add to that the effect that potting, mounting,
: etc. will have. Add to that the unbalance inherent in every phase
: splitter known (to me, at least)....
It's really not that hard to do. The easiest way is to split the
transformer bobbin into two halves. One side of the primary is
wound on one half bobbin, the other side is wound on the other half
of the bobbin. The secondary could be split-wound also, but this is
probably not necessary.
The other way of doing it would be to wind the primary with bifilar
wire.
Donald Borowski WA6OMI Hewlett-Packard, Spokane Division
"Angels are able to fly because they take themselves so lightly."
-G.K. Chesterton
ke...@austin.ibm.com
In article <DGK7...@hpcvsnz.cv.hp.com>, boro...@spk.hp.com (Don Borowski) writes:
> James Mcshane (MLJ...@prodigy.com) wrote:
> : The previous post is the reason top notch output transformers cost $$$.
> : When you model the equivalent circuit of an OPT the resulting model is
> : very complex in terms of its AC behavior. Now imagine the goal is to
> : produce identical circuitry for both halves of a push-pull OPT!
> :
> : Add to that the mechanical difficulties encountered when assembling and
> : winding the transformer. Add to that the effect that potting, mounting,
> : etc. will have. Add to that the unbalance inherent in every phase
> : splitter known (to me, at least)....
>
> It's really not that hard to do. The easiest way is to split the
> transformer bobbin into two halves. One side of the primary is
> wound on one half bobbin, the other side is wound on the other half
> of the bobbin. The secondary could be split-wound also, but this is
> probably not necessary.
>
> The other way of doing it would be to wind the primary with bifilar
> wire.
>
read on sectionalizing transformer windings for particular purposes. It turns
out that different applications have different symmetry requirements. Just
split-winding the primary is not enough, unfortunately. You have to split
primary and secondary into several layers which run the full width of the
bobbin and interleave them to get equivalent leakage inductance down. In
class AB and B applications, the leakage from each half primary to the
secondary must be the same, or you get changing frequency and time domain
responses as the "off" tube turns off.
This field is fairly well plowed in the literature.
R.G.
Scott Frankland
: In article <45sa1q$r...@borg.svpal.org>, audi...@svpal.svpal.org (Scott
: Frankland) wrote:
: > Bear in mind that an SE amp is inherently unbalanced and will produce
: > prodigious quantities of even-order distortion unless negative feedback is
: > applied.
: What exactly is "prodigious quantities of even-order distortion"? At what
: magnitude does the amount of distortion become "prodigious"? Also under
: what specific conditions does this phenomenon prevail?
Webster defines "prodigious" as "extraordinary in bulk, quantity, or
degree; exciting amazement or wonder." There is no particular magnitude.
I don't think anyone would deny that the quantity of distortion is
amazing, considering how good many of these amps sound. The specific
conditions are as I indicated: an SE amp without feedback.
: Cracks me up that many of the players in the high end whom 10 to 15 years
: ago...were quite willing to say "...it's not solely in the numbers" that
: differentiated their products in a qualitative sense from the mass
: produced or solid state guys which had diminshingly low distortion
: products compared to any tube amp (save perhaps that old KrohnHite
: amp)...now with the advent of a new (or rather a "new old") approach many
: of these same players are saying...."look, look at those alarming
: distortion stats" on the SE amps. Funny isn't it how we sometimes go full
: circle?
It is not solely in the numbers. I said it then, I say it again.
: My sense is that if you value ultra low distortion then just skip all
: vacuum tube amps and go buy a nice adcom solid state critter. Pretty darn
: low distortion specs last time I looked.
Ultra-low distortion is not the issue here, Mike. The issue is the
specific mix of distortion products that gives a component its sonic
signature; and how to decipher that signature.
: Responsible designs in the SE arena will
: frequently have less than 5% THD...and that will only prevail under max
: power at low freq test conditions. At and up to that first magical watt
: test sample circuits that I have seen measurements for will frequently be
: well under 1% THD.
Then we both agree that at some point distortion figures *do* become
meaningful.
: My advice don't get sucked in by the numbers. Go out and audition for
: yourself whatever amplifier your interested in. The same good advice that
: held twenty years ago when a Dyna ST 120 could put forth better numbers on
: paper than an ARC D76 amp holds true today....maybe the numbers (and a
: rather limited set of numbers at that) won't provide a complete picture of
: the subjective performance capabilities of the amp in question.
My advice as well. Nonetheless, it is in a designer's best interest to
try to find meaning in numbers. I gained a considerable amount of
knowledge by comparing, in the Jan '94 Stereophile, the charts of the
Cary CAD-805 to those of the Krell KSA-300S and YBA 2 HC. Sometimes
numbers do lead to new discoveries. But as you say, Mike, and as Hafler and
Keroes said in 1951, "Excellent measurements are a necessary but not a
sufficient condition for quality of sound. This means that the listening
test is the one of most importance--it is the most stringent test of all."
("An Ultra-Linear Amplifier," Aud Eng, Nov 1951).
: Michael LaFevre
Scott Frankland
Kurt Strain
: Sorry to bend your noodle, Kurt. The key word here is "effect."
No biggie. It's all understood.
Kurt
Scott Frankland
:Is it possible that the presence of lower even harmonics (octaves, etc)
:affords the same 'bridge' to the unavoidable odd-order distortion
:products?
Wonderful observation, Mark. Some work in this vein was done by Ladner,
"The Analysis and Synthesis of Musical Sounds" (Electronic Eng, Oct 1949).
:This IS the root of why odd harmonics are regarded as less pleasant.
Agreed. In fact, it was Pythagorous who ranked dissonance according to
the complexity of the ratio between any given pair of frequencies (tones).
>Many people believe that the push-pull amplifier divides the signal in
>half abruptly at the zero-crossing and that the resulting distortion is
>due to misalignment of the two halves. This is the classic definition of
>crossover distortion and occurs only in Class B amplifiers. Increasing
>the idle current creates overlap between the two halves; eliminating the
>crossover distortion.
:I think I understand what you are saying in principle, but I don't
:agree with the above statement verbatim. Crossover distortion is a
:big factor in class AB amps. Maybe I'm misinterpreting?
I think you may be confusing Class B with Class AB, Mark.
Class AB was the *solution* to crossover distortion. Crossover
distortion is eliminated by the overlap resulting from the
increased idle current. It doesn't take much overlap to eliminate
crossover distortion; just enough to overcome the difference in
cut-in characteristics of the devices. Millman and Halkias call it
a "standby current" (*Electronic Devices and Circuits*, McGraw-Hill,
1967, pp. 564--565). The problem of crossover distortion was resolved
shortly after the appearance of Barton's landmark paper on high-power
amplifiers ("High Audio Power From Relatively Small Tubes," Proc IRE,
Jul 1931). Barton himself suggested the solution in "Application of
the Class B Amplifier to A-C Operated Receivers," Proc IRE, Jul 1932.
:Since power tube curves don't conform to ideal straight lines, there's
:no point at which the output tube suddenly turns class B. (I'll even
:argue whether the crossover point to class A can be precisely defined
:without spec'ing input voltages.)
You're right about Class A; but in a push-pull amplifier the composite
tube curves *are* straight--even in Class B (see Gray, *Applied Electronics*,
2nd ed., Wiley, pp. 460--476; 609--619).
:Probably a semantic side-track though. I'm more interested in the following
:comment:
>The odd-order distortion that remains is due to the
>comb-filter effect of the push-pull action. Even-order products cancel
:The term 'comb-filter' caught my attention, since I haven't heard many
:references to it in this context. Interesting in that it would imply
:that the cause is phase shift between the two output stages. It would
:manifest more at low levels as the two output tubes were passing the
:baton.
See my thread in response to Kurt. Not phase shift between the two sides;
although phase does play a key role in determining how the transformer
segregates even from odd-order distortion products. I believe you are
thinking again of crossover distortion, which *does* increase at low
levels (and is centered about the zero-crossing). The cancellation of
even-order distortion is not a function of signal level, so far as I
know.
>There are only one or two papers, of which I am aware, that explain this
>analytically; if anyone is interested, I'll be happy to list these in a
>future post.
:Please do, Scott. I'd appreciate it.
A number of graphical analyses were published prior to 1933; the
most famous being: B.J. Thompson, "Graphical Determination of
Performance of Push-Pull Audio Amplifiers" (*Proc. IRE*, 21.4,
Apr 1933, pp. 591-600).
Later, an important addendum to Thompson's paper was published:
H.L. Kraus, "Class-A Push-Pull Amplifier Theory" (*Proc. IRE*,
Jan 1948, pp. 50-52). Kraus's paper utilized a combination of
graphical and network analysis.
The definitive analysis was: M.A. Melehay, "Push-Pull Audio
Amplifier Theory" (*Trans. IRE on Audio*, Jul-Aug 1957, pp.
86-89). Melehay's paper is the first to allow for the non-linear
characteristics of output tubes.
Recommended textbooks that cover the subject are:
1. John D. Ryder, *Engineering Electronics* (McGraw-Hill Book
Co., Inc., NY, 1957).
2. Truman S. Gray, *Applied Electronics* (John Wiley and Sons,
Inc., NY, 2nd ed., 1954).
3. Frederick Emmons Terman, *Electronic and Radio Engineering*
(McGraw-Hill Book Co., Inc., NY, 4th ed., 1955).
4. Albert Preisman, *Graphical Constructions for Vacuum Tube
Circuits* (McGraw-Hill Book Co., Inc., NY, 1943).
5. Austin V. Eastman, *Fundamentals of Vacuum Tubes*
(McGraw-Hill Inc., NY, 1937).
:Also bears mentioning in the single-end vs push-pull discussion that many
:single end and class A proponents refer to the system-shock of one side
:'slamming shut' as a major problem. This affects power supplies,
:transformers etc. Single-ended systems are always on, of course, and
:immune to this.
As are Class A push-pull amps; and the McIntosh circuit, which is Class
AB. In Class B and AB amps, this effect is more pronounced with pentodes
than with triodes, as Sah explained in his ominous white paper, "Quasi
Transients in Class B Audio-Frequency Push-Pull Amplifiers" (Proc. IRE,
Nov 1936). The McIntosh amplifier, introduced in 1949, solved,
definitively, the problem of quasi transients--or as Frank McIntosh called
it, "notch distortion" (McIntosh and Gow, "Description and Analysis of a
New 50-Watt Amplifier Circuit," Audio Eng, Dec 1949). This whole question
of notch distortion revolves around three factors: (1) internal output
impedance; (2) cut-in characteristic of the tube type; and (3) leakage
inductance of the OPT. Notch distortion resembles crossover distortion but
differs in that the former predominates at high power levels and the
latter at low power levels.
Scott
Scott Frankland
:"A perfectly balanced push pull amplifier produces purely
:odd-order distortion." In other words, any lack of symmetry in
:the OPT construction will add distortion products that need not
:be present if perfect AC balance is achieved. To me, this
:is undesirable. In a single ended amp there is a highly
:predictable distortion signature produced by the designed in
:imbalance. This distortion, even-ordered and decreasing at
:higher orders, has been (arguably) shown to be more "musical."
:If zero distortion is unobtainable, then I want to create the
:most satisfying spectrum I can.
I won't disagree with you, James. As Joe Roberts observes
(regularly, in *Sound Practices*), everyone has his own
definition of what is "satisfying." Double-blind testing is fine
if what you're looking for is a statistical mean. But we are not
here quantifying the Fermi function. We are talking about highly
individual perceptions by experienced connoisseurs. That is
what, at bottom, we audiophiles are. True, there is a thread of
commonality among our perceptions, but any attempt to level such
experiences can only degrade them. When people of like
sensibility get together to compare sonic notes and jointly
share in a musical experience, I call that group a *circle of
adepts*. This concept is explained at length in my article "Who
We Are" in Positive Feedback Vol. 6, No. 1.
:And I agree with Mike LaFevre that single ended amps get
:unfairly characterized as highly distorted.
To get to the bottom of this trend of "unfair" characterization
we must go back to 1934, when the article "High Quality
Amplification" appeared, by W.T. Cocking (*WW, May 4, 1934). In
this article Cocking lays down the law for the "attainment of a
high standard of fidelity."
Cocking argues for a lower level of distortion "than the level
usually tolerated." Push-pull amplifiers are compared to
single-ended amplifiers and push-pull operation is found to be
necessary in order to reduce the large second-harmonic
distortion produced by single-ended amplifiers (negative
feedback was not yet commonly used to reduce distortion).
Pentodes are then compared to triodes and the following
conclusion is reached: "It will be clear, therefore, that the
ideal output stage for present types of loud speaker [sic] is a
pair of triodes connected in push-pull."
Cocking's articles had the same kind of wide-ranging effect on
the hi-fi world of the 30's that Williamson's were to exert in
the 40's. As late as 1947, with the need for negative feedback
well established, even Williamson continued to operate under
Cocking's sway; considering that Williamson's amplifer used KT66
pentodes converted to triode.
Cocking's influence dominates the thinking of the hi-fi community
up until at least 1955. In 1948, Cocking wrote the
definitive survey of phase-splitters, "Push-Pull Input Circuits"
(*WW*, Jan--May, 1948). In 1955, Cocking wrote what, to the best
of my knowledge, is the *only* analytical treatment of the
ultra-linear circuit (in an editorial for Wireless Engineer, Aug
1955). Even Keroes quoted him in one of his own articles later
that same year ("Adapting the 'Ultra-Linear' Williamson to 6550
Operation," *Radio & TV News*, Nov, 1955).
Beginning in 1951, with the advent of the ultra-linear
amplifier, the old triode vs. pentode controversy was, for a time,
resolved. The resolution of the conflict brought both camps together
as if long lost brothers of the Fraternal Tube Lodge. In the immortal
words of David Hafler and Herbert Keroes: "We have achieved a
new tube type without designing a new tube. This tube is
neither triode nor tetrode, but its improved linearity over
either of those types justifies the designation 'ultra-linear'."
("An Ultra-Linear Amplifier," *Aud Eng*, Nov 1951).
The Ultra-linear amplifier of Hafler and Keroes resolved the
pentode/triode controversy for years to come (by 1955 more than
twenty manufacturers had licensed the design); until
it was to erupt again in the 90's as the driving force behind
the new tube renaissance.
Scott
Michael S. LaFevre
Frankland) wrote:
> James McShane wrote:
>
> :And I agree with Mike LaFevre that single ended amps get
> :unfairly characterized as highly distorted.
>
to which Scott Frankland replied:
> To get to the bottom of this trend of "unfair" characterization
> we must go back to 1934, when the article "High Quality
> Amplification" appeared, by W.T. Cocking (*WW, May 4, 1934). In
> this article Cocking lays down the law for the "attainment of a
> high standard of fidelity."
>
> Cocking argues for a lower level of distortion "than the level
> usually tolerated." Push-pull amplifiers are compared to
> single-ended amplifiers and push-pull operation is found to be
> necessary in order to reduce the large second-harmonic
> distortion produced by single-ended amplifiers (negative
> feedback was not yet commonly used to reduce distortion).
> Pentodes are then compared to triodes and the following
> conclusion is reached: "It will be clear, therefore, that the
> ideal output stage for present types of loud speaker [sic] is a
> pair of triodes connected in push-pull."
>
MSL replies: Notwithstanding the support of Mr. Cocking the continual
focus on harmonic distortion as a criteria of excellence trivializes both
the engineering paradigms extolled as well as our individual subjective
experience of music.
Within the engineering community there have also been competing voices and
thoughts from other well respected engineers. Check out the article which
I sent Joe and he published in one of the earlier Sound Practices from the
founder of the old Audax company (based in NYC). Shit...forget the guy's
name but in a nutshell he argued that the highly touted flatness of freq
resp itself was not a very useful evaluative tool in terms of listener
satisfaction and lifelike reproduction. Wonder what he might have also
said of the harmonic distortion "crutch".
Reliance on simplistic measurements of HD or freq resp does not begin to
capture the essence of what makes one amp more musical or satisfying than
another. Music and it's reproduction is a very complex technical matter
which is trivialized technically by focusing on only these two issues.
Why not expand it to include the PDI (Partridge Distortion Index), the mix
of odd to even distortion, it's distribution, and the magnitude of the
separate products? But this may also be a very meager and impoverished
technical view of a circuit.
Push pull tranneys have their own host of challenges. Read Partridges
seminal work in WW in 1939. The linearity of the primary inductance is
comprimised in most tradtitional push pull output tranneys. Additionally
the push-pull guys at very low levels of excitation operate in the
non-linear region of the BH curve way down at the bottom. SE tranneys
with the DC magnetizing currents (if properly designed) will exhibit a
much enhanced linear inductance plot as well as having the advantage of
being jump started (due to the unbalanced dc present in the primary
winding) well into the linear region of the BH curve if designed
carefully. Check out Partridge's article...he wound up essentially
recommending that push-pull tranneys be built with a deliberate air gap to
linearize the inductance.
But I hate technical arguments because it is impossible (short of writing
books or perhaps volumes of books) to adequately examine and argue the
many technical issues that must be addressed. Everyone always seems able
to find some text or authority to support their thesis...but at the
expense of the scope of the inquiry being myopic.
I very much enjoyed Scott's preceding paragraph (which I deleted for sake
of space and bandwidth) in which he talks about the primacy of individual
aesthetic experience and judgement. Not to be crude but oft times I think
of an analogy with sex. Suppose we made everyone on this group read the
complete studies of Masters and Johnson...do you think this academic work
would begin to describe adequately the subjective experience of
sexuality? Crudely, the phenemonology of sex is such that it is what
happens above the shoulders which defines the experience for us (ignoring
medical maladies or organic dysfunctionalities). All the shop talk in the
world by the authorities Masters and Johnson will strike most of us (as
connoseurs of the sexual experience) as remote at best and downright alien
at worst. For most consumers of hi-fi (whatever that is) bar graphs of
distortion spectra or the Infiniti car commericial bouncing bar graphs of
freq response gives no one I know of a single clue as to how it will
actually sound in real life.
Now if you said I've listened to single ended amps (and more than
one....none of us would judge the solid state camp by the efforts of say
it's worst example) and I find the distortion to be so great that it
absolutely interferes with the music and diminishes the aesthetic
experience...then we should allpay heed to the cries of the complaintant.
But there is much preliminary field data that suggests (note I say
suggests) that SE amps have met with a significant amount of critical
acceptance by the members (or at least some portion of) of the audio
community.
Again...use your own best judgements, use your own ears, forget the
simplistic grossly reductionistic distortion spectra that the pundits pay
homage to and trust your own sense of what is right or satisfying.
And I don't mean to imply here that I am anti-engineering or theory...far
from it. But the one thing I have learned at least in my journeys of
learning about transformer engineering is that it is often not the most
commonly hoisted technical claims that constitutes excellence in its
entirety at all. Rather engineering is a very rich, complex subject
matter well beyond the simplistic analysis of a circuit or device on
isolated technical grounds such as distortion or frequency response.
Mike LaFevre
Mark Garvin
>>Many people believe that the push-pull amplifier divides the signal in
>>half abruptly at the zero-crossing and that the resulting distortion is
>>due to misalignment of the two halves. This is the classic definition of
>>crossover distortion and occurs only in Class B amplifiers. Increasing
>>the idle current creates overlap between the two halves; eliminating the
>>crossover distortion.
>Mark Garvin (mga...@panix.com) wrote:
>:I think I understand what you are saying in principle, but I don't
>:agree with the above statement verbatim. Crossover distortion is a
>:big factor in class AB amps. Maybe I'm misinterpreting?
In <468kga$d...@borg.svpal.org> audi...@svpal.svpal.org (Scott Frankland) writes:
>I think you may be confusing Class B with Class AB, Mark.
>Class AB was the *solution* to crossover distortion. Crossover
>distortion is eliminated by the overlap resulting from the
>increased idle current. It doesn't take much overlap to eliminate
Hi Scott,
Not much chance I'll confuse AB with class B. I was just trying
to be tactful. The onset of crossover distortion is gradual
--hence bias adjust pots. I don't know of any class B audio amps,
misconceptions about MacIntoshes included. In all the class AB amps
I've ever worked with, the tubes cannot be turned off even at the
coldest bias settings. But you can still get huge amounts of xover
distortion.
We may still end up talking semantics here: I was speaking of class B
as push-pull stages operating with nearly zero quiescent current.
In most class AB amps (6L6 or EL34 for instance), crossover distortion
is evident with anything less than 30 ma or so quiescent.
>You're right about Class A; but in a push-pull amplifier the composite
>tube curves *are* straight--even in Class B (see Gray, *Applied Electronics*,
>2nd ed., Wiley, pp. 460--476; 609--619).
Despite the reference to class B above, I've never seen this happen
in real life. Tubes just don't mesh smoothly near cutoff.
>:...I'm more interested in the following comment:
>>ScottF:
>>The odd-order distortion that remains is due to the
>>comb-filter effect of the push-pull action. Even-order products cancel
>:mgarvin:
>:The term 'comb-filter' caught my attention, since I haven't heard many
>:references to it in this context. Interesting in that it would imply
>:that the cause is phase shift between the two output stages. It would
>:manifest more at low levels as the two output tubes were passing the
>:baton.
Note: Usually 'comb filter' implies frequency-dependent cancellations.
Split signal in half, time delay one half, add back together == comb filter.
This will cancel a fundamental frequency and all harmonics.
Since there are no true time delays involved in the circuits we were
discussing, I assumed you were speaking of phase delays.
>See my thread in response to Kurt. Not phase shift between the two sides;
>although phase does play a key role in determining how the transformer
>segregates even from odd-order distortion products. I believe you are
>thinking again of crossover distortion, which *does* increase at low
>levels (and is centered about the zero-crossing). The cancellation of
No, I was not thinking of xover distortion. I was speaking of
signals at low enough level that both output tubes are always on
(ala Class A). If post-phase-splitter signals were subjected to
uneven phase shifts, conceivably some *signal* frequencies could
end up being IN phase, and getting cancelled--as in a comb filter.
[good reading list snipped]
Thanks for taking the time to post the references, Scott.
Regards,
Mark Garvin
Kurt Strain
: focus on harmonic distortion as a criteria of excellence trivializes both
: the engineering paradigms extolled as well as our individual subjective
: experience of music.
[snip]
[snip]
[snip]
[snip]
: And I don't mean to imply here that I am anti-engineering or theory...far
: from it. But the one thing I have learned at least in my journeys of
: learning about transformer engineering is that it is often not the most
: commonly hoisted technical claims that constitutes excellence in its
: entirety at all. Rather engineering is a very rich, complex subject
: matter well beyond the simplistic analysis of a circuit or device on
: isolated technical grounds such as distortion or frequency response.
: Mike LaFevre
[Kurt points to Mike LaFevre]
Kurt: What he said!
Kurt
Scott Frankland
: Notwithstanding the support of Mr. Cocking the continual focus
: on harmonic distortion as a criteria of excellence trivializes
: both the engineering paradigms extolled as well as our
: individual subjective experience of music.
No one is using Cocking as support here. My allusion to
Cocking was meant to show that this is a very old argument;
rooted in the engineer's prejudice toward efficiency. In the
early 30's, the control of distortion was the "final frontier."
Little did they know.
: Everyone always seems able to find some text or authority to
: support their thesis...but at the expense of the scope of the
: inquiry being myopic.
I don't recall having propounded a thesis. If I had, my inquiry
would not have been myopic.
: I very much enjoyed Scott's preceding paragraph (which I
: deleted for sake of space and bandwidth) in which he talks about
: the primacy of individual aesthetic experience and judgement."
Thanks. This paragraph gives my fundamental view of the subject.
Nonetheless, I believe that the spectral distribution of
distortion products, assuming a flat complex f-response, goes
far to explain the differences we hear. We just don't know how
to fully interpret our measurements. Nor do we know quite *how*
to measure in a way that adequately corresponds to *everything*
that may be audible.
Notwithstanding this inadequacy, I believe we should continue to
seek clues that link technology with perception. Such as the one
given recently by Mark Garvin: "Dissonance is more pronounced
when the ear has no 'bridge' to the reference point. Try this:
grab the closest musical instrument and play an F and an E
together. The E should be ALMOST an octave up (it's a major
7th). Sounds dissonant. Now play them again, but plug in a C
(or both A and C) in between. The C tends to bridge the
harmonic series and provide a path for the ear to make sense of
the more complex interval. Is it possible that the presence of
lower even harmonics (octaves, etc) affords the same 'bridge' to
the unavoidable odd-order distortion products?...Could be that
even-order harmonics have a masking effect, or possibly help the
human ear to integrate the odd-order harmonics into a more
musical context."
Clues like this, and those given by researchers such as Ladner
[1] and Shorter [2], allow designers to find repeatable solutions
to common engineering problems; resulting in better, cheaper
audio products.
Scott
[1] "The Analysis and Synthesis of Musical Sounds," *Electronic
Eng*, Oct 1949.
[2] "The Influence of High-Order Products in Non-Linear
Distortion," *Electronic Eng*, Apr 1950.