[tubegarden]

Hi RATs!

John Broskie started this last year. I finally found myself listening
to an amp with one shared cathode resistor per pair of push-pull output
tubes, the Heathkit A7.

http://www.tubecad.com/2005/May/blog0046.htm

Broskie is using Blumlein's design to enter his own world of bias
design circuits, which, unfortunately, not enough of my brain can pull
together to even try to understand, much less covet

I remembered after a few weeks there was something I wanted to try with
a shared cathode resistor circuit. Only took a few months to remember
what ... sigh.

But, the basic idea, of a balanced P-P circuit that does absolutely
nothing dynamically much better than any other, really captures my
imagination. Even if it sounds like Eyesore's 'Old Testament
Pronouncements', it at least 'rests' closer to Truth than meer mortal
circuits The next step is to get the AC operation to follow, perhaps
a bit trickier.

I intend to try this in my real world amps. I will replace the single
470 Ohm resistor and cap with two parallel pairs of two series 1K
resistors. Each series pair will be bypassed by a cap.

The grid resistor of each output tube will be tied to the central
junction of the two series resistors in the cathode circuit of the
other output tube. Cross-pollinization, of sorts

The basic idea is to create an environment where the output transformer
has balanced current through the two sides of the push-pull pair of
windings. I have no clue if real transformers, especially mine, really
'want, or need' balance. We shall hear

As per my usual strict Scientificult Principia, if neither amp sounds
really wretched after the mod, I will declare complete success!

Many sure fire enhancements may follow

Happy Ears!
Al

[Ian Iveson]

Bret Ludwig wrote

The basic idea is to create an environment where the output
transformer
has balanced current through the two sides of the push-pull pair of
windings. I have no clue if real transformers, especially mine,
really
'want, or need' balance. We shall hear

They very certainly do. Unless you have a fixed resistance in each
cathode leg to act as a shunt to measure against in addition to the
cathode resistance providing bias you cannot measure this.


Near enough, but no nearer. Arguably, the point of zero magnetic field
should not occur at zero signal.

Awaiting Al's listening report with interest...

cheers, Ian

[Chris Hornbeck]

On Sat, 25 Nov 2006 02:50:34 GMT, "Ian Iveson"
wrote:

Near enough, but no nearer. Arguably, the point of zero magnetic field
should not occur at zero signal.

This is the rabbit hole that leads to the the White Queen's
argument for single-ended output stages.

It worked for me, at least until she said "Off with her head".

Awaiting Al's listening report with interest...

Natch. Much thanks, as always,

Chris Hornbeck
"What's the line? 'When the student is ready, the teacher appears.'
Maybe we should try to figure out how to be "ready" more often.
-Ty Ford

[Ian Iveson]

Chris Hornbeck wrote

Near enough, but no nearer. Arguably, the point of zero magnetic
field
should not occur at zero signal.

This is the rabbit hole that leads to the the White Queen's
argument for single-ended output stages.

It worked for me, at least until she said "Off with her head".

Remember what the white rabbit said.

Al is dangerously close to ambush by the notion of precision. He'll be
measuring something next.

I've been wondering about symmetry. Fred's ideas on assymetric output
stages, Patrick's on gapping PP OPTs, and Cambridge Audio's "class XD"
recently expounded by Douglas Self. Optimise for class B, then offset
bias to give SE class A for the first few watts. Daft idea for valves,
but an idea all the same.

Perhaps the process of painstakingly resetting the bias current for 8
output valves makes me grumpy, but the amps never seem to be at their
best afterwards.

cheers, Ian

[Chris Hornbeck]

On Sat, 25 Nov 2006 04:51:01 GMT, "Ian Iveson"
wrote:

It worked for me, at least until she said "Off with her head".

Remember what the white rabbit said.

Sadly, I've gotten too old. But I remember. Some changes
are permanent.


Al is dangerously close to ambush by the notion of precision. He'll be
measuring something next.

Heresy!


I've been wondering about symmetry. Fred's ideas on assymetric output
stages, Patrick's on gapping PP OPTs, and Cambridge Audio's "class XD"
recently expounded by Douglas Self. Optimise for class B, then offset
bias to give SE class A for the first few watts. Daft idea for valves,
but an idea all the same.

Perhaps the process of painstakingly resetting the bias current for 8
output valves makes me grumpy, but the amps never seem to be at their
best afterwards.

I'd suggest that the issue isn't really one of symmetry, but one
of monotonicity. It's perhaps the Numero Uno overlooked problem
in modern electronics.

A transformer without DC current has its BH flat spot around
the signal's zero crossing. Dancing this flat spot somewhere
else has its charms.

Much thanks, as always,

Chris Hornbeck
"What's the line? 'When the student is ready, the teacher appears.'
Maybe we should try to figure out how to be "ready" more often.
-Ty Ford

[Patrick Turner]

tubegarden wrote:

Hi RATs!

John Broskie started this last year. I finally found myself listening
to an amp with one shared cathode resistor per pair of push-pull output
tubes, the Heathkit A7.

http://www.tubecad.com/2005/May/blog0046.htm

Broskie is using Blumlein's design to enter his own world of bias
design circuits, which, unfortunately, not enough of my brain can pull
together to even try to understand, much less covet

I remembered after a few weeks there was something I wanted to try with
a shared cathode resistor circuit. Only took a few months to remember
what ... sigh.

But, the basic idea, of a balanced P-P circuit that does absolutely
nothing dynamically much better than any other, really captures my
imagination. Even if it sounds like Eyesore's 'Old Testament
Pronouncements', it at least 'rests' closer to Truth than meer mortal
circuits The next step is to get the AC operation to follow, perhaps
a bit trickier.

I intend to try this in my real world amps. I will replace the single
470 Ohm resistor and cap with two parallel pairs of two series 1K
resistors. Each series pair will be bypassed by a cap.

The grid resistor of each output tube will be tied to the central
junction of the two series resistors in the cathode circuit of the
other output tube. Cross-pollinization, of sorts

The basic idea is to create an environment where the output transformer
has balanced current through the two sides of the push-pull pair of
windings. I have no clue if real transformers, especially mine, really
'want, or need' balance. We shall hear

As per my usual strict Scientificult Principia, if neither amp sounds
really wretched after the mod, I will declare complete success!

Many sure fire enhancements may follow

Happy Ears!
Al


Blummy offers us a way of making sure the bias tends to stay balanced.

In his Garter circuit if one tube decides to pull more current,
its rise in cathode voltage is 1/2 transfered to the other tube
so its bias increases, but then it passes 1/2 its rise in bias to the
one that
decided to conduct more.

Sorry folks, but I can't endorse Blumlein's idea here.

Give me plain old dual cathode R, each bypassed with about 1,000 uF and
you have the simplest and best sounding cathode topology.

Some say the common R does wonders to help the signal 2H of each
tube balance, and while this is true, if one must use one darn R
for the 2 cathodes, its better to have two,
and have two C also, but instead of grounding each 1,000 uF connect
their
-Ve ends to a 22k and ground that.
Thus you have separate dc biasing for each tube, and you have
a single ac R for the two tubes.
The single R is of benefit only while the circuit
works in class A, and it all falls apart for AB, unless you have fixed
bias.
Servo biasing to change the grid bias for AB is all BS.
Grid bias must me kept constant for AB and B circuit working.

Various methods have been devised for bias balancing while fiddling with
the grid voltage
but none work well and the only one that does work
to be equal to fixed bias for AB or B is at my website, see
the pages onj my 300 watt amps with the unique circuit found nowhere
else on the net.
Its very applicable to two tube output circuits.

Patrick Turner.

[tubegarden]

Patrick Turner wrote:
tubegarden wrote:

Hi RATs!

John Broskie started this last year. I finally found myself listening
to an amp with one shared cathode resistor per pair of push-pull output
tubes, the Heathkit A7.

http://www.tubecad.com/2005/May/blog0046.htm

Broskie is using Blumlein's design to enter his own world of bias
design circuits, which, unfortunately, not enough of my brain can pull
together to even try to understand, much less covet

I remembered after a few weeks there was something I wanted to try with
a shared cathode resistor circuit. Only took a few months to remember
what ... sigh.

But, the basic idea, of a balanced P-P circuit that does absolutely
nothing dynamically much better than any other, really captures my
imagination. Even if it sounds like Eyesore's 'Old Testament
Pronouncements', it at least 'rests' closer to Truth than meer mortal
circuits The next step is to get the AC operation to follow, perhaps
a bit trickier.

I intend to try this in my real world amps. I will replace the single
470 Ohm resistor and cap with two parallel pairs of two series 1K
resistors. Each series pair will be bypassed by a cap.

The grid resistor of each output tube will be tied to the central
junction of the two series resistors in the cathode circuit of the
other output tube. Cross-pollinization, of sorts

The basic idea is to create an environment where the output transformer
has balanced current through the two sides of the push-pull pair of
windings. I have no clue if real transformers, especially mine, really
'want, or need' balance. We shall hear

As per my usual strict Scientificult Principia, if neither amp sounds
really wretched after the mod, I will declare complete success!

Many sure fire enhancements may follow

Happy Ears!
Al


Blummy offers us a way of making sure the bias tends to stay balanced.

In his Garter circuit if one tube decides to pull more current,
its rise in cathode voltage is 1/2 transfered to the other tube
so its bias increases, but then it passes 1/2 its rise in bias to the
one that
decided to conduct more.

Sorry folks, but I can't endorse Blumlein's idea here.

Give me plain old dual cathode R, each bypassed with about 1,000 uF and
you have the simplest and best sounding cathode topology.

Some say the common R does wonders to help the signal 2H of each
tube balance, and while this is true, if one must use one darn R
for the 2 cathodes, its better to have two,
and have two C also, but instead of grounding each 1,000 uF connect
their
-Ve ends to a 22k and ground that.
Thus you have separate dc biasing for each tube, and you have
a single ac R for the two tubes.
The single R is of benefit only while the circuit
works in class A, and it all falls apart for AB, unless you have fixed
bias.
Servo biasing to change the grid bias for AB is all BS.
Grid bias must me kept constant for AB and B circuit working.

Various methods have been devised for bias balancing while fiddling with
the grid voltage
but none work well and the only one that does work
to be equal to fixed bias for AB or B is at my website, see
the pages onj my 300 watt amps with the unique circuit found nowhere
else on the net.
Its very applicable to two tube output circuits.

Patrick Turner.

Hi RATs!

Into the future, slowly. I got motivated and went thru my large Watt
Resistor stash. Sigh. Only two 1K 5W were left. So, I put them in one
amp. Each with its own new cap. I did find lots of 2K 5W, but, cooler
heads prevailed, and I will wait to order some on Monday.

Meanhile, running the two amps, one with 470 ohms and one cap and one
with parallel 1K/cap sounds OK.

It is just a hobby. The Music is always nice

Happy Ears!
Al

[Ian Iveson]

Chris wrote:

A transformer without DC current has its BH flat spot around
the signal's zero crossing. Dancing this flat spot somewhere
else has its charms.

Magnetism is still a mystery to me.

What about hysteresis?

The flat spots are generally at the extremes of magnetisation, not in
the middle, surely? What is more, whatever the magnetic bias,
hysteresis occurs around the bias point.

Points of zero current will always be on the curvy parts of the loop.
Where points of zero voltage are, I wonder.

The width of a hysteresis loop is least, and its linearity best,
around a bias point on the steepest part of the DC curve.

cheers, Ian

[tubegarden]

Ian Iveson wrote:
Chris wrote:

A transformer without DC current has its BH flat spot around
the signal's zero crossing. Dancing this flat spot somewhere
else has its charms.

Magnetism is still a mystery to me.

What about hysteresis?

The flat spots are generally at the extremes of magnetisation, not in
the middle, surely? What is more, whatever the magnetic bias,
hysteresis occurs around the bias point.

Points of zero current will always be on the curvy parts of the loop.
Where points of zero voltage are, I wonder.

The width of a hysteresis loop is least, and its linearity best,
around a bias point on the steepest part of the DC curve.

cheers, Ian


Hi RATs!

The idea of the garters is to balance the transformer while idle.

Then both directions can contribute equally to the slaughter of the
signal

Happy Ears!
Al

[Ian Iveson]

Al wrote:

The idea of the garters is to balance the transformer while idle.

It's easy to be balanced when you're idle. But DC isn't music.

Then both directions can contribute equally to the slaughter of the
signal

DC and AC balance are rather different things, as are current and
magnetic balances.

Anyway, have you done it yet? How does it sound?

cheers, Ian
"tubegarden" wrote in message
oups.com...
Ian Iveson wrote:
Chris wrote:

A transformer without DC current has its BH flat spot around
the signal's zero crossing. Dancing this flat spot somewhere
else has its charms.

Magnetism is still a mystery to me.

What about hysteresis?

The flat spots are generally at the extremes of magnetisation, not
in
the middle, surely? What is more, whatever the magnetic bias,
hysteresis occurs around the bias point.

Points of zero current will always be on the curvy parts of the
loop.
Where points of zero voltage are, I wonder.

The width of a hysteresis loop is least, and its linearity best,
around a bias point on the steepest part of the DC curve.

cheers, Ian


Hi RATs!


Happy Ears!
Al

[Chris Hornbeck]

On Mon, 27 Nov 2006 17:18:09 GMT, "Ian Iveson"
wrote:

A transformer without DC current has its BH flat spot around
the signal's zero crossing. Dancing this flat spot somewhere
else has its charms.

Magnetism is still a mystery to me.

What about hysteresis?

The flat spots are generally at the extremes of magnetisation, not in
the middle, surely? What is more, whatever the magnetic bias,
hysteresis occurs around the bias point.

Indeed! So at small signal the B/H flat spot is everywhere.

Coaxing and persuading the dynamic flat-spot away from
signal zero-crossing is (perhaps) part of the otherwise
antithetical magic of the single-ended chism.

Monotonicity is the bete noir of modern electronics.

Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

[Patrick Turner]

Chris Hornbeck wrote:

On Mon, 27 Nov 2006 17:18:09 GMT, "Ian Iveson"
wrote:

A transformer without DC current has its BH flat spot around
the signal's zero crossing. Dancing this flat spot somewhere
else has its charms.

Magnetism is still a mystery to me.

What about hysteresis?

The flat spots are generally at the extremes of magnetisation, not in
the middle, surely? What is more, whatever the magnetic bias,
hysteresis occurs around the bias point.

Indeed! So at small signal the B/H flat spot is everywhere.

Coaxing and persuading the dynamic flat-spot away from
signal zero-crossing is (perhaps) part of the otherwise
antithetical magic of the single-ended chism.

Monotonicity is the bete noir of modern electronics.

Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

In my tests of ac behaviour of chokes and iron wound windings,
one finds that the measured L reduces as the applied voltage is reduced.

The L is measured by measuring the impedance or inductive reactance of
the L in question,
and L = Z / ( 6.28 x F ).

A coil may have 100H at a high field strength of say 1 Tesla, which
could be the case
with an OPT with say 500Vrms across it at say 50Hz.
But at 0.01Vrms across the same coil, the L may measure only 1H, or some
very low value.
If the applied signal to the coil is a current source then the wave form
will show severe 3H distortion as the voltage is increased because the
reactance of the coil rises dynamically with voltage rise during each
wave,
and shows that the inductance changes as the voltage or current changes
which
simply means that the iron presence gives the coil a non linear
character.
There is a lot about ironj caused distortion in RDH4 and how to
calculate it.
To avoid it requires that we keep the source impedance low so the use of
pentodes driving OPTs isn't good practice without a shirtload of NFB.

Readers wanting to build better amplifiers should ignore the drivel
posted by
Ian Iveson.

Regardless of the BH curves, its important to understand the
simple manifestation of the non-linearities of the iron wound
componentsand 1/2 an hour with a variety of resistors, voltmeter, old
tranny
and a signal gene will reveal what is important about iron wound items.

The use of GOSS core materials greatly reduces the THD generated by
audio
transformers so that it becomes smaller than that generated by the tubes
or other active components driving the tranny, and thus it becomes
negligible.
But with an IST, its not unusual for 0.1% thd to exist across all F
at normal listening levels, and its very like crossover distortion in
underbiased
class Ab circuits, ie, the 3H makes the original wave form more pointy
on the wave crests rather than tend to flatten the wave crests which
occurs
when the 3H waves are 180 degrees different.

But why should I explain it all?
I suggest those who dunno what happens with iron should go away
and observe what happens in practical experiments.
One may need a low distortion oscillator and distortion tester to see
the small quantities of distortion generated.

Patrick Turner.

[Ian Iveson]

Naughty Patrick blethered:

Readers wanting to build better amplifiers should ignore the drivel
posted by
Ian Iveson.

Oi!

And why didn't you ignore it, anyway?

The trouble with being so bitter and twisted is that you are inclined
to miss the point. In this case you have missed it by process of
averaging.

At what *point*, with respect to a sine input voltage, is the
inductance least? Where, exactly, is the BH curve flattest?

What is the effect of a little DC bias on the primary of a PP OPT, on
the distortion of a smallish signal?

RDH4 has already been written and can be acquired easily. I see no
point in repeating it.

Ian

[Patrick Turner]

Ian Iveson wrote:

Naughty Patrick blethered:

Readers wanting to build better amplifiers should ignore the drivel
posted by
Ian Iveson.

Oi!

And why didn't you ignore it, anyway?

The trouble with being so bitter and twisted is that you are inclined
to miss the point. In this case you have missed it by process of
averaging.

At what *point*, with respect to a sine input voltage, is the
inductance least? Where, exactly, is the BH curve flattest?

What is the effect of a little DC bias on the primary of a PP OPT, on
the distortion of a smallish signal?

RDH4 has already been written and can be acquired easily. I see no
point in repeating it.

Ian

I will not answer your questions on this subject because
you prefer that I ignore you, so you can get way with drivel
without being challenged,
and I am sure you have the means to discover for yourself what the
answers are.
And unless you do some really careful technical testing, nobody could
take much
meaning from any results you obtain.

Patrick Turner.

[Chris Hornbeck]

On Sun, 03 Dec 2006 09:29:30 GMT, Patrick Turner
wrote:

In my tests of ac behaviour of chokes and iron wound windings,
one finds that the measured L reduces as the applied voltage is reduced.

lots of other good stuff about large-signal behavior of
iron-cored inductors snipped for bandwidth

There's a separate issue for iron-cored inductors not
oft' enough discussed (although well-understood and
referenced in the earliest Frank McIntosh papers,
before I was born, for instance).

The issue is one of a flat spot in the B/H curve across
zero-crossing. It's analogous to what's called "backlash"
in gear trains; the driving gear, if idled, must first
take up all the gear slack in the drive train before
power can be transmitted.

Note that this is an issue separate from and unrelated to
hysteresis, which is a dynamic problem occuring at all
signal levels. Hysteresis is more analogous to the
overshoot of a mass/compliance pair.

Threshold effects aren't necessarily fatal (good transformers
exist for microphone-level signals, ferinstance), but, in
general, issues of monotonicity are the most overlooked ones
"in these days of modern times". (.0001% distortion implies
perfection; pay no attention to that man behind the curtain).

Monotonicity issues are well understood in the
world of analog-to-digital conversion (and D/A) but
are (lately) mostly ignored in the purely analog world.
I believe this to be a preventable misperception.

Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

[tubegarden]

Hi RATs!

I lost the thread when the A7 in play did not respond to my attempts at
garter implementation ...

I only have so much juice ... so, I took it out of the system and
plugged in a long silent SE amp.

I cannot swing my perceptual leg over the monotonicity horse. I grew up
in Minneapolis. Talk about Monotony City ... they just built a new live
performance theatre, call it the Guthrie, just like the old theatre.
Maybe they will call the next one the Gu3 ... Don't get excited, that
is a BIG maybe

So, one A7 is now using seperate R/C for each output tube's cathode
bias. The other A7 is pouting, silently. Just because I can not see the
need for triple grounding, doesn't mean it is not there ...

There is a nearby billboard ad with Santa blowing across an open bottle
of Coca-Cola and snow appearing on the far side. Snow is charming, in
ads ... Phoenix gets hot, but, I never have to shovel it 8^D

Happy Ears!
Al

[Ian Iveson]

Chris wrote:

The issue is one of a flat spot in the B/H curve across
zero-crossing. It's analogous to what's called "backlash"
in gear trains; the driving gear, if idled, must first
take up all the gear slack in the drive train before
power can be transmitted.

Note that this is an issue separate from and unrelated to
hysteresis, which is a dynamic problem occuring at all
signal levels. Hysteresis is more analogous to the
overshoot of a mass/compliance pair.

Of course it is related to hysteresis.

When you say "flat spot at zero crossing", what do you mean is zero,
and what is flat? Some clarity might help.

A common misconception seems to be that there is a flat spot on the BH
curve where it crosses the H axis; i.e. that dB/dH is at a minimum
when the BH curve where H is zero. This would be true were it not for
hysteresis.

"Flat spot at zero crossing" and "hysteresis" may not be *necessarily*
related conceptually, but they are in fact entangled, both by cause
and in effect.

And a flat spot in a gearbox, rather like a thermostat with a gap
between switch-on and switch-off, gives rise to a phenomenon which is
commonly called hysteresis.

Anyway, I introduced hysteresis because it seems to me that the two
effects combine to complicate the outcome.

I avoided the slack gearbox analogy because it can't stand much
analysis. Otherwise I might have suggested you consider steering gear.
The flat spot is always around the current position of the steering
wheel. Bias doesn't help. Ah, but the teeth are worn in proportion to
their proximity to the straight-ahead steering position, so there
isn't as much slack on a corner...

cheers, Ian

[tubegarden]

Ian Iveson wrote:
Chris wrote:

The issue is one of a flat spot in the B/H curve across
zero-crossing. It's analogous to what's called "backlash"
in gear trains; the driving gear, if idled, must first
take up all the gear slack in the drive train before
power can be transmitted.

Note that this is an issue separate from and unrelated to
hysteresis, which is a dynamic problem occuring at all
signal levels. Hysteresis is more analogous to the
overshoot of a mass/compliance pair.

Of course it is related to hysteresis.

When you say "flat spot at zero crossing", what do you mean is zero,
and what is flat? Some clarity might help.

A common misconception seems to be that there is a flat spot on the BH
curve where it crosses the H axis; i.e. that dB/dH is at a minimum
when the BH curve where H is zero. This would be true were it not for
hysteresis.

"Flat spot at zero crossing" and "hysteresis" may not be *necessarily*
related conceptually, but they are in fact entangled, both by cause
and in effect.

And a flat spot in a gearbox, rather like a thermostat with a gap
between switch-on and switch-off, gives rise to a phenomenon which is
commonly called hysteresis.

Anyway, I introduced hysteresis because it seems to me that the two
effects combine to complicate the outcome.

I avoided the slack gearbox analogy because it can't stand much
analysis. Otherwise I might have suggested you consider steering gear.
The flat spot is always around the current position of the steering
wheel. Bias doesn't help. Ah, but the teeth are worn in proportion to
their proximity to the straight-ahead steering position, so there
isn't as much slack on a corner...

cheers, Ian

Hi RATs!

So, I have all the parts I need to implement and listen to the Blumlein
garters, but, feast and famine, my bod is not in the mood to sit up. I
have gotten used to not being able to go to work. (I never loved it as
much as I pretended) But, it still frustrates me when I am unable to do
anything useful. Complaining on Newsgroups is popular, but, I don't
really love it, neither

Trying to describe perceived phenomenon in a manner which makes
knowledge applicable to real circuits is a Noble Goal. I admit I reread
the basic literature many times before I got anywhere near being able
to build a circuit which performed. Well, occasionally. Actually,
building circuits from scratch is pretty successful ... modifying old
stuff occasionally leaves me wishing I had just gutted it and started
over. Then I gut it and start over and all is well. Now I am less
energetic, but, since I believe I can do it, I will, but, time is just
not tightly coupled to enthusiasm these days.

The sequencial analysis of the balancing action of the circuit is
simply incorrect. Both cathodes are actively adjusting at all times.
Whenever an imbalance appears, if ever , each cathode adjusts the
other's bias voltage. As one sends a boost, the other sends a
reduction. It ain't a runaway feedback loop. I don't know how it will
sound, but I do know how to accept two things happening at the same
time ...

I must admit I am biased in favor of the idea that a shared cathode
resistor/capacitor was not selected on the wonderful performance of
anything but COGS.

We dream and smile and listen to Music. Somebody is busy trying to suck
up the ladder by saving the company money ... the true nature of the
human conflict, people who want to do the best job and people who want
to get ahead. Not the same trip. Of course, both ends of that spectrum
want their audio system to humble all others

I am overjoyed that my system humbles me

Happy Ears!
Al

[Chris Hornbeck]

On Tue, 05 Dec 2006 05:33:26 GMT, "Ian Iveson"
wrote:

Note that this is an issue separate from and unrelated to
hysteresis, which is a dynamic problem occuring at all
signal levels. Hysteresis is more analogous to the
overshoot of a mass/compliance pair.

Of course it is related to hysteresis.

much good analysis snipped 'cause you've alreardy read it

Hysteresis is dynamic; it occurs at all times and
at all points on the B/H curve.

In addition, there is a non-linearity in magnetization vs.
field strength unique to the near-zero-crossing area.

The defining practical difference between the (what word
would you agree to use for this phenomenon?) and hysteresis
is that hysteresis operates everywhere, and the (name to
be decided by popular referendum) is only significant
around zero crossing.

The n.t.b.d.b.p.r effect can be moved away from the small
signal area by biasing the magnetization up onto the straight
line part. Hysteresis is everywhere, including the small signal
area.

The difference could also be thought of as the difference
between a static and a dynamic effect, but I'm afraid
that this also won't pass muster by a determined critic
of over-simplified analogies. Arf!


Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

[Ian Iveson]

Chris wrote below

You think I'm being pernickety, I can tell.

But we haven't quite grasped what I'm saying. Perhaps because it's so
simple.

What is zero at your zero crossing, please?

Thanks, Ian

"Chris Hornbeck" wrote in message
...
On Tue, 05 Dec 2006 05:33:26 GMT, "Ian Iveson"
wrote:

Note that this is an issue separate from and unrelated to
hysteresis, which is a dynamic problem occuring at all
signal levels. Hysteresis is more analogous to the
overshoot of a mass/compliance pair.

Of course it is related to hysteresis.

much good analysis snipped 'cause you've alreardy read it

Hysteresis is dynamic; it occurs at all times and
at all points on the B/H curve.

In addition, there is a non-linearity in magnetization vs.
field strength unique to the near-zero-crossing area.

The defining practical difference between the (what word
would you agree to use for this phenomenon?) and hysteresis
is that hysteresis operates everywhere, and the (name to
be decided by popular referendum) is only significant
around zero crossing.

The n.t.b.d.b.p.r effect can be moved away from the small
signal area by biasing the magnetization up onto the straight
line part. Hysteresis is everywhere, including the small signal
area.

The difference could also be thought of as the difference
between a static and a dynamic effect, but I'm afraid
that this also won't pass muster by a determined critic
of over-simplified analogies. Arf!


Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

[Chris Hornbeck]

On Thu, 07 Dec 2006 03:25:43 GMT, "Ian Iveson"
wrote:

You think I'm being pernickety, I can tell.

On the contrary; I think you're making an important
point. I was just riffing on some old discussion, the
details of which are now forgotten. Wasn't all that
funny anywho; sorry.

But we haven't quite grasped what I'm saying. Perhaps because it's so
simple.

What is zero at your zero crossing, please?

Zero magnetization vs. zero field strength. Sorry I didn't
respond to this earlier. In America, "zero crossing" is a
common usage with an accepted meaning. We're "separated
by a common language".


The n.t.b.d.b.p.r effect can be moved away from the small
signal area by biasing the magnetization up onto the straight
line part. Hysteresis is everywhere, including the small signal
area.

The above was particularly poorly expressed, even for me.
If you'll allow me to try again:

The flat spot in the magnetization curve around zero-
crossing is mapped into the small-signal region in
transformers (or any other iron-cored inductors) where
no "bias" or standing magnetization exists without
signal.

In designs where "bias" or standing magnetization exists
at zero signal, the small-signal region is moved "up" the
curve, away from the non-linearities of the zero-crossing
region.

Iron-cored inductors exact quite a price for this luxury,
in cost, weight, inductance, yada-yada, but the price buys
monotonicity.


Hysteresis differs from the above effect in two ways. First,
it cannot be "moved" somwhere else by bias. It exists at
all points on the curve.

And second, it is "dynamic" in the sense that it's a memory
effect. Hysteresis is an inability to retrace the original
curve when applying the exact same stimulus a short time
later.

The flat spot on the magnetization curve occurs "statically",
independent of history and exactly repeating the same curve
coming from either direction.

Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

[tubegarden]

Chris Hornbeck wrote:
On Thu, 07 Dec 2006 03:25:43 GMT, "Ian Iveson"
wrote:

You think I'm being pernickety, I can tell.

On the contrary; I think you're making an important
point. I was just riffing on some old discussion, the
details of which are now forgotten. Wasn't all that
funny anywho; sorry.

But we haven't quite grasped what I'm saying. Perhaps because it's so
simple.

What is zero at your zero crossing, please?

Zero magnetization vs. zero field strength. Sorry I didn't
respond to this earlier. In America, "zero crossing" is a
common usage with an accepted meaning. We're "separated
by a common language".


The n.t.b.d.b.p.r effect can be moved away from the small
signal area by biasing the magnetization up onto the straight
line part. Hysteresis is everywhere, including the small signal
area.

The above was particularly poorly expressed, even for me.
If you'll allow me to try again:

The flat spot in the magnetization curve around zero-
crossing is mapped into the small-signal region in
transformers (or any other iron-cored inductors) where
no "bias" or standing magnetization exists without
signal.

In designs where "bias" or standing magnetization exists
at zero signal, the small-signal region is moved "up" the
curve, away from the non-linearities of the zero-crossing
region.

Iron-cored inductors exact quite a price for this luxury,
in cost, weight, inductance, yada-yada, but the price buys
monotonicity.


Hysteresis differs from the above effect in two ways. First,
it cannot be "moved" somwhere else by bias. It exists at
all points on the curve.

And second, it is "dynamic" in the sense that it's a memory
effect. Hysteresis is an inability to retrace the original
curve when applying the exact same stimulus a short time
later.

The flat spot on the magnetization curve occurs "statically",
independent of history and exactly repeating the same curve
coming from either direction.

Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."


Hi RATs!

OK, I am now listening to two amps:
one P-P pentode in UL (not yet with "garters") NFB OT to driver
cathode.
one triode in SE

This string quartet is very good, but, I cannot hear any hysterisis nor
zero-crossing flat spots, so I ask:

On what recording do you detect these phenomena while listening?

How may these two technolgies produce one exquisite voice? Sex???

Happy Ears!
Al

"Almost kaput, still not hip!"

[Ian Iveson]

Al wrote:

On what recording do you detect these phenomena while listening?

How may these two technolgies produce one exquisite voice? Sex???

Ballet. I think they dance together. Zero crossing is the one in the
tutu; hysteresis wears the tight pants.

Music is all about dance but it's a silent chorus.

It seems to be established that crossover distortion is less offensive
if moved away from the point of zero signal voltage. Applied to
transformer output stages, some argue that sound reproduction can be
improved if the transformer is biased. If your garters succeed in
ensuring balanced current, then your transformer will have zero
magnetic bias too, goes the argument, so the perfect cancellation of
distortion from the valves is disrupted by the spectre of magnetic
crossover effects exactly where you least want them. I may have put
the argument badly because I don't quite follow it. Seems to me it
leaves out remanence.

cheers, Ian

[Ian Iveson]

Chris Hornbeck wrote: [below]

Zero crossing is fine here, but I think everywhere it is important to
define the axes and say which is being crossed.

I've been thinking, and looking for pictures, again.

The beginning of the story is simple enough. A lowish frequency,
smallish amplitude voltage signal appears at the primary of a
transformer that happens to have initially zero magnetisation. The BH
curve is shallow so inductance is low, and consequently the reflected
secondary load is significantly shunted by the primary inductance. As
current increases through the inductance, then H increases, moving
away from the flat spot, so inductance rises and the shunt becomes
less significant. So far we have distortion of the first quadrant of
the signal.

From the point at which primary current, and hence H, begins to
reverse, the BH curve becomes a hysteresis loop. This shifts the flat
spot from the point of zero current, to the points of positive and
negative current maxima. The curve is now steepest, and hence
inductance greatest, where B crosses the H axis.

What I have had difficulty with is visualising how this 90 degree
shift of the flat spot coincides with the phase shift between voltage
and current. Assuming the flat spot remains at the point of zero
primary voltage, then so far I agree with you.

However, the flat spot is not at zero magnetisation, but at its
maxima. Not so flat there.

So it seems the flat spot is a first-cycle
problem...perhaps it could be described as transient distortion,
considering that it coincides with the transient response.

The pictures I really want would show a family of BH loops derived
from the first couple of cycles of an applied sine voltage, of various
amplitudes, with and without bias. A plot of inductance v applied *AC*
voltage would be handy too.

So don't think I agree with you, for the moment, that there is always
a flat spot at zero magnetisation.

Most links I can find that show the loop omit the flat spot from the
initial curve, and vice-versa. This is one that shows both:

http://pioneer.netserv.chula.ac.th/~.../Magnetics.pdf

If I could find a way of building a subcircuit in spice that stores
energy, and loses some of it, without introducing an unwanted delay, I
would make my own pictures.

cheers, Ian




On Thu, 07 Dec 2006 03:25:43 GMT, "Ian Iveson"
wrote:

You think I'm being pernickety, I can tell.

On the contrary; I think you're making an important
point. I was just riffing on some old discussion, the
details of which are now forgotten. Wasn't all that
funny anywho; sorry.

But we haven't quite grasped what I'm saying. Perhaps because it's
so
simple.

What is zero at your zero crossing, please?

Zero magnetization vs. zero field strength. Sorry I didn't
respond to this earlier. In America, "zero crossing" is a
common usage with an accepted meaning. We're "separated
by a common language".


The n.t.b.d.b.p.r effect can be moved away from the small
signal area by biasing the magnetization up onto the straight
line part. Hysteresis is everywhere, including the small signal
area.

The above was particularly poorly expressed, even for me.
If you'll allow me to try again:

The flat spot in the magnetization curve around zero-
crossing is mapped into the small-signal region in
transformers (or any other iron-cored inductors) where
no "bias" or standing magnetization exists without
signal.

In designs where "bias" or standing magnetization exists
at zero signal, the small-signal region is moved "up" the
curve, away from the non-linearities of the zero-crossing
region.

Iron-cored inductors exact quite a price for this luxury,
in cost, weight, inductance, yada-yada, but the price buys
monotonicity.


Hysteresis differs from the above effect in two ways. First,
it cannot be "moved" somwhere else by bias. It exists at
all points on the curve.

And second, it is "dynamic" in the sense that it's a memory
effect. Hysteresis is an inability to retrace the original
curve when applying the exact same stimulus a short time
later.

The flat spot on the magnetization curve occurs "statically",
independent of history and exactly repeating the same curve
coming from either direction.

Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."


Hi RATs!

OK, I am now listening to two amps:
one P-P pentode in UL (not yet with "garters") NFB OT to driver
cathode.
one triode in SE

This string quartet is very good, but, I cannot hear any hysterisis
nor
zero-crossing flat spots, so I ask:

On what recording do you detect these phenomena while listening?

How may these two technolgies produce one exquisite voice? Sex???

Happy Ears!
Al

"Almost kaput, still not hip!"