[Ian Iveson]

Just an occasional reminder.

cheers, Ian

[Andre Jute]

Ian Iveson wrote:
Just an occasional reminder.

cheers, Ian

Shhh! You'll wake up Wilson; he'll insist that we breathe negative
feedback.

Andre Jute

[Prune]

A Google search for triode internal feedback disagrees.

[Ian Iveson]

Prune wrote

A Google search for triode internal feedback disagrees.

That's who I was reminding.

The truth is only useful for countering lies.

Actually, I forgot the Miller effect this time round. The Miller effect is
feedback.

But I bet plenty of those Google ppl you found think that sagging under load is
evidence of feedback. Engineers know better.

cheers, Ian

[Bob H.]

We don't need no stinking feedback with a good triode.


Bob H.


Ian Iveson wrote:
Just an occasional reminder.

cheers, Ian

[22/7]

By the same token, there is no feedback in an operational amplifier
either.

22/7

Ian Iveson wrote:
Just an occasional reminder.

cheers, Ian

[Ian Iveson]

22/7 wrote

By the same token, there is no feedback in an operational amplifier
either.


What token is that?

An opamp without feedback is only useful in a circuit in which feedback has been
added, AFAIK, BTW. Some argue that triodes are useful without feedback.

cheers, Ian

[Patrick Turner]

Ian Iveson wrote:

Prune wrote

A Google search for triode internal feedback disagrees.

That's who I was reminding.

The truth is only useful for countering lies.

Actually, I forgot the Miller effect this time round. The Miller effect is
feedback.

But I bet plenty of those Google ppl you found think that sagging under load is
evidence of feedback. Engineers know better.

Miller effect is NFB as F increases, thus limiting RF gain in triodes.

But due to electrostatic NFB as distinct from Miller capacitance effects the load
voltage sags
less with triodes than with any other devices.

If the load has too low a value, little anode voltage change occurs, so there
is little NFB voltage. With a high value of load there is also no sag in anode
voltage, voltage gain
approaches µ, or gm x Ra of the triode as load approaches a CCS.
The distortion is very low because the distortion is all fed back by the anode
to the electron stream to cancel its own production. Lowest THD occurs because
the applied VOLTAGE NFB is maximal when no current change occurs in the triode.
The load which allows maximum AF output POWER is usually between
3 and 5 x Ra, or about ( Ea / Ia ) - 2Ra and allows the production of about 5% THD
at clipping.
The electrostatic field effects of anode and grid have a NET effect on the electron
stream feedback
While not perfect, triodes can have more voltage linearity than any other devices
which all require external loops of NFB to make them linear and act as low
impedance
signal generators.

Since you maintain there is no NFB in a triode, could you
point out your detailed logic to proove why not?

Patrick Turner.



cheers, Ian

[Patrick Turner]

"Bob H." wrote:

We don't need no stinking feedback with a good triode.

Bob H.

Indeed you may not need external smelly NFB loops but that's because of
the NFB
ALREADY installed within a triode.

Patrick Turner.



Ian Iveson wrote:
Just an occasional reminder.

cheers, Ian

[Patrick Turner]

22/7 wrote:

By the same token, there is no feedback in an operational amplifier
either.

Indeed there is little NFB within an opamp.
They are useless without applied NFB loops, so we connect
what loops we want to give whatever functions we want; sometimes its
a combination of positive and negative voltage and current FB.

Patrick Turner.



22/7

Ian Iveson wrote:
Just an occasional reminder.

cheers, Ian

[robert casey]

Ian Iveson wrote:
Just an occasional reminder.


Depends on how you define "feedback".

A triode's plate current is influenced by the voltage on its plate. And
if you have a resister on the plate to B+, the resistor will have an
amplified signal voltage on it, which variance influences plate current
some more. Maybe that's feedback, maybe not, your call.

[Eeyore]

Ian Iveson wrote:

22/7 wrote

By the same token, there is no feedback in an operational amplifier
either.

What token is that?

Your stupid idea.

An opamp without feedback is only useful in a circuit in which feedback has been
added,

Not necessarily true.

Graham

[Eeyore]

Ian Iveson wrote:

Prune wrote

A Google search for triode internal feedback disagrees.

That's who I was reminding.

The truth is only useful for countering lies.

Actually, I forgot the Miller effect this time round. The Miller effect is
feedback.

But I bet plenty of those Google ppl you found think that sagging under load is
evidence of feedback. Engineers know better.

You're evidently no engineer.

Graham

[Eeyore]

Just an occasional reminder.

Graham

[Ian Iveson]

robert wrote

Depends on how you define "feedback".

Quite so.

A triode's plate current is influenced by the voltage on its plate. And if
you have a resister on the plate to B+, the resistor will have an amplified
signal voltage on it, which variance influences plate current some more.
Maybe that's feedback, maybe not, your call.

Same could be said of a potentiometer.

I have argued the point formally and in detail before.

I take my definition of feedback from standard control systems engineering,
which is the genus of which audio engineering is a species. It is the same
methodology adopted by all branches of real engineering, and the one assumed by
RDH4, for example.

As I have said before, it is possible to apply control systems theory to
anything, including resistors. You can use it to arrive at Ohm's law, even. That
doesn't mean that resistors have "internal feedback". In just the same way, you
can apply it to a triode, and with a fair amount of contrivance, you can derive
part of Child's law. But systems theory is not necessary for either derivation,
although it *is* necessary for understanding and using the feedback we apply in
our circuits.

The Miller effect accounts sufficiently for the direct effect of the anode
voltage on the grid. Even there, it is only significant if external impedance is
added to the grid.There is no summing point where the output voltage is compared
with the input voltage ahead of the forward transfer function.

Those who invent imaginary internal circuits made out of vacuum can never prove
that they are real, because they are not. They are mathematical contrivances and
have never, ever, been found to be useful in practice, either in circuit design
or in valve design. It is not up to me to prove that imagination is not real. It
is up to them to show it is useful.

It has been suggested, preposterously, that the writers of RDH4 didn't know
about systems theory. Actually the book cites several standard texts on control
systems engineering for background reference. It is where all the stuff like
Bode theory, feedback equations, Fourier, Laplace, Nyquist, etc etc comes from.

It has also been argued, equally desperately, that the writers of RDH4 didn't
know how triodes work. Also that the manufacturers of valves didn't know how
they worked. In fact they knew pretty much the same as we do, as far as it is
necessary to know anything at all.

It should be clear to everyone that the notion of internal triode feedback is a
source of stupid confusion. How many times have you come across an argument that
starts like:

A: I like SETs because I don't like to use feedback

B: But triodes have internal feedback.

Ya de ya, blah blah.

Often some useless debate ensues about the difference between one kind of
feedback and another, as if "internal" somehow is better than "external" or
"added" ...blah blah, whatever.

Let's keep it simple. For engineering purposes, the meaning of "feedback" could
not be more clearly defined. It is as tight as you could wish. Check out any
recognised general text on control systems engineering. You will find the same
strict definition, pretty much word for word, in any such reference.

I won't get round to replying to everyone, and I won't have arguments with those
who simply refuse to abandon the illusions they propagate for commercial or
other dishonest purposes.

You have put your finger on one essential point. If there is no delay, then in
the real world there is no feedback. Also, if there is no delay, then there is
absolutely no reason to resort to systems theory. None of that stuff...Bode,
Laplace, etc...is useful if there is no delay. Whatever may or may not be bad
about feedback therefore cannot be bad where there is no delay.

Not keen on analogies, but this may be a way of looking at it:

You want a car that has a linear response to the accelerator pedal. You find
that wind resistance introduces a problem of sagging. Wind resistance is not
feedback in your system because the throttle, which is your input, is not
sensitive to it. You can deal with the effect of the wind load in two ways. You
may add negative feedback, which is what cruise control does, or you can
tailgate, which is a bit like what a pentode's so-called screen does. Tailgating
is not feedback, and neither does it magically remove some kind of mystical
internal feedback inside your car.

cheers, Ian

[robert casey]

By the same token, there is no feedback in an operational amplifier
either.



What token is that?


Here in New York City they don't have tokens anymore. We have to talk
about metro cards now. :-)

[Arny Krueger]

"flipper" wrote in message

On Sat, 12 Aug 2006 17:08:14 GMT, robert casey
wrote:

Ian Iveson wrote:
Just an occasional reminder.


Depends on how you define "feedback".

A triode's plate current is influenced by the voltage on
its plate. And if you have a resister on the plate to
B+, the resistor will have an amplified signal voltage
on it, which variance influences plate current some
more. Maybe that's feedback, maybe not, your call.

Seems to me it's whether you consider the load to be 'a
part of the triode' because without it there is no
feedback.

Just because you don't drive it with a signal, doesn't mean that the
negative feedback isn't there.

[Patrick Turner]

Ian Iveson wrote:

robert wrote

Depends on how you define "feedback".

Quite so.

A triode's plate current is influenced by the voltage on its plate. And if
you have a resister on the plate to B+, the resistor will have an amplified
signal voltage on it, which variance influences plate current some more.
Maybe that's feedback, maybe not, your call.

Same could be said of a potentiometer.

I have argued the point formally and in detail before.

I take my definition of feedback from standard control systems engineering,
which is the genus of which audio engineering is a species. It is the same
methodology adopted by all branches of real engineering, and the one assumed by
RDH4, for example.

As I have said before, it is possible to apply control systems theory to
anything, including resistors. You can use it to arrive at Ohm's law, even. That
doesn't mean that resistors have "internal feedback". In just the same way, you
can apply it to a triode, and with a fair amount of contrivance, you can derive
part of Child's law. But systems theory is not necessary for either derivation,
although it *is* necessary for understanding and using the feedback we apply in
our circuits.

The Miller effect accounts sufficiently for the direct effect of the anode
voltage on the grid. Even there, it is only significant if external impedance is
added to the grid.There is no summing point where the output voltage is compared
with the input voltage ahead of the forward transfer function.

Those who invent imaginary internal circuits made out of vacuum can never prove
that they are real, because they are not. They are mathematical contrivances and
have never, ever, been found to be useful in practice, either in circuit design
or in valve design. It is not up to me to prove that imagination is not real. It
is up to them to show it is useful.

It has been suggested, preposterously, that the writers of RDH4 didn't know
about systems theory. Actually the book cites several standard texts on control
systems engineering for background reference. It is where all the stuff like
Bode theory, feedback equations, Fourier, Laplace, Nyquist, etc etc comes from.

It has also been argued, equally desperately, that the writers of RDH4 didn't
know how triodes work. Also that the manufacturers of valves didn't know how
they worked. In fact they knew pretty much the same as we do, as far as it is
necessary to know anything at all.

It should be clear to everyone that the notion of internal triode feedback is a
source of stupid confusion. How many times have you come across an argument that
starts like:

A: I like SETs because I don't like to use feedback

B: But triodes have internal feedback.

Ya de ya, blah blah.

Often some useless debate ensues about the difference between one kind of
feedback and another, as if "internal" somehow is better than "external" or
"added" ...blah blah, whatever.

Let's keep it simple. For engineering purposes, the meaning of "feedback" could
not be more clearly defined. It is as tight as you could wish. Check out any
recognised general text on control systems engineering. You will find the same
strict definition, pretty much word for word, in any such reference.

I won't get round to replying to everyone, and I won't have arguments with those
who simply refuse to abandon the illusions they propagate for commercial or
other dishonest purposes.

You have put your finger on one essential point. If there is no delay, then in
the real world there is no feedback. Also, if there is no delay, then there is
absolutely no reason to resort to systems theory. None of that stuff...Bode,
Laplace, etc...is useful if there is no delay. Whatever may or may not be bad
about feedback therefore cannot be bad where there is no delay.

Not keen on analogies, but this may be a way of looking at it:

You want a car that has a linear response to the accelerator pedal. You find
that wind resistance introduces a problem of sagging. Wind resistance is not
feedback in your system because the throttle, which is your input, is not
sensitive to it. You can deal with the effect of the wind load in two ways. You
may add negative feedback, which is what cruise control does, or you can
tailgate, which is a bit like what a pentode's so-called screen does. Tailgating
is not feedback, and neither does it magically remove some kind of mystical
internal feedback inside your car.

cheers, Ian

So what do you think about the writings of Professor Child in
Terman's Radio Engineering of 1937 where he discusses the phenomena of "self
regulation"
ie, NFB in a triode?

What you have said above explains nothing about NFB.
You say NFB doesn't exist in a triode but you have not prooved your position.

Patrick Turner.

[Patrick Turner]

flipper wrote:

On Sat, 12 Aug 2006 17:08:14 GMT, robert casey
wrote:

Ian Iveson wrote:
Just an occasional reminder.


Depends on how you define "feedback".

A triode's plate current is influenced by the voltage on its plate. And
if you have a resister on the plate to B+, the resistor will have an
amplified signal voltage on it, which variance influences plate current
some more. Maybe that's feedback, maybe not, your call.

Seems to me it's whether you consider the load to be 'a part of the
triode' because without it there is no feedback.

The NFB in a triode only works where there is voltage change caused by
current change.
so a load is essential for triode NFB to work, but the load is NOT part of
the triode, and need not be.

The triode FB path is a an electrostatic shunt voltage negative feedback
path.
Its very similar to having a pentode with a shunt resistor nework between the
input
voltage, grid, and anode, where the output is 180 degrees of opposite phase
to the input.
But with electrostatic field effects the impedance of the feedback path arms
are very high,
much higher than any resistance values one may choose in a shunt FB network.
The impepance path of electostatic effects from anode to the electron stream
is high, but then reduces because of grid to anode capacitance.
They are not the same thing though, the Cag and field effects.

Pentodes have a screen which prevent MOST but NOT ALL anode voltage changes
from having much effect on the electron stream.
A pentode will have a finite value of Ra even with a screen.
Ie, there will be a an Ia change with Ea change even though Eg1 and Eg2 is
kept fixed.

If the screen action was perfect, the Ra of the pentode would be infinite,
and
since for all tubes µ = gm x Ra, then pentode gain would be infinite
where the load was a CCS.

But under certain Ia & Ea idle conditions, pentode µ is indeed extremely
high,
perhaps 3,000, and linearity over a wide anode voltage swing is not very
good,
and appraoaching the appalling voltage non-linearity of a j-fet or bjt
without any NFB.
Its possible to build a line stage with a humble 6AU6 with its enormous open
loop gain
of several thousand reduced to just 10, which is a gain reduction of say 200
times, or 46dB.
If the THD was 0.1% at 1V out which is typical without any shunt FB and
(perhaps 5% at 20Vrms output) then
after FB with A = 10, its reduced 46db to 0.0005% at the 1V.
The Ra of 500k can be reduced to 2.5k.

External loops of NFB make just about any device more linear if gm is
sufficient
and the load allows for adequate open loop gain.

In triodes, there is NFB within, but loops of external NFB can also be added.



Patrick Turner.

[Ian Iveson]

What you have said above explains nothing about NFB.
You say NFB doesn't exist in a triode but you have not prooved your position.

I have previously cited references to standard texts on control systems
engineering. I have pointed out that RDH has similar references. I am not trying
on this occasion to explain what feedback is, especially not to you, considering
your uncouth, foul-mouthed, ignorant and childish response last time. Check back
and follow my references, or follow those in RDH.

Neither am I trying to prove my position. I merely offer relief from confusion.

So what do you think about the writings of Professor Child in
Terman's Radio Engineering of 1937 where he discusses the phenomena of "self
regulation"
ie, NFB in a triode?

You have put words in the mouth of a writer you clearly don't understand.

Feedback is not necessary for regulation. I am not surprised that you don't know
that. Perhaps you believe that a power transformer, for which a value may be
quoted for "regulation", also has internal feedback? Then all amplifiers with
power transformers would have feedback. The same could be argued for every
component, active or passive. Everything has feedback if that's all you mean by
it. But all cows are black only at night.

Some ppl say god exists. Since that cannot be proven either way, and since there
are perfectly adequate, simpler and less mysterious explanations for the way the
world is, it is safe to assume it is not true for practical purposes.

You have often confused triodes with god. You are also dishonest, poorly
educated, dim, and not worth talking to.

cheers, Ian

[Ian Iveson]

Arny wrote

Just because you don't drive it with a signal, doesn't mean that the negative
feedback isn't there.

There? Where? It's a vacuum.

Since you are a leading exponent of naive realism in audio, I shouldn't be
surprised to find you lost in metaphysics.

cheers, Ian

[Phil]

Patrick Turner wrote:


Ian Iveson wrote:


Prune wrote


A Google search for triode internal feedback disagrees.

That's who I was reminding.

The truth is only useful for countering lies.

Actually, I forgot the Miller effect this time round. The Miller effect is
feedback.

But I bet plenty of those Google ppl you found think that sagging under load is
evidence of feedback. Engineers know better.


Miller effect is NFB as F increases, thus limiting RF gain in triodes.

But due to electrostatic NFB as distinct from Miller capacitance effects the load
voltage sags
less with triodes than with any other devices.

If the load has too low a value, little anode voltage change occurs, so there
is little NFB voltage. With a high value of load there is also no sag in anode
voltage, voltage gain
approaches µ, or gm x Ra of the triode as load approaches a CCS.
The distortion is very low because the distortion is all fed back by the anode
to the electron stream to cancel its own production. Lowest THD occurs because
the applied VOLTAGE NFB is maximal when no current change occurs in the triode.
The load which allows maximum AF output POWER is usually between
3 and 5 x Ra, or about ( Ea / Ia ) - 2Ra and allows the production of about 5% THD
at clipping.
The electrostatic field effects of anode and grid have a NET effect on the electron
stream feedback
While not perfect, triodes can have more voltage linearity than any other devices
which all require external loops of NFB to make them linear and act as low
impedance
signal generators.

Since you maintain there is no NFB in a triode, could you
point out your detailed logic to proove why not?

Feedback exists IF AND ONLY IF, you can change the input voltage, i.e.,
control voltage, by changing the output voltage. In a triode, this
occurs only at frequencies where the Miller capacitance starts to take
effect. Note that in all standard solid state devices, the ONLY way to
achieve low output impedance is by using feedback. And yes, if you
change the output voltage of a cathode/emitter/source follower, you
change the cathode-grid voltage, so these are feedback circuits (which
is why it is sometimes too easy to make oscillators out of them!). To my
knowledge, only power J-Fets, also known as Static Induction
Transistors, are exceptions, and they have curves that look EXACTLY like
a triode's except with a "plate" impedance (drain impedance) of 2 ohms,
instead of 200 to 20,000 ohms, like you have in a tube.

Also note that whereas a triode can use a transformer to reduce its
"natural," feedbackless output impedance down to 1 or 2 ohms, not even a
transformer can save standard solid state devices, because in order to
use a transformer, the output signal itself must have a high voltage to
current ratio, which is not the case for solid states. This "voltage to
current ratio impedance" is different from the output impedance. For
example, a pentode has both a high voltage to current ratio, and a high
output impedance, but if you triode-strap it, its output impedance
drops, even though its voltage to current ratio stays the same.

Phil

Patrick Turner.



cheers, Ian

[Andre Jute]

Ian Iveson wrote:
Patick Turner wrote:
What you have said above explains nothing about NFB.
You say NFB doesn't exist in a triode but you have not prooved your position.

I have previously cited references to standard texts on control systems
engineering. I have pointed out that RDH has similar references. I am not trying
on this occasion to explain what feedback is, especially not to you, considering
your uncouth, foul-mouthed, ignorant and childish response last time. Check back
and follow my references, or follow those in RDH.

Neither am I trying to prove my position. I merely offer relief from confusion.

Iveson, I sincerely hope that even the misguided British government has
enough smarts not to let you loose on the children of the nation as a
teacher.

You made a wildly provocative post, you don't have the engineering to
back it up, the list of those you refuse to discuss it with includes
everone capable of discussing the matter (except John Byrns, who is
absent, and I'm sure you will find a reason to include him in your list
of banned experts should he return), so what will you do now? Waffle on
mindlessly and bask in the glow of those who have not yet grasped that
words have precise meanings? If you insist that your ramblings should
appear unedited and not be subject to review, how can anyone believe
anything you say? You reputation has been in shreds ever since you
declared yourself the slew rate expert; now even the shreds have fallen
away to leave you naked and pretty clearly clueless. You disappoint me,
man. I would have bet that, if you can't escape your ignorance, you
would at least be glib enough to put a good face on it.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"wonderfully well written and reasoned information
for the tube audio constructor"
John Broskie TubeCAD & GlassWare
"an unbelievably comprehensive web site
containing vital gems of wisdom"
Stuart Perry Hi-Fi News & Record Review


So what do you think about the writings of Professor Child in
Terman's Radio Engineering of 1937 where he discusses the phenomena of "self
regulation"
ie, NFB in a triode?

You have put words in the mouth of a writer you clearly don't understand.

Feedback is not necessary for regulation. I am not surprised that you don't know
that. Perhaps you believe that a power transformer, for which a value may be
quoted for "regulation", also has internal feedback? Then all amplifiers with
power transformers would have feedback. The same could be argued for every
component, active or passive. Everything has feedback if that's all you mean by
it. But all cows are black only at night.

Some ppl say god exists. Since that cannot be proven either way, and since there
are perfectly adequate, simpler and less mysterious explanations for the way the
world is, it is safe to assume it is not true for practical purposes.

You have often confused triodes with god. You are also dishonest, poorly
educated, dim, and not worth talking to.

cheers, Ian

[Patrick Turner]

Ian Iveson wrote:

What you have said above explains nothing about NFB.
You say NFB doesn't exist in a triode but you have not prooved your position.

I have previously cited references to standard texts on control systems
engineering. I have pointed out that RDH has similar references. I am not trying
on this occasion to explain what feedback is, especially not to you, considering
your uncouth, foul-mouthed, ignorant and childish response last time. Check back
and follow my references, or follow those in RDH.

Calling me a lot of names does not constitute any proof of the correctness
of your argument that NFB does not exist in a triode.

Calling me names indicates that you hate your pet theories challenged.

If you don't prrove your points, then it could be said that you think people like
Professor Child is a liar.
Now you both can't be right, but I back the old professor, not you.




Neither am I trying to prove my position. I merely offer relief from confusion.

You have ZERO validity of any position unless you proove it.
You create confusion by stating there isn't any NFB in a triode.



So what do you think about the writings of Professor Child in
Terman's Radio Engineering of 1937 where he discusses the phenomena of "self
regulation"
ie, NFB in a triode?

You have put words in the mouth of a writer you clearly don't understand.

There is very little understanding of a triode in your mind it would seem.



Feedback is not necessary for regulation.

But the regulation effects of the interaction of the two electrostaic fields in a
triode, one
created at the grid and the other by the anode voltage is NFB, and results in the
low Ra of a triode. Without the electrostatic field effect from the anode acting on
the
electron stream there is very little regulation, and gain is merely gm x RL.
Ra is extremely high.

I am not surprised that you don't know
that. Perhaps you believe that a power transformer, for which a value may be
quoted for "regulation", also has internal feedback?

There is no electrostatic feedback effect in a power tranny.

Then all amplifiers with
power transformers would have feedback. The same could be argued for every
component, active or passive. Everything has feedback if that's all you mean by
it. But all cows are black only at night.

You remain ignorant of the interaction of two fields in a triode.



Some ppl say god exists. Since that cannot be proven either way, and since there
are perfectly adequate, simpler and less mysterious explanations for the way the
world is, it is safe to assume it is not true for practical purposes.

But certainly some larger power or being or force, presence known as God must have
been responsible for
the universe's existance.
No evidence has been found of a cordial but stern old giza with white flowing beard.

But you fail to proove there is no God.



You have often confused triodes with god. You are also dishonest, poorly
educated, dim, and not worth talking to.

As the debate unfolds the audience will judge for themselves who is right and wrong,

and your insults don't win any points in this debate.

Patrick Turner.





cheers, Ian

[Patrick Turner]

Phil wrote:

Patrick Turner wrote:


Ian Iveson wrote:


Prune wrote


A Google search for triode internal feedback disagrees.

That's who I was reminding.

The truth is only useful for countering lies.

Actually, I forgot the Miller effect this time round. The Miller effect is
feedback.

But I bet plenty of those Google ppl you found think that sagging under load is
evidence of feedback. Engineers know better.


Miller effect is NFB as F increases, thus limiting RF gain in triodes.

But due to electrostatic NFB as distinct from Miller capacitance effects the load
voltage sags
less with triodes than with any other devices.

If the load has too low a value, little anode voltage change occurs, so there
is little NFB voltage. With a high value of load there is also no sag in anode
voltage, voltage gain
approaches µ, or gm x Ra of the triode as load approaches a CCS.
The distortion is very low because the distortion is all fed back by the anode
to the electron stream to cancel its own production. Lowest THD occurs because
the applied VOLTAGE NFB is maximal when no current change occurs in the triode.
The load which allows maximum AF output POWER is usually between
3 and 5 x Ra, or about ( Ea / Ia ) - 2Ra and allows the production of about 5% THD
at clipping.
The electrostatic field effects of anode and grid have a NET effect on the electron
stream feedback
While not perfect, triodes can have more voltage linearity than any other devices
which all require external loops of NFB to make them linear and act as low
impedance
signal generators.

Since you maintain there is no NFB in a triode, could you
point out your detailed logic to proove why not?

Feedback exists IF AND ONLY IF, you can change the input voltage, i.e.,
control voltage, by changing the output voltage.

This is the case with any amplifier with a standard shunt voltage negative feedback
network.
Imagine a high gain pentode with a 10k input series R between the signal input and grid.
Then imagine a 100k resistor between grid and anode, and imagine the open gain of the
pentode
is above 1,000.
The closed loop gain will be about -10, because +1V input is amplified to become -10V
output.
Any distortion, noise, or load change voltage appearing at the anode is divided to 1/11
of voltage at the anode and applioed to the grid and thus opposes its own production.
The anode signal does not cause a change to the source input signal but causes a cjange
to the grid signal
which is remote from the input source because of the 10k R.


In the case of the triode, the input souce signal is connected to the grid which at AF is
always a high
Z input port to the amp.
But extending from the grid is the electrostatic arm of the grid, and extending from the
anode
there is the electrostatic arm of the anode, and their NET value is the actual resultant
field effect that causes the change in electron flow from cathode to anode.

The net electrostatic value acting on the electron stream is the REAL or equivalent
input point as described for the case above with pentode and simple resistor divider
shunt NFB network.

In a triode, this
occurs only at frequencies where the Miller capacitance starts to take
effect.

The Miller effect is present at even the lowest F but its effect on circuit gain only
becomes an issue when the Miller C impedance reduces and a divider between
series source resistance and the Miller C form a low pass filter.
In audio preamps using say a 12AX7 the Miller C may be effectively 120pF and
so the high audio F are seriously attenuated when the source R value of input to the grid
is say
from another high R anode signal.
The 120pF of miller C has an enormous Z at 50Hz, so Miller effects at LF are negligible.

But the electrostatic FB acts at all F.

Note that in all standard solid state devices, the ONLY way to
achieve low output impedance is by using feedback.

Agreed.

And yes, if you
change the output voltage of a cathode/emitter/source follower, you
change the cathode-grid voltage, so these are feedback circuits (which
is why it is sometimes too easy to make oscillators out of them!). To my
knowledge, only power J-Fets, also known as Static Induction
Transistors, are exceptions, and they have curves that look EXACTLY like
a triode's except with a "plate" impedance (drain impedance) of 2 ohms,
instead of 200 to 20,000 ohms, like you have in a tube.

Nearly all SS devices WITHOUT applied NFB have collector or drain resistance curves that
have no resemblance to triodes, but a close resemblance to a set of pentode Ra lines.
After series/shunt NFB is applied, the Ra lines or collector/drain lines resemble those
of a triode.
The data sheets for all thse devices are given without any NFB applied.
Where series current NFB is applied to a triode or other device the slope of the devices
output
Ra, Rc, Rs become more horizontal than they were to begin with without NFB applied.





Also note that whereas a triode can use a transformer to reduce its
"natural," feedbackless output impedance down to 1 or 2 ohms, not even a
transformer can save standard solid state devices, because in order to
use a transformer, the output signal itself must have a high voltage to
current ratio, which is not the case for solid states. This "voltage to
current ratio impedance" is different from the output impedance. For
example, a pentode has both a high voltage to current ratio, and a high
output impedance, but if you triode-strap it, its output impedance
drops, even though its voltage to current ratio stays the same.

The pentode such as EL34 has Ra = 12k.
So a 6k to 6 ohm OPT will transform the Ra from 12k to 12k/1,000 = 12 ohms at the
secondary.
This Rout at the sec is way too high to drive speakers to give a wanted flat speaker
response
so some folks would just connect the screen to the anode to allow the anode voltage to
deliver the full possible amount of internal electrostatic NFB.
The result is that the Ra drops to about 1,200 ohms and the Ra at the sec becomes 1.2
ohms
plus whatever OPT winding resistances exist.

Effectively a pentode is a current source where gain = gm x RL approx.
With the NFB of the triode applied it becomes substantially a voltage source
where gain = µ x RL / ( RL + Ra ).
This formula is valid for pentodes but because Ra is so high, the formula can be
simplified to A = gm x RL, and the pentode µ which is very high and somewhat variable
need not
be considered. The µ of the triode and pentode are considered very different amp factors,

and are each the product of gm x Ra, where gm and Ra are for either triode and pentode.
The triode NFB in an EL34 reduces the µ from about 130 in pentode to about 10 in triode.

The term current source describes signal sources where Rout is much higher than RL,
and voltage sources describes where Rout is much less than RL.

The UL connection is between the two extremes and where the partial anode signal voltage
applied to the
screen upon the electron stream gives Ra approx = RL.

Patrick Turner.







Phil

Patrick Turner.



cheers, Ian

[Patrick Turner]

flipper wrote:

On Sun, 13 Aug 2006 08:02:53 GMT, Patrick Turner
wrote:



flipper wrote:

On Sat, 12 Aug 2006 17:08:14 GMT, robert casey
wrote:

Ian Iveson wrote:
Just an occasional reminder.


Depends on how you define "feedback".

A triode's plate current is influenced by the voltage on its plate. And
if you have a resister on the plate to B+, the resistor will have an
amplified signal voltage on it, which variance influences plate current
some more. Maybe that's feedback, maybe not, your call.

Seems to me it's whether you consider the load to be 'a part of the
triode' because without it there is no feedback.

The NFB in a triode only works where there is voltage change caused by
current change.
so a load is essential for triode NFB to work,

Precisely.

but the load is NOT part of
the triode,

That's my opinion too.

and need not be.

Right, but without it there's no NFB, as you just established.

And if there is a case when it isn't there, and there is, then it
strikes me that it can't be called 'inherent', or at least not 'fully
inherent'.

Triode NFB only works when the triode is set up to have voltage gain.

All amplifiers need gain to allow the applied NFB to work.
An opamp with a NFB loop connected but with a shorted output
has no voltage gain, ( although it may have huge current gain.
If the NFB network is set up for series voltage NFB, then no
NFB is in operation when no gain exists.

The NFB network is still there, and inherent in the circuit, but just cannot do
anything.

Patrick Turner.



snip



Patrick Turner.

[Chris Hornbeck]

On Sat, 12 Aug 2006 20:52:30 GMT, "Ian Iveson"
wrote:

A triode's plate current is influenced by the voltage on its plate. And if
you have a resister on the plate to B+, the resistor will have an amplified
signal voltage on it, which variance influences plate current some more.
Maybe that's feedback, maybe not, your call.

Same could be said of a potentiometer.

Or even just a simple resistor.

You're in my always-read list, so I can't escape r.a.t forever,
apparently. Reading through this thread, I see two possible
arguments, both weak.

First is the analogy to Miller capacitance. This is weak
because capacitance is parasitic (ie, not intrinsic to the
triode; or to put it another way, tolerant of the triode's
characteristics and completely dependent on the combination
of triode and external circuitry).

Second is the analogy to multi-grid valves. This argument
turns the electron flow through the valve upside-down
and argues from "received wisdom", as it were. Poor
analogy, from several angles.


But there's no need for any discussion when a simple test
will suffice. Arrange a triode with hot filament, grid
supplied with a conventional, appropriate negative voltage
fed through high impedance. Vary its plate/ anode voltage
over a conventional, appropriate range.

Does the grid voltage vary? If so, feedback occurs. If not, not.


Arguments that the intrinsic characteristics of devices result
from "feedback" are exactly as nebulous as arguments that
claim that feedback is inherently evil.

Hoping all's well with you and yours, and with the newsgroup,

Chris Hornbeck
"This has been an account for those who don't keep them"
J-LG, _Tout Va Bien_ 1972

[Patrick Turner]

Chris Hornbeck wrote:

On Sat, 12 Aug 2006 20:52:30 GMT, "Ian Iveson"
wrote:

A triode's plate current is influenced by the voltage on its plate. And if
you have a resister on the plate to B+, the resistor will have an amplified
signal voltage on it, which variance influences plate current some more.
Maybe that's feedback, maybe not, your call.

Same could be said of a potentiometer.

Or even just a simple resistor.

You're in my always-read list, so I can't escape r.a.t forever,
apparently. Reading through this thread, I see two possible
arguments, both weak.

First is the analogy to Miller capacitance. This is weak
because capacitance is parasitic (ie, not intrinsic to the
triode; or to put it another way, tolerant of the triode's
characteristics and completely dependent on the combination
of triode and external circuitry).

Second is the analogy to multi-grid valves. This argument
turns the electron flow through the valve upside-down
and argues from "received wisdom", as it were. Poor
analogy, from several angles.

But there's no need for any discussion when a simple test
will suffice. Arrange a triode with hot filament, grid
supplied with a conventional, appropriate negative voltage
fed through high impedance. Vary its plate/ anode voltage
over a conventional, appropriate range.

Does the grid voltage vary? If so, feedback occurs. If not, not.

I think you are missing the point.
If you set up a 12AX7 triode in a normal common cathode circuit with gain = 70
and Miller C = 120pF, and feed the grid signal from a signal source of 4Mohms
source resistance,
then the Miller effect will give the circuit a low pass filter character with pole

determined by the 4M and 120pF RC time constant.

If you just have a 4M grid bias resistor to bias the 12AX7 and no input signal,
but apply a signal to the anode circuit, then that signal will cause a grid signal
change
determined by the 3pF of capacitance between anode and grid, and there is more
grid signal change as F rises. So that if a +ve RF signal is applied to the anode,
the
grid also swings +ve, increasing Ia, which opposes a rising anode voltage
application.
So feedback is evident on the principle of a moving grid voltage.
So the Ra of a triode at RF can be different at RF if the grid isn't fed by a low
source resistance signal.

But the NFB in the triode works at all F and at DC.
If we bias the 12AX7 with a lower R of say 100k, then alter the DC voltage of the
anode,
no grid voltage change is seen to occur; the increase in +ve anode voltage causes
more Ia to flow
which tries to oppose the attempt to raise the anode voltage.

But where we have the evidence of NFB is also seen in the DC and AC operation.
A 12AX7 set up with Ia = 0.6mA at Ea = 120V looks like a resistor
whose value = 120V / 0.0006A = 200k.
But it isn't like a resistor if we attempt to change the anode voltage while
keeping Eg1 constant.
We would measure that for small changes in Ea the Ra of the tube isn't simply Ea /
Ia,
but we may find it is about 65k. Resistors just don't act like this.
The triode is an active device with an ability to control their Ia by Ea.


The triode when used as an amplifier at AF right down to DC and with a low
impedance
source signal applied to the grid to ensure the Va output CANNOT change the
applied grid voltage
displays evidence of NFB.
Inside the triode, the electron stream is affected by TWO applications of
electrostatic fields.
These fields are just VOLTAGE effects, without requiring large current flows to
establish
the changes of voltages.
The beauty of the vacuum tube is that the grid voltage changes take almost no
power at all
to create, and the electrostatic field effect of the voltage change at the anode
also requires
virtually no power.
The TWO changing voltages inside the triode have a COMBINED effect on the
electrons around thre cathode If the grid swings +ve, more electrons are allowed
to flow to the anode.
The anode voltage swings negative with increasing load current.
The -ve going anode voltage tends to oppose the grid caused increase in Ia.
Electrons are less attracted to the anode because its voltage has become less
positive.
From an electron's point of view, the grid is say "Go Johnny Go!"
But the anode is telling "Hey, not so fast".
Electrons are subject to two masters, one is the grid, the other is the anode.

The electron controlling element inside the triode is the NET EFFECT of the TWO
changing voltages
at the two electrodes that are able to move. The cathode is assumed remain at 0V.
So where you have a triode as an amplifier, we will always see some THD because
the laws of current change in a triode relative to electrostatic field effects
is slightly non linear, as seen by the Ra lines for a tube diode.
The Ra lines fo diodes and triodes for fixed values of grid follow a graph
of x = cube root of y squared, rather inconvenient for us all but that's nautre.
The distortion appearing at the anode in a class A SET signal amp
applies itself to the electron stream to oppose its own production.
The amount of applied FB depends on the relative distances of the
grid and anode to the space charge of electrons around the cathode.
When the load of a triode is increased its gain increases so there is a greater
application of the NFB and greater reduction of distortion. The triode has a
maximum amount
of NFB applied when there is NO current change in the load which is a CCS
and the device is operating as a purely voltage in and voltage out device.
The voltage gain with CCS is then equal to the amplification factor, µ, which is
also
equal to gm x Ra, where gm is the transconductance of the triode, ie, mA/V of Ia
change
needed for 1V of change at the anode.
Ra is the measured dynamic anode resistance of the triode for the given Ea & Ia
set up conditions.
So what about where we have a CCS anode load? What is the gm then?
Would not the gm be zero because for one volt of anode change, there is ZERO Ia
change?
So Ia and Ea changes with grid changes result in gm being a very variable factor
in a tube.
But why do we see the fabulous natural linearity of a triode where there is zero
Ia change
even though Va change can be up to +/- 100V with less than 1% thd?
Why is the µ so easily maintained in any triode made to the same recipe
all around the world? Its NFB.
If the distances are relatively the same, the same FB is applied.

I doubt I have answered all the questions about triodes.
What happens where the grid is grounded, and input is applied to the cathode?
What happens to the NFB and how is such a circuit linearized maximally?

Rather than insult me and have argy bargy fights over who is or is not a BS artist

I urge those who doubt the existance of triode NFB to do a little more reading
to improve their basic knowledge of triodes and their use.grid v

A J-fet also is a device which controls electron current flow from an applied
field effect
to an electron flow. But the amount of NFB applied is very low,
since the drain has very little effect on the electron flow in the solid state
material.

In pentodes the anode has very little effect on electron flow because the variable
anode electrostatic
effect is interupted by the screen grid kept at a fixed voltage, so from the
electron's point of view they think the anode voltage isn't changing much, and
take their orders mainly only
from the grid voltage. Hence the gain of pentodes is much higher, but they also
have no mechanism to reduce the dynamic Ra, another result of zero NFB.
But without NFB and with very much reduced Miller C the pentode liberated
mankind to be able to explore high RF for communications, something
rather difficult with triodes with their NFB and instablity at RF due to phase
shift
and the Miller effect.



Arguments that the intrinsic characteristics of devices result
from "feedback" are exactly as nebulous as arguments that
claim that feedback is inherently evil.

Triodes sound their best when their signal current barely changes
which is where their internal applied NFB is greatest.

There is no need for external loops of NFB in such circumstances,
although were one to use a pentode for anything one is normally forced into
using loops of NFB because there isn't much NFB within a pentode.

Patrick Turner.



Hoping all's well with you and yours, and with the newsgroup,

Chris Hornbeck
"This has been an account for those who don't keep them"
J-LG, _Tout Va Bien_ 1972

[Andre Jute]

A lotta big snippage; take it for granted that I'm standing shoulder to
shoulder with Patrick except for some very minor kibbitzing.

Patrick Turner replied (to a post by Chris Hornbeck in reply to one by
Ian Iveson, summarized in the title of the thread):

I think you are missing the point.
If you set up a 12AX7 triode in a normal common cathode circuit with gain = 70
and Miller C = 120pF, and feed the grid signal from a signal source of 4Mohms
source resistance,
then the Miller effect will give the circuit a low pass filter character with pole
determined by the 4M and 120pF RC time constant.

This would depend on the allowance you make for strays. That classical
allowance of about 30pF assumed that the guy doing the wiring had the
right tools and components and all the time in the world to make a tidy
job (it was usually done by the appie, who was told to do it over until
he got it right). For your modern DIYer the stray capacitance can
easily run to several times the classical allowance. It is one of the
reasons I like a lot of current on my drivers -- then whoever builds
the design cannot go wrong.

Also, in this thread it has been said (not by you) that Miller works
only at HF. That's nonsense, either ignorant or provocative; Miller
works well down into the presence band, another reason for dumping all
those little zero-current 12AX7 drivers and crushing them underfoot
before binning the shards.

Triodes sound their best when their signal current barely changes
which is where their internal applied NFB is greatest.

I've been saying this for years; it is the subtext to my HVHCHL mantra,
that high voltage and high current and especially high load (I don't
even blink before I speciify 5K6 on SE 300B) kills noise.

There is no need for external loops of NFB in such circumstances,
although were one to use a pentode for anything one is normally forced into
using loops of NFB because there isn't much NFB within a pentode.

The more I hear these futile arguments, the more convinced I am that
we're putting apples and oranges in the same basket. Small ZNFB SE amps
and big pentode/beam PP amps with NFB serve different purposes and are
*necessarily* designed to different parameters which then meet the
different tastes of different customers.

..Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"wonderfully well written and reasoned information
for the tube audio constructor"
John Broskie TubeCAD & GlassWare
"an unbelievably comprehensive web site
containing vital gems of wisdom"
Stuart Perry Hi-Fi News & Record Review

[Andre Jute]

Chris Hornbeck wrote:

Welcome back, Chris.

Arguments that the intrinsic characteristics of devices result
from "feedback" are exactly as nebulous

If you observe that the characteristics of NFB are in play around an
isolated component under certain conditions, then the components
otherwise deliberately used to arrange NFB must be inside the visible,
real component. It's a black box argument, already well fleshed out
with hard, incontestible numbers by Patrick.

exactly as nebulous as arguments that
claim that feedback is inherently evil.

If you're pointing a finger at me, four of your own fingers point back
at you to accuse you of humourlessness. I have a provocative sig that
calls NFB "undesirable" but even the choleric Wilson doesn't fall for
it any more. As for the article on my SET site describing the monkey on
the back of the monkey that is NFB, that relates to a particular
topology and way of thinking; it is also signalled to be amusing by
overstatement via the joke heading the entire article. But I don't
claim NFB is inherently evil: I use it as appropriate to topology and
taste; I merely avoid devices so inherently poor that they *depend* on
megadecibels of NFB to work at all, never mind pleasingly: a kludge is
a kludge is a kludge regardless (or perhaps because) of the fact that
it provides a living to millions of engineers. And I gladly use NFB
elsewhere, even in real life; who nowadays drives a car without power
steering? It amuses me that the "engineers" who arrive here hotfoot to
announce that they're "on Jute's case" (like your pal Poopie) never
grasp that I apply negative feedback to them: the harder they are on my
case, the more they hurt, until they break; NFB is such a fundamental
tenet of their mullahcracy, it is an easy whip to reach for.

Hoping all's well with you and yours, and with the newsgroup,

We missed you.

Chris Hornbeck
"This has been an account for those who don't keep them"
J-LG, _Tout Va Bien_ 1972

Andre Jute
Our legislators managed to criminalize fox-hunting and smoking; when
they will get off their collective fat arse and criminalize negative
feedback? It is clearly consumed only by undesirables.

[Chris Hornbeck]

On 14 Aug 2006 17:40:47 -0700, "Andre Jute" wrote:

If you observe that the characteristics of NFB are in play around an
isolated component under certain conditions, then the components
otherwise deliberately used to arrange NFB must be inside the visible,
real component. It's a black box argument, already well fleshed out
with hard, incontestible numbers by Patrick.

A triode is a three terminal device. If you cannot observe
something on its terminals, that something does not exist.

The argument working backward from multigrid devices is
seductive, but false. It's a useful analogy, but flawed
as a model. You'll never, for example, hear anyone trying to
make the same case for junction FET's, whose pinchoff region
extends through both source and drain. "Electrostatic effects"
anyone?

FWIW, the problem with the multigrid analogy is a common
one in many fields, working backward from known results to
reach false conclusions about how those results obtain.

If you doubt me, do a simple test: heat up a triode's cathode,
apply a typical amount of fixed negative voltage to its
grid through a large resistor, apply varying typical voltages
to its anode, and observe grid voltage.

If the grid's voltage varies, feedback has occurred. If not,
handwaving has occurred. QED (quick and easy demonstration).

All the best,

Chris Hornbeck
"Then there is the competition level: blasters, flutterblasters, and
double flutterblasters.

Oh. we were talking about micing classical guitar."
-Willie K. Yee, MD

[Phil]

Patrick Turner wrote:

[snip]


Since you maintain there is no NFB in a triode, could you
point out your detailed logic to proove why not?

Feedback exists IF AND ONLY IF, you can change the input voltage, i.e.,
control voltage, by changing the output voltage.


This is the case with any amplifier with a standard shunt voltage negative feedback
network.
Imagine a high gain pentode with a 10k input series R between the signal input and grid.
Then imagine a 100k resistor between grid and anode, and imagine the open gain of the
pentode
is above 1,000.
The closed loop gain will be about -10, because +1V input is amplified to become -10V
output.
Any distortion, noise, or load change voltage appearing at the anode is divided to 1/11
of voltage at the anode and applioed to the grid and thus opposes its own production.
The anode signal does not cause a change to the source input signal but causes a cjange
to the grid signal
which is remote from the input source because of the 10k R.


In the case of the triode, the input souce signal is connected to the grid which at AF is
always a high
Z input port to the amp.
But extending from the grid is the electrostatic arm of the grid, and extending from the
anode
there is the electrostatic arm of the anode, and their NET value is the actual resultant
field effect that causes the change in electron flow from cathode to anode.

The net electrostatic value acting on the electron stream is the REAL or equivalent
input point as described for the case above with pentode and simple resistor divider
shunt NFB network.

Yes, but the coulombs needed to change this electrostatic field from one
value to another give you the effective grid-plate capacitance. So this
"feedback" is merely the Miller capacitance. What I think you are
missing is that a triode can be modelled as a pentode in parallel with a
resistor from plate to ground. This parallel resistance reduces the
impedance seen by any load, but it is not the same as using feedback.
You could never use this parallel resistor to make an oscillator, and it
does not transform amplitude distortions into phase distortions, which
is something that Matti Otala proved many years ago occurs in ALL
negative feedback circuits. Again, EXCEPT for the Miller capacitance, a
triode has no feedback, and does not obtain its low output impedance
from feedback. To express this from another point of view, do you
actually believe that it is a pentode's low grid-plate capacitance that
is responsible for its high output impedance? The same mechanism that
reduces its capacitance increases its impedance, but the low capacitance
is a CORRELATION of the high impedance, not a CAUSATION. In other words,
the mechanism that CAUSES low capacitance also CAUSES high impedance.
You could add some extra capacitance from the plate to the grid, and
guess what? That's right, the impedance would remain as high as ever,
until you get way higher than the audio frequencies we are discussing
and into the range where Miller capacitance starts to become effective,
which does indeed start to give triodes a SIGNIFICANT negative feedback.


In a triode, this
occurs only at frequencies where the Miller capacitance starts to take
effect.


The Miller effect is present at even the lowest F but its effect on circuit gain only
becomes an issue when the Miller C impedance reduces and a divider between
series source resistance and the Miller C form a low pass filter.
In audio preamps using say a 12AX7 the Miller C may be effectively 120pF and
so the high audio F are seriously attenuated when the source R value of input to the grid
is say
from another high R anode signal.
The 120pF of miller C has an enormous Z at 50Hz, so Miller effects at LF are negligible.

But the electrostatic FB acts at all F.

As stated above, your "electrostatic FB" IS Miller capacitance, and yes,
it works at all frequencies, but so? At lower frequencies, this has an
insignificant effect, and yet the output impedance is as low as ever,
clearly not CAUSED by this infinitesimal effect.


Note that in all standard solid state devices, the ONLY way to
achieve low output impedance is by using feedback.


Agreed.


And yes, if you
change the output voltage of a cathode/emitter/source follower, you
change the cathode-grid voltage, so these are feedback circuits (which
is why it is sometimes too easy to make oscillators out of them!). To my
knowledge, only power J-Fets, also known as Static Induction
Transistors, are exceptions, and they have curves that look EXACTLY like
a triode's except with a "plate" impedance (drain impedance) of 2 ohms,
instead of 200 to 20,000 ohms, like you have in a tube.


Nearly all SS devices WITHOUT applied NFB have collector or drain resistance curves that
have no resemblance to triodes, but a close resemblance to a set of pentode Ra lines.

Yep. EXCEPT for power J-Fets/static induction transistors. Let me know
if you "have to see it to believe it," and I'll either send you a photo
of the drain curves for one of these devices or post it where you can
see it. You will be amazed ...

After series/shunt NFB is applied, the Ra lines or collector/drain lines resemble those
of a triode.
The data sheets for all thse devices are given without any NFB applied.
Where series current NFB is applied to a triode or other device the slope of the devices
output
Ra, Rc, Rs become more horizontal than they were to begin with without NFB applied.





Also note that whereas a triode can use a transformer to reduce its
"natural," feedbackless output impedance down to 1 or 2 ohms, not even a
transformer can save standard solid state devices, because in order to
use a transformer, the output signal itself must have a high voltage to
current ratio, which is not the case for solid states. This "voltage to
current ratio impedance" is different from the output impedance. For
example, a pentode has both a high voltage to current ratio, and a high
output impedance, but if you triode-strap it, its output impedance
drops, even though its voltage to current ratio stays the same.


The pentode such as EL34 has Ra = 12k.
So a 6k to 6 ohm OPT will transform the Ra from 12k to 12k/1,000 = 12 ohms at the
secondary.
This Rout at the sec is way too high to drive speakers to give a wanted flat speaker
response
so some folks would just connect the screen to the anode to allow the anode voltage to
deliver the full possible amount of internal electrostatic NFB.

Truly, I hope you quit saying this, because it really is silly, and your
other posts are not silly. The awesome Miller capacitance CONTROLS the
output at 20 Hz, where at 2 gigaohms (4 pF), it overwhelms the 10 Kohms
output impedance of the previous stage, giving the triode a plate
impedance of 600 ohms! WHAT-ever ...

Phil

The result is that the Ra drops to about 1,200 ohms and the Ra at the sec becomes 1.2
ohms
plus whatever OPT winding resistances exist.

Effectively a pentode is a current source where gain = gm x RL approx.
With the NFB of the triode applied it becomes substantially a voltage source
where gain = µ x RL / ( RL + Ra ).
This formula is valid for pentodes but because Ra is so high, the formula can be
simplified to A = gm x RL, and the pentode µ which is very high and somewhat variable
need not
be considered. The µ of the triode and pentode are considered very different amp factors,

and are each the product of gm x Ra, where gm and Ra are for either triode and pentode.
The triode NFB in an EL34 reduces the µ from about 130 in pentode to about 10 in triode.

The term current source describes signal sources where Rout is much higher than RL,
and voltage sources describes where Rout is much less than RL.

The UL connection is between the two extremes and where the partial anode signal voltage
applied to the
screen upon the electron stream gives Ra approx = RL.

Patrick Turner.







Phil

Patrick Turner.



cheers, Ian

[Patrick Turner]

Andre Jute wrote:

A lotta big snippage; take it for granted that I'm standing shoulder to
shoulder with Patrick except for some very minor kibbitzing.

Patrick Turner replied (to a post by Chris Hornbeck in reply to one by
Ian Iveson, summarized in the title of the thread):

I think you are missing the point.
If you set up a 12AX7 triode in a normal common cathode circuit with gain = 70
and Miller C = 120pF, and feed the grid signal from a signal source of 4Mohms
source resistance,
then the Miller effect will give the circuit a low pass filter character with pole
determined by the 4M and 120pF RC time constant.

This would depend on the allowance you make for strays.

Well Andre, I try to keep things simple at first and leave the stray C out of trh
picture.
As more knowledge is gained, readers would be introduced to extra ad on info
to complete the detail of the total picture.

That classical
allowance of about 30pF assumed that the guy doing the wiring had the
right tools and components and all the time in the world to make a tidy
job (it was usually done by the appie, who was told to do it over until
he got it right).

30pF is a bit on the high side for Cg-k, and Cgrid to whatever else.
30pF would not seriously affect what i said.

For your modern DIYer the stray capacitance can
easily run to several times the classical allowance.

Only if the diyer was a complete dolt.
Its actually rare to find more than 30pF stray C at the input of any average 12AX7
stage.
One could have an interconnect cable a metre long between the anode circuit of a preamp
and a power amp
and this could be an unwanted low pass filter.
Solution? place the preamp stage on the same chassis as the power amp to remove the need
for
a cable, two plug/socket joins, and a cathode follower buffer.

With two mono amps this can be a problem for providing attenuation
but not if the attenuator is a 10k pot in a box fed from the low Rout of a CD player or
phono stage with µ
follower or CF output.


It is one of the
reasons I like a lot of current on my drivers -- then whoever builds
the design cannot go wrong.

I used to think as long as the open loop bw without compensations/phase tweakings
of the driver and output stage at full voltage needed for clipping is 3Hz to 100kHz,
then all is well.
But better is to have more driver current than is needed to fulfill the above condition.

It simply sounds more right, and dynamic.
I like using 6V6/6BQ5 in tride as drivers instead of a lone 1/2 of a 6SN7....
Never would I use a 12AX7 as a driver in any amp, unless it was a guitar amp
and then it fattens the sound, which tastes good to muso ears.
For hi-fi I want panache, and lowish µ triode drivers do it for me best.
Nevertheless, a recent tweak of a Quad II amp showed me how good
EF86 pentodes could sound as drivers......



Also, in this thread it has been said (not by you) that Miller works
only at HF. That's nonsense, either ignorant or provocative; Miller
works well down into the presence band, another reason for dumping all
those little zero-current 12AX7 drivers and crushing them underfoot
before binning the shards.

I described the effects of the Miller effects with 12AX7 and with a high source
impedance to the grid.
The 120pF of Miller C compared to a very high source R causes the pole for the
RC filter to be well below 20kHz.
But where a 12AX7 input tube used in a power amp is fed from a cathode follower of a
line stage amp
with source R = 600 ohms, then the interconnect cable C and Miller C and stray C does
not cause
any loss of detail below 50kHz. The input voltage to a power amp is below a volt and so
current
requirements to charge or discharge Cin of an average power amp are very low; a few mA
is plenty of Ia for the CF line stage output.




Triodes sound their best when their signal current barely changes
which is where their internal applied NFB is greatest.

I've been saying this for years; it is the subtext to my HVHCHL mantra,
that high voltage and high current and especially high load (I don't
even blink before I speciify 5K6 on SE 300B) kills noise.

There is no need for external loops of NFB in such circumstances,
although were one to use a pentode for anything one is normally forced into
using loops of NFB because there isn't much NFB within a pentode.

The more I hear these futile arguments, the more convinced I am that
we're putting apples and oranges in the same basket. Small ZNFB SE amps
and big pentode/beam PP amps with NFB serve different purposes and are
*necessarily* designed to different parameters which then meet the
different tastes of different customers.

Well, beam tetrodes and pentodes in class A power output stages can sound OK.
FB MUST be used because Rout is hopelessly high.

Magnificent recordings have been made in the past using studio gear based around
multiple EF86 and other pentodes with NFB, and hardly a triode to be seen.
Quad II is based around beam tubes driven by EF86.
Is this combo poor? The jury may still be out after 50 years.
One can make Quad II sound better with PS improvements and
the operation of the input pair of EF86 as a true long tail pair, not a
floating paraphase inverter.
I've also triode using an all triode line up with Quad II with PS improvements.

If I had a pair of horns I'd not want more than a pair of SE 2A3 amps, they are plenty,
and using a six pack of 6550 is just showing off.
I've seen guys using 96dB sensitive JBL studio monitors with 2 x 15" basses and bipolar
"bum" shaped horn tweeter in a 4M x 4M room and biamp driven with two Yamaha 2200
or capacity for 880 watts where only 1 watt total is ever likely.

The guy likes "muscle" in his amp. He also worked out at the gym to keep himself
looking good, or dangerous, but to what purpose was all this muscle put towards
apart from showing off?
I mentioned SET use and he didn't bite; to him that would be accepting a wimp
to his stable of muscle.......

For those who want to know more about what i said a couple of years ago about
NFB in triodes they could do worse to search the archives under NFB,triodes, and my
name.

Patrick Turner.






.Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"wonderfully well written and reasoned information
for the tube audio constructor"
John Broskie TubeCAD & GlassWare
"an unbelievably comprehensive web site
containing vital gems of wisdom"
Stuart Perry Hi-Fi News & Record Review

[Patrick Turner]

Chris Hornbeck wrote:

On 14 Aug 2006 17:40:47 -0700, "Andre Jute" wrote:

If you observe that the characteristics of NFB are in play around an
isolated component under certain conditions, then the components
otherwise deliberately used to arrange NFB must be inside the visible,
real component. It's a black box argument, already well fleshed out
with hard, incontestible numbers by Patrick.

A triode is a three terminal device. If you cannot observe
something on its terminals, that something does not exist.

With all due respect, this line of reasoning does not stand up to scrutiny.

Many many stars and other things in the observable universe and particles
in quantum mechanics are not so easily
observable but have been calculated as being there, and have then been
later observed, and some things
unobservable are known to be there because of the effects thay cause to the
observable around them,
such as black holes or perhaps strings in sub-atomic theory.

With 3 terminals, the effects or any TWO can have a net effect on a flow of
electrons between
two terminals. To say because we cannot observe the net effect is
balderdash because
we COULD PROOVE the net effect occurs by means of observation of several
gain measurements and related voltage measurements.
I am not a maths guru and I don't pretend to know more than all the famous
electronic theoreticians who
have more intelligence and university education than I do; Professor Child
is one.

In a simple engineer's world where practical people like myself are faced
with making a working amp
I need not always consider the model of a triode including its
electrostatic NFB internal loop.
I need only know the closed loop model where Ra, and gm and µ are nominated
for the
triode, ie, after the feedback is applied.

The triode is a self contained voltage generator with its own internal NFB
loop
which ned not be mentioned or considered in detail when designing any amp
with a triode.

There are other models of a triode ( or pentode )such as the equivalent
generator model where the
tube is considered a low output voltage generator producing µ x Vg output
volts, and Ra is a series
R between the gene output and anode terminal, with the other two terminals
being
cathode and grid.
The NFB could be drawn up into such a model but there is no need.
MOre about that is at my website.

The amount of operational NFB can be established in a triode when gain is
known
and µ and gm and Ra. The equations are all a bit tedious, but much better
brains than I have
did proove there is NFB in a triode well before WW2.



The argument working backward from multigrid devices is
seductive, but false. It's a useful analogy, but flawed
as a model. You'll never, for example, hear anyone trying to
make the same case for junction FET's, whose pinchoff region
extends through both source and drain. "Electrostatic effects"
anyone?

The voltage that causes the fet pinch off is applied as a voltage field
effect as opposed to a base current flow.

There is no need to revert to multigrids to support the idea of triode NFB
existance.





FWIW, the problem with the multigrid analogy is a common
one in many fields, working backward from known results to
reach false conclusions about how those results obtain.

If you doubt me, do a simple test: heat up a triode's cathode,
apply a typical amount of fixed negative voltage to its
grid through a large resistor, apply varying typical voltages
to its anode, and observe grid voltage.

It depends how big the bias resistor is.
A value of 10M will generate a volt of two of negative grid bias which
changes little if the
anode voltage is changed.
But with bias R = 100k, and negligible grid current biasing, and with the
grid kept -ve with respect
to the cathode, then what changes do you see with Ea changes applied?
Not much change at all.
But when the grid voltage changes say +ve 1V, and the anode voltage say
changes -60V for a typical 12AX7,
then there are TWO electrode voltage changes which control the Ia, not just
the grid's action.

If we were to place a mesh screen between anode and grid and keep that at
say +100V
so that any anode V change is NOT FELT BY THE ELECTRON STREAM BETWEEN
CAHTODE AND SCREEN,
then the Ia change will be determined solely by the grid V change.
The anode voltage change OPPOSES the action of the grid voltage, since they
are 180 degrees
opposite phase.
With the screen if inserted, the gain of a 12AX7 would be vastly boosted,
indeed it would become a tetrode,
and the gain may be 1,000 instead of just 60 depending on loading.
Because of the NFB in the 12AX7, the gain cannot ever exceed the triode µ
of the tube.





If the grid's voltage varies, feedback has occurred. If not,
handwaving has occurred. QED (quick and easy demonstration).

Sorry, but many things exist as intangibles.
I cannot see your logic in denying that mutually combined grid and anode
voltage changes
will sum to form a net effect to control Ia, and hence Va where a load is
used.

Consider that we have a low impedance voltage souce adjustable for between
+100V and +200V
connected to a 12AX7 anode. Consider the grid also connected to a low
impedance voltage source
adjustable between -1.5V and -0.5V.

Now we can randomly set either voltage of the anode and grid to any
voltages we wanted.
Ia could be measured through a 1 ohm R in the anode circuit.
The Ia will be the result of the combined action in terms of anode gm and
grid gm.
The maths wouldn't be too hard to work out, but after recording several
values of Ia for several settings for Ea and Eg, algebra could be used to
work out the
various gm of the active electrodes, and the net effect on the Ia prdicted
for
any new values of Ea and Eg. And this would all proove what exists within a
triode which
you fail to mention or seem to grasp,
ie, that the voltages of anode and grid combine to give a NET controlling
action on the
Ia, and in fact the Ia is controlled BY A VIRTUAL GRID within the triode.

Virtual input ports in opamp use occur wher you have a two R network for
plain negative shunt FB.
Of the two input ports of the opamp, one is grounded ( like the cathode of
a triode ), and the other
is the live input, but in fact is a "virtual earth" because such a tiny
signal exists there since
the larger in and out signal are at the two ends of the two shunt NFB R.
There is a virtual earth input port in a triode and it is the net
electrostatic field created by
the two fields of the anode and grid.

Rather than complicate the use of triodes by continually giving ther anode
gm and grid gm
as well as a complex formula, for the 12AX7 the µ, gm, and Ra are the
resultant parameters
after FB has been applied, and this leads to easy design
with the two main formulae,
µ = gm x Ra,
and Voltage Gain = µ x RL / ( Ra + RL ).

For a pentode the Ra changes very variably for ia values and also gm.
µ for a pentode is rarely ever given in the data because the product
of gm x Ra varies so much.
Since pentode loads are usually much less than the high Ra of the pentode
then their VG = approx gm x RL.
Where the RL is a series triode set up to be a high impedance active load
the gain
of the pentode approaches the product gm x Ra, ie, pentode µ.
With a load approaching a CCS, the pentode gain becomes huge,
and somehwhat prone to shunt capacitance and inductance so the gain profile
with F
is an arched graph, not a flat line like a triode.
Simple shunt FB can be applied around a pentode high gain unit to make it
behave almost identically to the triode
and so it displays the same triode like Ra lines for given Eg values low
thd, and low Ra like a triode's.

This external shunt NFB looping has been done in countless line stage amps
to set the gain
at between say 20 and unity and where the open loop gain at 1 kHz may be
200 or greater.
In many phono stages, NFB is applied for RIAA eq around a single pentode
stage, eg, Quad 22.
This is more difficult to do with a single triode because the triode
already has NFB within
so gain has already been reduced by internal NFB and we cannot reach inside
the triode to connect up
some kind of frequency selective electrostatic NFB.
So many phono amps used a pair of 12AX7 with RIAA NFB series voltage NFB.
The best samples of these use a CF buffer at the output to drastcally
reduce Rout to a few ohms at
1 kHz, and to prevent the NFB network from loading the triode and reducing
the open loop gain.
The same could be done with a pentode where one has a CCS load for ther
pentode but a direct coupled CF output buffer, so that open loop gain of
about 1,500 could be achieved easily, and so that the shunt
FB input R becomes the loading for the cart, say 47k, and the RIAA FB
network is between the pentode
grid and its CF triode buffer output.

I digress, but triodes and pentodes and j-fets such as the 2SK369 have
their
advantages and dis-advantages, and I leave you all to try things yourselves
to find out which sounds best,
and I would advise there is more than one way to build a fabulous
amplifier.

Do not insist on being too simplistic; that is the unforgiving stance of
the shortsighted fanatic.
Always look for things that lie beneath the surface.

Patrick Turner.






All the best,

Chris Hornbeck
"Then there is the competition level: blasters, flutterblasters, and
double flutterblasters.

Oh. we were talking about micing classical guitar."
-Willie K. Yee, MD

[Chris Hornbeck]

On Tue, 15 Aug 2006 13:43:12 GMT, Patrick Turner
wrote:

snipped for sanity

Always look for things that lie beneath the surface.

Like Black Holes and Virtual Grids? Hint: a triode has
a real grid.

Suppose I were to try to sell you on the idea that
multigrid valves have *positive feedback*, and used
exactly your same arguments? And that I then invoked
the Black Hole defense when you pointed out to me
that my argument is Swiss cheese? I'd guess that you
might be, what's the polite word?, sceptical?

You're a smart guy and may believe as you wish, but
handwaving is completely unconvincing to me.

All the best,

Chris Hornbeck
"Then there is the competition level: blasters, flutterblasters, and
double flutterblasters.

Oh. we were talking about micing classical guitar."
-Willie K. Yee, MD

[Chris Hornbeck]

On Sun, 13 Aug 2006 08:43:54 GMT, "Ian Iveson"
wrote:

Since you are a leading exponent of naive realism in audio, I shouldn't be
surprised to find you lost in metaphysics.

A very simple test can confirm or deny the argument.
Vary anode voltage over typical linear operating regions
WRT cathode while biasing grid at an appropriate negative
voltage WRT cathode through high resistance.

Does grid voltage vary? "Feedback" requires that
voltage or current from an output terminal influences
voltage or current at an input terminal.

It really and truely ain't rocket surgery.

Much thanks, as always,

Chris Hornbeck
"Then there is the competition level: blasters, flutterblasters, and
double flutterblasters.

Oh. we were talking about micing classical guitar."
-Willie K. Yee, MD

[Chris Hornbeck]

On Tue, 15 Aug 2006 22:34:13 -0500, flipper wrote:

snipped for bandwidth

Two words:

No, on second thought, I won't.

All good fortune,

Chris Hornbeck
"Oh, life is a glorious cycle of song,
A medley of extemporanea;
And love is a thing that can never go wrong;
and I am Marie of Roumania."
- Dorothy Parker

[Chris Hornbeck]

On Tue, 15 Aug 2006 22:58:45 -0500, flipper wrote:

Yes, well. fact is I've built it, both with ("Looking Glass" amp) and
without (PC Speaker Hybrid and "Current Loop" amps) the series plate
resistor.

You're looking at the wrong input. The (voltage) feedback input is the
stem.

The banana's native output is plate current; which is determined by
two voltage input signals, peel and stem.

A practical non varying plate circuit is to operate the banana's stem
into a PNP current mirror (as in my "PC Speaker Tube/Bipolar current
mirror hybrid amplifier" amp design). You can get gobs of gain but the
linearizing effect of stem feedback is lost. (Of course, in the amp
design there's global cathode feedback)

Insert a series plate banana in the same circuit, to create the
voltage feedback signal, and you get precisely what classical feedback
theory says: reduced gain and reduced distortion.

Chris Hornbeck
"I expect a black silence that is almost as violent as laughter."
Jean Cocteau, 1932

[Chris Hornbeck]

And the missing words are "Whack Job."

But meant in the best possible way, this being simply
a technical discussion. Where *do* these theories
originate? Is it a result of all the Herbal Fragrances
of are we not all wearing our tin-foil hats at night?

Hard to guess, I'd guess,

Chris Hornbeck
"I expect a black silence that is almost as violent as laughter."
- Jean Cocteau, 1932

[Patrick Turner]

Chris Hornbeck wrote:

On Tue, 15 Aug 2006 13:43:12 GMT, Patrick Turner
wrote:

snipped for sanity

Always look for things that lie beneath the surface.

Like Black Holes and Virtual Grids? Hint: a triode has
a real grid.

And a triode has a net field effect, a virtual grid within itself.

Sure it has a real grid.

But the anode controls the electron stream in conjunction with the grid.

The feedback path with all triodes is the electrostatic field effect
from anode to
electron stream.



Suppose I were to try to sell you on the idea that
multigrid valves have *positive feedback*, and used
exactly your same arguments?

I smell a large rotten red herring.
Multigrids don't have positive FB.
They simply have very much reduced NFB
because of the screen which is at a fixed voltage.

But if you build a UL amp one can reverse the screen connections
from the OPT to send PFB to the screens rather than the conventional
NFB.
This usually causes violent oscillations in a UL amp with 43% taps,
but where the taps are a lower % the PFB boosts the output stage gain
and thus makes the applied global NFB more effective, so that
12dB of gain boost means the NFB is 12dB more effective.
Its a real fudge of a trick this one and nobody does PFB in any
tube amps these days.

And that I then invoked
the Black Hole defense when you pointed out to me
that my argument is Swiss cheese? I'd guess that you
might be, what's the polite word?, sceptical?

I definately am sceptical.



You're a smart guy and may believe as you wish, but
handwaving is completely unconvincing to me.

There are those I may convert to the faith, and heathens who will
always remain pagan.
I loose no sleep.

I really do suggest you read a little more on what greater minds than
mine have written about
NFB in triodes.

Please don't equate me to a flamboozler. I merely relate the world to
you through
my experiences.

Patrick Turner





All the best,

Chris Hornbeck
"Then there is the competition level: blasters, flutterblasters, and
double flutterblasters.

Oh. we were talking about micing classical guitar."
-Willie K. Yee, MD