On Fri, 17 Dec 2004 00:48:51 +0000, Ian Iveson wrote:

"John Byrns" wrote in message
...

Hi Ian,

I have two problems with your view of "NFB".

More than two, in my view.

First were is it written that in order to have "NFB" you must sum
voltages?

The control input to a triode is a voltage.

snip

You are perhaps getting confused here.

The voltage is applied to the grid of a triode in order to create an
electrostatic charge on the grid, relative to the cathode.

The actual device controlling electron flow is the field around the grid.
It is thus correct to state that the feedback mechanism can be either
internal (directly summated via the field around the grid) or external
(via a voltage which will be used to change the field around the grid).

Note that the voltage *itself* is *not* a controlling device in its own
right, merely a method of maintaining it. Granted that the field cannot
exist without a voltage, but this is fact.

--
Mick
(no M$ software on here... :-) )
Web: http://www.nascom.info
Web: http://projectedsound.tk

On Fri, 17 Dec 2004 21:31:33 +1100, Patrick Turner wrote:

snip
or even
as linear

as a 6L6 set up SE, no FB, and in beam tetrode mode.

snip


grin - tongue unjammed out of cheek yet Patrick? :-)

--
Mick
(no M$ software on here... :-) )
Web: http://www.nascom.info
Web: http://projectedsound.tk

"Stewart Pinkerton" wrote in message
...
: On Fri, 17 Dec 2004 01:52:17 GMT, "Ian Iveson"
: wrote:
:
: "Patrick Turner" wrote
:
: Ah, but greater minds than yours or mine suggest there **is**
: NFB in a triode.
:
: Maybe greater than your mind. If any of them claim what you are
: claiming, their minds are definitely not greater than mine. There is
: dynamic equilibrium in a triode, that some may loosely, maliciously,
: dishonestly, or foolishly call NFB. I have accepted all along that
: this loose definition has currency, as you will see if you read my
: posts.
:
: If there is not an internal feedback loop in a triode, how is it that
: breaking the loop with a screen increases the gain?
:
: --
:
: Stewart Pinkerton | Music is Art - Audio is Engineering

Well, I'd not want to call it a loop, doesn't get much more
instantaneous then field addition in a few cm of space, does it ?

Rudy

"Stewart Pinkerton" wrote

If there is not an internal feedback loop in a triode, how is it
that
breaking the loop with a screen increases the gain?

By altering the conditions of dynamic equilibrium. For a control
system engineer, direction is defined as from controlling input to
controlled output. Hence an engineer should know the difference
between feed-forward and feedback. This is the point that students
can get hung up on.

The common mistake is to confuse this sense of direction with
another, such as the alignment of fields, or the flow of electrons.

Incidentally, your view of the difference between a tetrode and a
triode is misleadingly simplistic. The screen does not merely break
the "loop", but is an active element in its own right.

Ian

On Fri, 17 Dec 2004 17:25:58 +0100, "Ruud Broens"
wrote:


"Stewart Pinkerton" wrote in message
.. .
: On Fri, 17 Dec 2004 01:52:17 GMT, "Ian Iveson"
: wrote:
:
: "Patrick Turner" wrote
:
: Ah, but greater minds than yours or mine suggest there **is**
: NFB in a triode.
:
: Maybe greater than your mind. If any of them claim what you are
: claiming, their minds are definitely not greater than mine. There is
: dynamic equilibrium in a triode, that some may loosely, maliciously,
: dishonestly, or foolishly call NFB. I have accepted all along that
: this loose definition has currency, as you will see if you read my
: posts.
:
: If there is not an internal feedback loop in a triode, how is it that
: breaking the loop with a screen increases the gain?

Well, I'd not want to call it a loop, doesn't get much more
instantaneous then field addition in a few cm of space, does it ?

Not the point. If a pentode is simply a triode with screens preventing
the effect, why not call it feedback - because that's what it is.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

On Fri, 17 Dec 2004 15:41:40 GMT, mick wrote:

On Fri, 17 Dec 2004 21:31:33 +1100, Patrick Turner wrote:

snip
or even
as linear

as a 6L6 set up SE, no FB, and in beam tetrode mode.

snip

grin - tongue unjammed out of cheek yet Patrick? :-)

I know where he's going with that argument, but I don't see why
putting together a good-performing SS amp with a very low parts count
is 'cheating', just because it's not implemented in the same way that
you would use a 300B. I try to avoid using my chisels as if they were
screwdrivers!

I guess it depends what you're trying to prove with a KISS design,
actual simplicity, or forcing a SS copy of how you'd build a SE 300B
amp. It's not like two-stage SE 6SN7/300B amps are exactly
ground-breaking 'gee whizz, how did he do that' technology, so I'm not
really sure what the point of Andre's amp is in any case.............
--

Stewart Pinkerton | Music is Art - Audio is Engineering

On Fri, 17 Dec 2004 21:31:33 +1100, Patrick Turner
wrote:

Stewart Pinkerton wrote:

On Thu, 16 Dec 2004 15:00:59 -0600, (John Byrns) wrote:

Which rules have I changed? You may not agree with me that triodes
incorporate an internal negative feedback loop, but that doesn't mean I
have changed any rules, we just don't agree on the rules as they relate to
the operation of triodes, I might just as well accuse you of changing the
rules regardless of reality.

Apologies due, it was not *your* position with which I was in
conflict, but Ivesons. Which is unfortunate, since he and I are both
C&I engineers, but he doesn't seem to understand what goes on inside a
triode. I do agree that triodes have a large amount of internal
feedback (hence their low gain, and hence the pentode, which breaks
the loop). My position is that certain transistors, such as the Sanken
2SC2922/2SA1216 pair, are as linear as triodes when used correctly,
and can be employed to produce low parts count 'zero NFB' amps in
similar manner.

Well all you have to do is proove your position to be valid.

Please set up a single transistor in common emmitter mode, ie, 8 ohm load in
the
collector circuit, idle current about 2.5 amps, grounded emitter.

The input is applied to the base from a very low impedance source.
with less than 0.001% thd.
This will require you to use a low thd amp with low Ro, because the base
input impedance is very low
Then measure the thd products and draw a graph of the level of each
harmonic product as it increases from 0.0 watts to
a maximum of about 10 watts.

Let us know how you get on.

But not before you have done the above experiment will any of us know if the
bjts you have chosen are anywhere near as linear as a triode, or even as linear
as a 6L6 set up SE, no FB, and in beam tetrode mode.

Very amusing, but in fact what I would do to prove the point is set up
the above circuit as you suggest, *plus* an emitter resister which
sets the stage gain to the same as that of a 'naked' 300B, say about
0.47 ohms or so. I'd also be running the stage with a 3 amp bias
current, so it would indeed need quite a beefy driver, able to put out
a nice linear 60 mA or so.

As previously discussed, the use of local degeneration does not in my
view count as loop feedback, as it does not to my knowledge result in
the kind of discordant intermodulation products previously noted to be
produced by global NFB loops under certain conditions.

But the emitter follower topology is a case of a high amount of NFB, and
its use completely negates the purpose of you quest, to proove that a bjt is as
naturally linear
as a triode.

Different point being made here, which is how to achieve best
linearity with minimum parts count, using SS rather than tube
technology. You don't design a sailboat the same way you design a
motor boat.

If we were to accept you premise that EF topology is OK, then
we would then request that the 300B would also be allowed to have 30 dB of
applied global NFB around a second gain tube to make the playing field level
for the comparison.

Ah, but that would increase the circuit complexity, thereby negating
the KISS principle.

Hence, this
should not be a problem for the 'zero NFB' purist.

It remains a real problem for everyone except you, and you cannot
accept that modern transistors are woefully non linear unless
large amounts of NFB are employed, and then there is the problem
of overcoming the naturally low base input impedance of the bjt.

The low base impedance is indeed a problem, but then you have the high
gate capacitance of MOSFETs to consider, if you go that route.

Similarly, I don't
agree that cathode/emitter followers are the spawn of Satan, they seem
simply to be an excellent impedance transformer. Is there any
*logical* objection to them?

Some have no problem with follower stages. I don't have a problem with them
where I use them for the output of a preamp for example.
But in this case where you are floundering while you try to design a simple
bjt amp which is as simple as a 300B amp, follower topology is plainly
quite unacceptable.

I'm not 'floundering', I'm just kicking some tyres.

But your aim and zeal is to proove bjts can be used as easily as a 300B and its
simple circuitry,
and with as little FB, zero feedback in fact, and still end up with a simple
amp, two devices max,
and have thd lower than 300B, and still have Rin above 10k, and able to be
powered from a
volume pot straight after a CD player..
So, unless you wish to be seen by all as the universe's greatest cheat, please
do not
keep insisting on the use of emitter follower in this case in this thread is
OK.

Depends if you consider the use of a CCS load instead of an OPT to be
'cheating'. I've never been convinced that a very expensive OPT is not
to be counted as a 'component', and of course you also need a massive
choke to provide a CLC rail filter, if an active load is not used.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

Stewart Pinkerton wrote:
On Fri, 17 Dec 2004 15:41:40 GMT, mick wrote:


On Fri, 17 Dec 2004 21:31:33 +1100, Patrick Turner wrote:

snip
or even

as linear

as a 6L6 set up SE, no FB, and in beam tetrode mode.


snip

grin - tongue unjammed out of cheek yet Patrick? :-)


I know where he's going with that argument, but I don't see why
putting together a good-performing SS amp with a very low parts count
is 'cheating', just because it's not implemented in the same way that
you would use a 300B. I try to avoid using my chisels as if they were
screwdrivers!

I guess it depends what you're trying to prove with a KISS design,
actual simplicity, or forcing a SS copy of how you'd build a SE 300B
amp. It's not like two-stage SE 6SN7/300B amps are exactly
ground-breaking 'gee whizz, how did he do that' technology, so I'm not
really sure what the point of Andre's amp is in any case.............

I see no reason why a simple SS amp should have to meet any design
requirements other than "simple". And I suppose for comparison
purposes it should have a power output similar to what it's being
compared to.
Jeff T

Ruud Broens wrote:

"Stewart Pinkerton" wrote in message
...
: On Fri, 17 Dec 2004 01:52:17 GMT, "Ian Iveson"
: wrote:
:
: "Patrick Turner" wrote
:
: Ah, but greater minds than yours or mine suggest there **is**
: NFB in a triode.
:
: Maybe greater than your mind. If any of them claim what you are
: claiming, their minds are definitely not greater than mine. There is
: dynamic equilibrium in a triode, that some may loosely, maliciously,
: dishonestly, or foolishly call NFB. I have accepted all along that
: this loose definition has currency, as you will see if you read my
: posts.
:
: If there is not an internal feedback loop in a triode, how is it that
: breaking the loop with a screen increases the gain?
:
: --
:
: Stewart Pinkerton | Music is Art - Audio is Engineering

Well, I'd not want to call it a loop, doesn't get much more
instantaneous then field addition in a few cm of space, does it ?

More like a few mm.

And all personally installed by a God Of Triodes.

Yeah; fairly free of delays.....

Until you get to RF, and then the miller effect kicks in.
But triodes could have been used instead of inventing the pentode
if they had used a cascode circuit.
But a DHT in a cascode cascode means you have a heater power supply for the
top tube attatched to
the cathode of the top tube, which meant high noise and stray C, so it was
much easier to have a pentode, and achieve
low miller C and high Ro just with a screen.
And triodes in 1918 were hi-fallootin secret cutting edge technology and
bleeding expensive;
why just one triode cost the same as a cut crystal wine glass used by the
rich at their lavish dinners
while the boys played war games in France with triodes and guns and things.

The pentode, then the transistor, allowed a lot more to happen for many
more ppl.

The harmonic distortion in the pentode RF app didn't matter with a tuned
circuit load because the
tuned circuit only passed one F, and signal pentodes don't have to produce
much voltage
except in an AM reciever where to make 3 vrms of audio at 100% modulation
the
AM envelope wave at 455 kHz IF is 16.8v p-p.
Usually this is done with a vari-mu pentode, so 2H is increased.
But a sharp cut off pentode is better for lower thd in the audio.

The vari mu tube sounds a shirtload better than a transistor IF circuit.


And as F went higher and higher, even pentodes had to be slimmed down
to almost nothing in a bottle, 6AM6 and the acorn tubes were tiny
things.....

I have yet to try a 6AM6 as a triode. Maybe its a nice phono stage tube.
I have dozens of them.

Patrick Turner.





Rudy

Ian Iveson wrote:

"Stewart Pinkerton" wrote

If there is not an internal feedback loop in a triode, how is it
that
breaking the loop with a screen increases the gain?

By altering the conditions of dynamic equilibrium. For a control
system engineer, direction is defined as from controlling input to
controlled output. Hence an engineer should know the difference
between feed-forward and feedback. This is the point that students
can get hung up on.

Maybe you could define "altering the conditions of dynamic equilibrium."

To me the equilibrium whatsit is the result of NFB action.
Bright students will see for themselves the electrostatic summation in a
triode is a NFB
network.



The common mistake is to confuse this sense of direction with
another, such as the alignment of fields, or the flow of electrons.

Huh?



Incidentally, your view of the difference between a tetrode and a
triode is misleadingly simplistic. The screen does not merely break
the "loop", but is an active element in its own right.

It does break the loop though when used at a fixed voltage in relation
to the cathode voltage.
If the screen to cathode voltage is allowed to vary, then it too becomes
a third controlling
element in a tube.
But always the anode then has very little effect on the electron flow.

Triode connection of the pentode allows the FB to be fully applied, and
the NFB is then between the
screen and grid fields summing. Since the screen is directly connected
to the anode,
the action is quite triodic.

UL connection where the screen gets 1/2 the anode signal makes the
feeback about 1/2 what it is with the triode.
If one connects the anodes to the screen taps at 50%, and connects the
screens to where anodes are
normally connected at the ends of the primary windings, you get twice
the normal triode FB applied, and the
Ra of the tube is effectively lowered a whole lot more by the
transformed extra NFB.
Nobody does this because of other reasons such as it being too hard to
implement,
and its a waste of wire, and the power ability is reduced.....

Patrick Turner.




Ian

"mick" wrote

You are perhaps getting confused here.

The voltage is applied to the grid of a triode in order to create
an
electrostatic charge on the grid, relative to the cathode.

The actual device controlling electron flow is the field around
the grid.
It is thus correct to state that the feedback mechanism can be
either
internal (directly summated via the field around the grid) or
external
(via a voltage which will be used to change the field around the
grid).

Note that the voltage *itself* is *not* a controlling device in
its own
right, merely a method of maintaining it. Granted that the field
cannot
exist without a voltage, but this is fact.


No, mick.

You control a triode by varying Vgk. Don't see many amps that
control one any other way.

See my other posts, particularly the bit about confusing the
directions of fields, electron flows, or whatever else you fancy,
with the flow of control.

There is no feedback to the input control signal, save for the
miller effect.

It ought to be dawning on someone soon. I am quite genuinely shocked
that ppl don't know this already. Don't know how to explain it any
more. It should surely be clear that *close to the grid voltage*, or
*near to the grid voltage*, or *caused by* or *a reaction to*...
none of these are good enough. It's got to be *summed with the
input*, not *summed with something else that you believe might be
close to the input and maybe caused by it*.

Summed with the input voltage. Yes. That's what it's got to be.

Not somewhere close. Oh no, not near. That's not good enough. Not
summed with something else, nope, that's not sufficient. No, there
can be no doubt whatsoever. It just got to be

***summed with the input voltage***

If it were me, I'd sum it with the input voltage. Oh yes! That's
what I'd do.

Then I could design amplifiers without blunder.

cheers, Ian

Stewart Pinkerton wrote:

On Fri, 17 Dec 2004 15:41:40 GMT, mick wrote:

On Fri, 17 Dec 2004 21:31:33 +1100, Patrick Turner wrote:

snip
or even
as linear

as a 6L6 set up SE, no FB, and in beam tetrode mode.

snip

grin - tongue unjammed out of cheek yet Patrick? :-)

I know where he's going with that argument, but I don't see why
putting together a good-performing SS amp with a very low parts count
is 'cheating', just because it's not implemented in the same way that
you would use a 300B. I try to avoid using my chisels as if they were
screwdrivers!

Using more NFB than for a 300B amp is cheating.





I guess it depends what you're trying to prove with a KISS design,
actual simplicity, or forcing a SS copy of how you'd build a SE 300B
amp. It's not like two-stage SE 6SN7/300B amps are exactly
ground-breaking 'gee whizz, how did he do that' technology, so I'm not
really sure what the point of Andre's amp is in any case.............

You'll never be sure of anything.

But we'll be sure you could not string together a two transistor amp
with not a stitch of NFB, and which was not usable or listenable.

Patrick Turner.


--

Stewart Pinkerton | Music is Art - Audio is Engineering

Stewart Pinkerton wrote:

On Fri, 17 Dec 2004 21:31:33 +1100, Patrick Turner
wrote:

Stewart Pinkerton wrote:

On Thu, 16 Dec 2004 15:00:59 -0600, (John Byrns) wrote:

Which rules have I changed? You may not agree with me that triodes
incorporate an internal negative feedback loop, but that doesn't mean I
have changed any rules, we just don't agree on the rules as they relate to
the operation of triodes, I might just as well accuse you of changing the
rules regardless of reality.

Apologies due, it was not *your* position with which I was in
conflict, but Ivesons. Which is unfortunate, since he and I are both
C&I engineers, but he doesn't seem to understand what goes on inside a
triode. I do agree that triodes have a large amount of internal
feedback (hence their low gain, and hence the pentode, which breaks
the loop). My position is that certain transistors, such as the Sanken
2SC2922/2SA1216 pair, are as linear as triodes when used correctly,
and can be employed to produce low parts count 'zero NFB' amps in
similar manner.

Well all you have to do is proove your position to be valid.

Please set up a single transistor in common emmitter mode, ie, 8 ohm load in
the
collector circuit, idle current about 2.5 amps, grounded emitter.

The input is applied to the base from a very low impedance source.
with less than 0.001% thd.
This will require you to use a low thd amp with low Ro, because the base
input impedance is very low
Then measure the thd products and draw a graph of the level of each
harmonic product as it increases from 0.0 watts to
a maximum of about 10 watts.

Let us know how you get on.

But not before you have done the above experiment will any of us know if the
bjts you have chosen are anywhere near as linear as a triode, or even as linear
as a 6L6 set up SE, no FB, and in beam tetrode mode.

Very amusing, but in fact what I would do to prove the point is set up
the above circuit as you suggest, *plus* an emitter resister which
sets the stage gain to the same as that of a 'naked' 300B, say about
0.47 ohms or so. I'd also be running the stage with a 3 amp bias
current, so it would indeed need quite a beefy driver, able to put out
a nice linear 60 mA or so.

I leave you to the small mercies of your slide rule....



As previously discussed, the use of local degeneration does not in my
view count as loop feedback, as it does not to my knowledge result in
the kind of discordant intermodulation products previously noted to be
produced by global NFB loops under certain conditions.

But the emitter follower topology is a case of a high amount of NFB, and
its use completely negates the purpose of you quest, to proove that a bjt is as
naturally linear
as a triode.

Different point being made here, which is how to achieve best
linearity with minimum parts count, using SS rather than tube
technology. You don't design a sailboat the same way you design a
motor boat.

Both vessels are to have sails. Two sails only Powered by wind.
No electronic winches either.

And boy oh boy, is there some wind about!!!





If we were to accept you premise that EF topology is OK, then
we would then request that the 300B would also be allowed to have 30 dB of
applied global NFB around a second gain tube to make the playing field level
for the comparison.

Ah, but that would increase the circuit complexity, thereby negating
the KISS principle.

Hence, this
should not be a problem for the 'zero NFB' purist.

It remains a real problem for everyone except you, and you cannot
accept that modern transistors are woefully non linear unless
large amounts of NFB are employed, and then there is the problem
of overcoming the naturally low base input impedance of the bjt.

The low base impedance is indeed a problem, but then you have the high
gate capacitance of MOSFETs to consider, if you go that route.

Similarly, I don't
agree that cathode/emitter followers are the spawn of Satan, they seem
simply to be an excellent impedance transformer. Is there any
*logical* objection to them?

Some have no problem with follower stages. I don't have a problem with them
where I use them for the output of a preamp for example.
But in this case where you are floundering while you try to design a simple
bjt amp which is as simple as a 300B amp, follower topology is plainly
quite unacceptable.

I'm not 'floundering', I'm just kicking some tyres.

Ah, that's why your'e limping badly....



But your aim and zeal is to proove bjts can be used as easily as a 300B and its
simple circuitry,
and with as little FB, zero feedback in fact, and still end up with a simple
amp, two devices max,
and have thd lower than 300B, and still have Rin above 10k, and able to be
powered from a
volume pot straight after a CD player..
So, unless you wish to be seen by all as the universe's greatest cheat, please
do not
keep insisting on the use of emitter follower in this case in this thread is
OK.

Depends if you consider the use of a CCS load instead of an OPT to be
'cheating'. I've never been convinced that a very expensive OPT is not
to be counted as a 'component', and of course you also need a massive
choke to provide a CLC rail filter, if an active load is not used.

Two sails are allowed. Not 3.

Wire wound rigging is permitted.

Square rigged vessels are allowed providing contestants'
sail area does not exceed 50 sq metres.

Bamboo raft hull designs are permitted, providing weight does not
exceed 2 tonnes, the lift limit for the club's water crane, in the event of a
sinking.

Racing begins with a final canon shot at 10am sharp.

Your vessel accompanying your sailing contestant to the start line may have steam
engines.

Towing of your contestant during the race during daylight hours up to race time
expiry is not permitted.

The aforesaid steam powered vessel is permitted to tow your contestant back to your
slipway and shiprights after 6pm, when the time limits for race completion expires.

Martini service closes in the club lounge bar of the club at 6.30pm, after which
time strict
dress rules apply.

Patrick Turner



--

Stewart Pinkerton | Music is Art - Audio is Engineering

In article , "Ian Iveson"
wrote:

"John Byrns" wrote in message
...

Hi Ian,

I have two problems with your view of "NFB".

More than two, in my view.

First were is it written that in order to have "NFB" you must sum
voltages?

The control input to a triode is a voltage.

I have seen analog "NFB" systems where the signals being summed
were represented by air pressure, so why not create "NFB" by
summing
electric fields? In the triode the input and output voltages are
transduced to electric fields, and those electric fields are then
combined
within the triode to create "NFB".

The control input of a triode is not air pressure, nor an
electrostatic field. You cannot sum a voltage with an air pressure,
nor with an electrostatic field, nor indeed can you sum an air
pressure with an electrostatic field.

On the contrary, in the early days a common application of triodes was
amplifying air pressure. The air pressure was feed through a transducer
known as a microphone, which produced an electrical voltage that drove the
triode amplifier. In the case where negative feedback was used in the
triode amplifier circuit, beyond the inherent feedback within the triode
tube, they did effectively "sum a voltage with an air pressure", by the
agent of the transducer. A further transducer was used at the output of
one or more triode amplifiers to convert back to the air pressure domain.

Second, what is wrong with doing the summing within the tube, why
must the
summing be external?

I have not said that it is necessary for it to be external. I have
argued that none of the effects you speak of are internal as you
have claimed, but depend in part on external elements. I was arguing
that your claim that it is internal is false.

Huh? I think you seem to be getting confused now, you seemingly
contradict yourself in the last paragraph.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

In article , Stewart Pinkerton
wrote:

On Thu, 16 Dec 2004 15:00:59 -0600, (John Byrns) wrote:

Which rules have I changed? You may not agree with me that triodes
incorporate an internal negative feedback loop, but that doesn't mean I
have changed any rules, we just don't agree on the rules as they relate to
the operation of triodes, I might just as well accuse you of changing the
rules regardless of reality.

Apologies due, it was not *your* position with which I was in
conflict, but Ivesons. Which is unfortunate, since he and I are both
C&I engineers, but he doesn't seem to understand what goes on inside a
triode.

Apologies accepted, but what might I ask is a "C&I engineer"?


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

In article , "Ian Iveson"
wrote:

No, mick.

You control a triode by varying Vgk. Don't see many amps that
control one any other way.

Actually Ian that is not always true, there is such a thing as an inverse
triode. In this circuit the anode is used as the input electrode, and the
output is taken from the grid.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

On Fri, 17 Dec 2004 22:11:12 -0600, (John Byrns) wrote:

In article , Stewart Pinkerton
wrote:

On Thu, 16 Dec 2004 15:00:59 -0600, (John Byrns) wrote:

Which rules have I changed? You may not agree with me that triodes
incorporate an internal negative feedback loop, but that doesn't mean I
have changed any rules, we just don't agree on the rules as they relate to
the operation of triodes, I might just as well accuse you of changing the
rules regardless of reality.

Apologies due, it was not *your* position with which I was in
conflict, but Ivesons. Which is unfortunate, since he and I are both
C&I engineers, but he doesn't seem to understand what goes on inside a
triode.

Apologies accepted, but what might I ask is a "C&I engineer"?

Sorry (again!) for the use of industry jargon. Control and
Instrumentation engineer - which is why we're having hissy fits about
feedback and control mechanisms. The servomech equation is to C&I
engineers what the 300B is to valvies! :-)

Not sure about Ian, but I spent about fifteen years designing large
computer-controlled test instrumentation for Marconi Instruments and
Hughes Aircraft Company. I'm reasonably familar with feedback
mechanisms.

--

Stewart Pinkerton | Music is Art - Audio is Engineering

In message , John
Byrns writes
In article , "Ian Iveson"
wrote:

No, mick.

You control a triode by varying Vgk. Don't see many amps that
control one any other way.

Actually Ian that is not always true, there is such a thing as an inverse
triode. In this circuit the anode is used as the input electrode, and the
output is taken from the grid.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/


Or, much more obviously, a grounded-grid stage.
--
Chris Morriss

In message , Chris Morriss
writes
In message , John
Byrns writes
In article , "Ian Iveson"
wrote:

No, mick.

You control a triode by varying Vgk. Don't see many amps that
control one any other way.

Actually Ian that is not always true, there is such a thing as an inverse
triode. In this circuit the anode is used as the input electrode, and the
output is taken from the grid.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/


Or, much more obviously, a grounded-grid stage.


DUH! That also varies Vg-k of course....... (runs away and hides!)
--
Chris Morriss

On Fri, 17 Dec 2004 16:57:56 GMT, "Ian Iveson"
wrote:

"Stewart Pinkerton" wrote

If there is not an internal feedback loop in a triode, how is it that
breaking the loop with a screen increases the gain?

By altering the conditions of dynamic equilibrium.

aka feedback..............

For a control
system engineer, direction is defined as from controlling input to
controlled output. Hence an engineer should know the difference
between feed-forward and feedback. This is the point that students
can get hung up on.

And this particular situation may be regarded as either, or neither.
Fact remains, there is a mechanism internal to the triode valve, which
both reduces its gain and increases its linearity.

To simplify matters, perhaps we can agree that a reasonable way to
compare devices for linearity is to build a simple common
emitter/source/cathode circuit with a voltage gain at a fixed value of
say ten, with bias current optimised for each device, then compare the
linearities of the BJT, FET and triode valve under those equalised
conditions. Anyone disagree that this is a fair comparison?

The common mistake is to confuse this sense of direction with
another, such as the alignment of fields, or the flow of electrons.

Incidentally, your view of the difference between a tetrode and a
triode is misleadingly simplistic. The screen does not merely break
the "loop", but is an active element in its own right.

Once you do something with it, of course it is. Otherwise, merely a
screen preventing the gain reduction mechanism from occurring. Save
your pomposity for your unfortunate students.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

On Fri, 17 Dec 2004 18:14:13 -0600, Jeff Thompson
wrote:

Stewart Pinkerton wrote:

I know where he's going with that argument, but I don't see why
putting together a good-performing SS amp with a very low parts count
is 'cheating', just because it's not implemented in the same way that
you would use a 300B. I try to avoid using my chisels as if they were
screwdrivers!

I guess it depends what you're trying to prove with a KISS design,
actual simplicity, or forcing a SS copy of how you'd build a SE 300B
amp. It's not like two-stage SE 6SN7/300B amps are exactly
ground-breaking 'gee whizz, how did he do that' technology, so I'm not
really sure what the point of Andre's amp is in any case.............

I see no reason why a simple SS amp should have to meet any design
requirements other than "simple". And I suppose for comparison
purposes it should have a power output similar to what it's being
compared to.

Agreed that the output voltages should be roughly comparable for a
meaningful comparison. My target of roughly 10 watts into 4 ohms is
the same voltage as Andre's 5 watts into 8 ohms, it just has more
current available for driving conventional speakers, and is a result
of the optimum 2-3 amps bias current of my preferred BJT.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

On Sat, 18 Dec 2004 12:36:14 +1100, Patrick Turner
wrote:

Stewart Pinkerton wrote:

I know where he's going with that argument, but I don't see why
putting together a good-performing SS amp with a very low parts count
is 'cheating', just because it's not implemented in the same way that
you would use a 300B. I try to avoid using my chisels as if they were
screwdrivers!

Using more NFB than for a 300B amp is cheating.

Who says I'll be using 'more feedback'? As noted elsewhere, if the
*stage gain* is the same, what's the difference between the internal
degeneration of the triode, and an external emitter resistor on the
BJT? Unless of course you have an agenda................

I guess it depends what you're trying to prove with a KISS design,
actual simplicity, or forcing a SS copy of how you'd build a SE 300B
amp. It's not like two-stage SE 6SN7/300B amps are exactly
ground-breaking 'gee whizz, how did he do that' technology, so I'm not
really sure what the point of Andre's amp is in any case.............

You'll never be sure of anything.

Oh, there are a couple of things I'm sure about............... :-)

But we'll be sure you could not string together a two transistor amp
with not a stitch of NFB, and which was not usable or listenable.

Oh no, I could definitely do that, but that's not my intention, which
is to string together a two-stage BJT amp with not a stitch of global
NFB, and which *is* useable and listenable. Don't not never use no
double negatives, ocker!
--

Stewart Pinkerton | Music is Art - Audio is Engineering

On Sat, 18 Dec 2004 12:59:56 +1100, Patrick Turner
wrote:

But your aim and zeal is to proove bjts can be used as easily as a 300B and its
simple circuitry,
and with as little FB, zero feedback in fact, and still end up with a simple
amp, two devices max,
and have thd lower than 300B, and still have Rin above 10k, and able to be
powered from a
volume pot straight after a CD player..
So, unless you wish to be seen by all as the universe's greatest cheat, please
do not
keep insisting on the use of emitter follower in this case in this thread is
OK.

Depends if you consider the use of a CCS load instead of an OPT to be
'cheating'. I've never been convinced that a very expensive OPT is not
to be counted as a 'component', and of course you also need a massive
choke to provide a CLC rail filter, if an active load is not used.

Two sails are allowed. Not 3.

In other words, unless it's an exact copy of how you'd implement a
300B, it's 'cheating'? Do you realise how pathetically defensive and
stupid is that argument? A BJT is *not* a 300B, you don't implement it
in the same way. You can however get as good results with a similar
parts count, by using the BJT (Or MOSFET) in the manner best suited to
its characteristics. It's called 'design'.

In particular, an output transformer is *essential* to connect a 300B
to a conventional loudspeaker. Use a BJT, and that heavy, expensive,
and intrinsically nonlinear component is no longer necessary. Hence, a
good KISASS design would not use it, but might very well use a CCS to
provide both a suitable collector load and a useful isolation from PSU
ripple. Simple, elegant and definitely not 'cheating' by any
reasonable measure.

--

Stewart Pinkerton | Music is Art - Audio is Engineering

In article , Chris Morriss
wrote:

In message , John
Byrns writes
In article , "Ian Iveson"
wrote:

No, mick.

You control a triode by varying Vgk. Don't see many amps that
control one any other way.

Actually Ian that is not always true, there is such a thing as an inverse
triode. In this circuit the anode is used as the input electrode, and the
output is taken from the grid.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/


Or, much more obviously, a grounded-grid stage.

Actually that is neither obvious or true, as the input to a grounded-grid
stage is applied between the grid and cathode just as in an ordinary
grounded cathode stage.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

On Sat, 18 Dec 2004 11:24:43 +0000, Stewart Pinkerton wrote:

On Sat, 18 Dec 2004 12:59:56 +1100, Patrick Turner
wrote:

But your aim and zeal is to proove bjts can be used as easily as a
300B and its simple circuitry,
and with as little FB, zero feedback in fact, and still end up with a
simple amp, two devices max,
and have thd lower than 300B, and still have Rin above 10k, and able
to be powered from a
volume pot straight after a CD player.. So, unless you wish to be seen
by all as the universe's greatest cheat, please do not
keep insisting on the use of emitter follower in this case in this
thread is OK.

Depends if you consider the use of a CCS load instead of an OPT to be
'cheating'. I've never been convinced that a very expensive OPT is not
to be counted as a 'component', and of course you also need a massive
choke to provide a CLC rail filter, if an active load is not used.

Two sails are allowed. Not 3.

In other words, unless it's an exact copy of how you'd implement a 300B,
it's 'cheating'? Do you realise how pathetically defensive and stupid is
that argument? A BJT is *not* a 300B, you don't implement it in the same
way. You can however get as good results with a similar parts count, by
using the BJT (Or MOSFET) in the manner best suited to its
characteristics. It's called 'design'.

In particular, an output transformer is *essential* to connect a 300B to a
conventional loudspeaker. Use a BJT, and that heavy, expensive, and
intrinsically nonlinear component is no longer necessary. Hence, a good
KISASS design would not use it, but might very well use a CCS to provide
both a suitable collector load and a useful isolation from PSU ripple.
Simple, elegant and definitely not 'cheating' by any reasonable measure.

I'm with you on this one. Its no ruling that a BJT should be used in a
manner in a manner for which it wasn't designed then slating it for poor
performance. You should be given a fair chance! Likewise, if you feel that
an OPT and/or choke-filtered supply would be helpful for your
implementation then you should be allowed to specify them. Use of a CCS
isn't cheating for either side as these have been around for a l-o-n-g
time!

--
Mick
(no M$ software on here... :-) )
Web: http://www.nascom.info
Web: http://projectedsound.tk

John Byrns wrote:

In article , Stewart Pinkerton
wrote:

On Thu, 16 Dec 2004 15:00:59 -0600, (John Byrns) wrote:

Which rules have I changed? You may not agree with me that triodes
incorporate an internal negative feedback loop, but that doesn't mean I
have changed any rules, we just don't agree on the rules as they relate to
the operation of triodes, I might just as well accuse you of changing the
rules regardless of reality.

Apologies due, it was not *your* position with which I was in
conflict, but Ivesons. Which is unfortunate, since he and I are both
C&I engineers, but he doesn't seem to understand what goes on inside a
triode.

Apologies accepted, but what might I ask is a "C&I engineer"?

Competent and Intelligent.

Patrick Turner.



Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

Chris Morriss wrote:

In message , John
Byrns writes
In article , "Ian Iveson"
wrote:

No, mick.

You control a triode by varying Vgk. Don't see many amps that
control one any other way.

Actually Ian that is not always true, there is such a thing as an inverse
triode. In this circuit the anode is used as the input electrode, and the
output is taken from the grid.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

Or, much more obviously, a grounded-grid stage.

True, but the Zin is very low.

So for gain, and high Zin, there cab be two triodes set up like an LTP.
But the one on the input side has an electolytic to bypass the anode to ground
so the input tube is a CF that works the cathode of the other tube which has
its grid grounded.
This makes a wide BW non inverting amplifier.
The commoned cathodes are taken to a CCS or R load to a -ve supply voltage.

Patrick Turner.


--
Chris Morriss

Stewart Pinkerton wrote:

On Fri, 17 Dec 2004 16:57:56 GMT, "Ian Iveson"
wrote:

"Stewart Pinkerton" wrote

If there is not an internal feedback loop in a triode, how is it that
breaking the loop with a screen increases the gain?

By altering the conditions of dynamic equilibrium.

aka feedback..............

For a control
system engineer, direction is defined as from controlling input to
controlled output. Hence an engineer should know the difference
between feed-forward and feedback. This is the point that students
can get hung up on.

And this particular situation may be regarded as either, or neither.
Fact remains, there is a mechanism internal to the triode valve, which
both reduces its gain and increases its linearity.

To simplify matters, perhaps we can agree that a reasonable way to
compare devices for linearity is to build a simple common
emitter/source/cathode circuit with a voltage gain at a fixed value of
say ten, with bias current optimised for each device, then compare the
linearities of the BJT, FET and triode valve under those equalised
conditions. Anyone disagree that this is a fair comparison?

You sound like you want to use 3 sails, not two....

A single power mosfet only has a gm of about 1A/V.
the gain can never be more than 8 for an 8 ohm load.

The power BJT will have enormous gain maybe 240,
if gm = 30A/V.

The 300B though has a net voltage loss between its grid and the speaker
voltage.
About 45 vrms at g1 is needed to put 6.3 v into 8 ohms, ( 5 watts ).

How do you reconcile these connundrums?

I said in the beginning that the producer of the SE two bjt amp should be
allowed to
use whatever amount of NFB he wishes to get to the same Ro
as the 300B amp, and since the collector Ro of a power bjt is real high
compared to the 8 ohm load,
then some considerable amount of NFB has to be used.
Your expected amp will be impossible to listen to without NFB; it would be
definately
quite ****eful and an article of displeasure.

But we plan to keep you honest, so you cannot use more NFB
than in the 300B, or more than 2 devices.

Think about your design, draw it up, and post it to a site where we can
spend an hour to pick it to peices.

Patrick Turner.


The common mistake is to confuse this sense of direction with
another, such as the alignment of fields, or the flow of electrons.

Incidentally, your view of the difference between a tetrode and a
triode is misleadingly simplistic. The screen does not merely break
the "loop", but is an active element in its own right.

Once you do something with it, of course it is. Otherwise, merely a
screen preventing the gain reduction mechanism from occurring. Save
your pomposity for your unfortunate students.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

Stewart Pinkerton wrote:

On Sat, 18 Dec 2004 12:36:14 +1100, Patrick Turner
wrote:

Stewart Pinkerton wrote:

I know where he's going with that argument, but I don't see why
putting together a good-performing SS amp with a very low parts count
is 'cheating', just because it's not implemented in the same way that
you would use a 300B. I try to avoid using my chisels as if they were
screwdrivers!

Using more NFB than for a 300B amp is cheating.

Who says I'll be using 'more feedback'? As noted elsewhere, if the
*stage gain* is the same, what's the difference between the internal
degeneration of the triode, and an external emitter resistor on the
BJT? Unless of course you have an agenda................

Well if we allowed you to use 30 dB of series voltage NFB, then
any comparison between AJ's amp and yours is meaningless.

Then AJ should be allowed to add whatever NFB is needed to keep level with
your
NFB.

There has to be a level playing field.

But you seem miles awau from the slightest prospective design.

Include an emitter R if you like; it increases Ro, and reduces gain
since its series current NFB.

You will still need series voltage NFB to make your amp at all usable.





I guess it depends what you're trying to prove with a KISS design,
actual simplicity, or forcing a SS copy of how you'd build a SE 300B
amp. It's not like two-stage SE 6SN7/300B amps are exactly
ground-breaking 'gee whizz, how did he do that' technology, so I'm not
really sure what the point of Andre's amp is in any case.............

You'll never be sure of anything.

Oh, there are a couple of things I'm sure about............... :-)

But we'll be sure you could not string together a two transistor amp
with not a stitch of NFB, and which was not usable or listenable.

Oh no, I could definitely do that, but that's not my intention, which
is to string together a two-stage BJT amp with not a stitch of global
NFB, and which *is* useable and listenable. Don't not never use no
double negatives, ocker!

We ain't seen any sign of any schematic yet....

Too hard for ya, eh?

Patrick Turner.


--

Stewart Pinkerton | Music is Art - Audio is Engineering

Stewart Pinkerton wrote:

On Sat, 18 Dec 2004 12:59:56 +1100, Patrick Turner
wrote:

But your aim and zeal is to proove bjts can be used as easily as a 300B and its
simple circuitry,
and with as little FB, zero feedback in fact, and still end up with a simple
amp, two devices max,
and have thd lower than 300B, and still have Rin above 10k, and able to be
powered from a
volume pot straight after a CD player..
So, unless you wish to be seen by all as the universe's greatest cheat, please
do not
keep insisting on the use of emitter follower in this case in this thread is
OK.

Depends if you consider the use of a CCS load instead of an OPT to be
'cheating'. I've never been convinced that a very expensive OPT is not
to be counted as a 'component', and of course you also need a massive
choke to provide a CLC rail filter, if an active load is not used.

Two sails are allowed. Not 3.

In other words, unless it's an exact copy of how you'd implement a
300B, it's 'cheating'? Do you realise how pathetically defensive and
stupid is that argument? A BJT is *not* a 300B, you don't implement it
in the same way. You can however get as good results with a similar
parts count, by using the BJT (Or MOSFET) in the manner best suited to
its characteristics. It's called 'design'.

In particular, an output transformer is *essential* to connect a 300B
to a conventional loudspeaker. Use a BJT, and that heavy, expensive,
and intrinsically nonlinear component is no longer necessary. Hence, a
good KISASS design would not use it, but might very well use a CCS to
provide both a suitable collector load and a useful isolation from PSU
ripple. Simple, elegant and definitely not 'cheating' by any
reasonable measure.

You'd be better off designing rather than whinging about the flak.

Patrick Turner.



--

Stewart Pinkerton | Music is Art - Audio is Engineering

On Sun, 19 Dec 2004 04:08:09 +1100, Patrick Turner wrote:


Using more NFB than for a 300B amp is cheating.


....but surely such "cheating" is normal for ss amplifiers? IMHO you can't
tell him off for using normal design techniques.


Stewart:
Who says I'll be using 'more feedback'? As noted elsewhere, if the
*stage gain* is the same, what's the difference between the internal
degeneration of the triode, and an external emitter resistor on the BJT?
Unless of course you have an agenda................

Patrick:

Well if we allowed you to use 30 dB of series voltage NFB, then any
comparison between AJ's amp and yours is meaningless.

Then AJ should be allowed to add whatever NFB is needed to keep level with
your
NFB.


No problem with that if he wants to imho...


There has to be a level playing field.


Yep, both to use standard design techniques for their respective
amplification devices...


But you seem miles awau from the slightest prospective design.

Include an emitter R if you like; it increases Ro, and reduces gain
since its series current NFB.

You will still need series voltage NFB to make your amp at all usable.



Note though, Patrick, that anyone using a lot of NFB runs the risk of
causing IMD problems. It's a two-edged sword. Comparison of the two
amplifiers is still surely valid, as the one with the most global NFB will
have the worst IMD figures too - pointing out the inferior design even if
its thd% is way smaller. This isn't a competition to find the lowest thd%.

--
Mick
(no M$ software on here... :-) )
Web: http://www.nascom.info
Web: http://projectedsound.tk

In article , mick
wrote:

On Sun, 19 Dec 2004 04:08:09 +1100, Patrick Turner wrote:

But you seem miles awau from the slightest prospective design.

Include an emitter R if you like; it increases Ro, and reduces gain
since its series current NFB.

You will still need series voltage NFB to make your amp at all usable.



Note though, Patrick, that anyone using a lot of NFB runs the risk of
causing IMD problems. It's a two-edged sword. Comparison of the two
amplifiers is still surely valid, as the one with the most global NFB will
have the worst IMD figures too - pointing out the inferior design even if
its thd% is way smaller. This isn't a competition to find the lowest thd%.

If the THD is high, the IMD will also be high, they stem from the same non
linearities. I don't believe there is such a beast as a low THD high IMD
amplifier.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

On Sat, 18 Dec 2004 21:34:23 GMT, mick wrote:

Note though, Patrick, that anyone using a lot of NFB runs the risk of
causing IMD problems. It's a two-edged sword. Comparison of the two
amplifiers is still surely valid, as the one with the most global NFB will
have the worst IMD figures too

Where do concepts like this originate?

Chris Hornbeck

"Chris Morriss" wrote

Or, much more obviously, a grounded-grid stage.


DUH! That also varies Vg-k of course....... (runs away and
hides!)
--

You raise an important point.

Varying Vgk via the cathode also varies Vak directly. This tends to
get overlooked.

The effects of these two inputs will be superimposed, so you don't
just get the effect of the genuine control nfb to Vgk, you also get
some output from the input to Vak.

Now, in the sense that a resistor can be properly contrived to be a
control system, so can a triode, when driven via Vak.

That is the same sense in which a thermionic diode with load can be
said to be a feedback system.

Just wondering about delay here. If I vary grid and cathode
together, so as to change Vak without changing Vgk. There's
resistance and capacitance in parallel. But not multiplied by gain.
Not very significant compared to the effect changing the dynamic
equilibrium point, I don't suppose.

So we have an effect similar to but less significant than the Miller
that is a path for genuine nfb, and which can be minimised in the
same way by using a low impedance source.

And the space charge thing is still not a source of significant nfb
because it causes no significant delay.

Not as sure about this though...need to think it through.

Incidentally, the most power can be got out of a pentode using CFB
if Vsk is kept constant I believe.


cheers, Ian

Air pressure is not the input to the triode in your example.

I haven't contradicted myself, I don't thing. Double checked before
I posted. It's an awkward paragraph, to be sure, but i think it
makes the right sense.

cheers, Ian

"John Byrns" wrote in message
...
In article , "Ian
Iveson"
wrote:

"John Byrns" wrote in message
...

Hi Ian,

I have two problems with your view of "NFB".

More than two, in my view.

First were is it written that in order to have "NFB" you must
sum
voltages?

The control input to a triode is a voltage.

I have seen analog "NFB" systems where the signals being summed
were represented by air pressure, so why not create "NFB" by
summing
electric fields? In the triode the input and output voltages
are
transduced to electric fields, and those electric fields are
then
combined
within the triode to create "NFB".

The control input of a triode is not air pressure, nor an
electrostatic field. You cannot sum a voltage with an air
pressure,
nor with an electrostatic field, nor indeed can you sum an air
pressure with an electrostatic field.

On the contrary, in the early days a common application of triodes
was
amplifying air pressure. The air pressure was feed through a
transducer
known as a microphone, which produced an electrical voltage that
drove the
triode amplifier. In the case where negative feedback was used in
the
triode amplifier circuit, beyond the inherent feedback within the
triode
tube, they did effectively "sum a voltage with an air pressure",
by the
agent of the transducer. A further transducer was used at the
output of
one or more triode amplifiers to convert back to the air pressure
domain.

Second, what is wrong with doing the summing within the tube,
why
must the
summing be external?

I have not said that it is necessary for it to be external. I
have
argued that none of the effects you speak of are internal as you
have claimed, but depend in part on external elements. I was
arguing
that your claim that it is internal is false.

Huh? I think you seem to be getting confused now, you seemingly
contradict yourself in the last paragraph.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

mick wrote:

On Sun, 19 Dec 2004 04:08:09 +1100, Patrick Turner wrote:


Using more NFB than for a 300B amp is cheating.


...but surely such "cheating" is normal for ss amplifiers? IMHO you can't
tell him off for using normal design techniques.

Then we may as well simply compare any old multi component
high power SS amp with a shirtload of NFB to AJ's 300B .

I thought the aim was to produce a two bjt amp capable of a few watts,
same amount of NFB, but certainly no more than in a 300B.

The bjt cannot be used without FB.



Include an emitter R if you like; it increases Ro, and reduces gain
since its series current NFB.

You will still need series voltage NFB to make your amp at all usable.



Note though, Patrick, that anyone using a lot of NFB runs the risk of
causing IMD problems.

Technically speaking, the use of a large amount of NFB reduces all distortions
of any kind
including the IMD which is worst when there is no NFB and thd is high.

It's a two-edged sword. Comparison of the two
amplifiers is still surely valid, as the one with the most global NFB will
have the worst IMD figures too - pointing out the inferior design even if
its thd% is way smaller. This isn't a competition to find the lowest thd%.

I think you ought to consider the facts about amplifier theory in general.
High NFB means low imd. High NFB means above 20+ dB of applied NFB.
100 dB of applied NFB in many SS amps makes it difficult to measure the thd and
imd,
Halcro claim 0.0001% thd at all F up to 20 kHz and 200 watts.


But if only a small amount of NFB, say 10 dB, applied around a horribly non
linear device without any NFB that gives say 10% thd will indeed add significant
extra products
to the thd spectra because of the IMD process and the amp will sound worse.
It would have been bad to start with, without any NFB.

But if the amp produces say 1% at 5 watts with no FB and we run a loop
of 10 dB, the extra thd products are still produced, but are insignificant.
10 dB is a gain reduction of 3.2 times.

20 dB of NFB around the same amp makes a fairly cleaned up
harmonic spectra. This is a gain reduction of 10 times.

NFB has been used liberally for 60+ years without too much complaint,
and I fear abandoning it for spurious or doubtful reasons is like throwing out
the bathwater
with the baby still sitting in the tub.
In bygone days when all recording was analog, not digital as it is now,
the recording industry had a very heavy reliance on NFB to
"keep the intermods down" as the engineers used to say.
After running the recorded sound a few times through gear without NFB and
finally cutting
a record without NFB applied to the cutting head, the music was somewhat
far less listenable than it was heard in the recording venue.
There were few recording studios which employed all triode recording amps
and all triode cutting head amps which were free of any loops of NFB.
Why? expense for one thing, and technical performance and sound quality and
noise being the others.

Later in the era of analog, between about 1960 and 1980,
recording studios used SS gear with the necessary high amount of NFB
to make it at all usable. I believe it was in vain, and they should have kept
their tube gear
which often sounded better, most of the time, but not always.
Digital is almost universal now, and for that we need zillions of transistors,
and I leave it up to others to argue if it sounds better than the best of what
was made using all
analog tubed gear on a good day when things all worked properly.



As many who have preceeded me, I have tried to point out there is a NFB loop
in all triodes, and it can be quantified in dB relative to the load value.

Although triodes are inefficient, they are capable of fabulous sound without
external loops of NFB, since they have NFB already within them.
In fact when we limit the efficiency to about 15% in class A, we
get the best sound and lowest thd.
But then without the NFB we must make the extra effort to make the driver stages

and the OPT of the amp give us sufficient bandwidth, and also to keep the noise
level low,
and this is so much more difficult without NFB.

This means using them with higher than conventional load values,
or using 4 triodes hooked up to the same RL as we would use for 2 triodes.

This allows 20 watts from 4 x EL34, and what fabulous watts they are!

Patrick Turner.






--
Mick
(no M$ software on here... :-) )
Web: http://www.nascom.info
Web: http://projectedsound.tk

Chris Hornbeck wrote:

On Sat, 18 Dec 2004 21:34:23 GMT, mick wrote:

Note though, Patrick, that anyone using a lot of NFB runs the risk of
causing IMD problems. It's a two-edged sword. Comparison of the two
amplifiers is still surely valid, as the one with the most global NFB will
have the worst IMD figures too

Where do concepts like this originate?

I do wonder.

I addressed the author's concerns above in another post.

Patrick Turner.



Chris Hornbeck

On Sun, 19 Dec 2004 04:08:09 +1100, Patrick Turner
wrote:

We ain't seen any sign of any schematic yet....

Too hard for ya, eh?

Not hard at all, but you won't be seeing any schematic until the amp
has been built and tested - which won't be until well after Christmas,
since I'm off to Venezuela for a couple of weeks mid-January, and I
have better things to do.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

"Patrick Turner" wrote

Bright students will see for themselves the electrostatic
summation in a
triode is a NFB
network.

A bright student would not have allowed Stew to trap him into
accepting that nfb around Stew's bjt is allowable because of your
stupid persistence with your internal triode nfb theory. Talk about
reductio ad absurdum!

You didn't even have the sense to wonder if the bjt already has its
own mysterious internal feedback mumbo jumbo. That would have at
least maintained the appearance of honesty.

Now you have allowed him to squirm out of the whole project. Maybe
Stew is cleverer than we thought...

Maybe you could define "altering the conditions of dynamic
equilibrium."

I'm a prophet, not a dictionary.

Consider two cards leaning up against each other, like the start of
a house of cards, on a flat surface. What is holding them up? Card A
stays up because it is leaning on B, so A depends on B. However, B
stays up because it is leaning on A, so B depends on A. Therefore A
depends on A, and B depends on B. Therefore they are both dependent
on each other, and at the same time independent. This is the oldest
problem of philosophy, enshrined in the pi symbol, Stonehenge, the
key symbol of the i ching, yin and yang, and the holy trinity. In
each of these, the top line (or the enclosing circle) denotes the
relationship. For believers in the holy trinity, the relationship is
god. For the cards, it is the bottom line of the triangle.

The cards are an example of static equilibrium. God may be more
dynamic...

Best person to describe dynamic equilibrium would be Hegel, but
perhaps an example would suffice.

Put a pressure vessel part filled with water on the hob. After a
while the water will boil and will continue to do so until the
pressure is so high that the temperature of the water and gases
rises to such a point that the heat lost to the surrounding air is
the same as that input through the hob. Keep the hob at that
setting. The pressure and temperature inside the vessel will remain
constant, even though the water will still be boiling, because there
will be a balance between pressure rise from boiling and pressure
loss by condensation. The relationship between temperature and
pressure is one of dynamic equilibrium. Two changing interdependent
variables in balance.

In each of these cases you could loosely call it feedback. Trouble
is, whichever card or variable you pick first in your head, the
feedback is in that direction. Could be either way. If it is
instantaneous, or if it's history is not known, or both, then you
can't distinguish one direction from its opposite.

Turn up the hob and the temperature will rise so the water will boil
faster so the pressure will rise so it will get hotter until the
equilibrium is restored at a new point. If you take the control
input as the hob setting, then this dynamic equilibrium is not
feedback in the strict sense. The forces effecting the equilibrium
do not impinge on the hob setting, so it is not a feedback control
system. If the input is taken to be the hob temperature, then there
is a minor nfb loop, because the rate of exchange of heat depends on
the difference in temperature between hob and vessel. This is a
minor effect if the hob is small and hot compared to the vessel. A
bit like the miller.

But if you consider the input to be gas pressure, and the output to
be water temperature, then the dynamic equilibrium has a control
direction, and since one participant in the equilibrium is the
input, and there is delay, it is a proper nfb loop. Raise the
pressure, water stops boiling but condensation continues so pressure
tends to fall. Meanwhile water temperature rises because its not
boiling, then it boils again so pressure tends to rise. Eventually a
new equilibrium point is reached, depending partly on the input
impedance of whatever you are using to vary the pressure.

To me the equilibrium whatsit is the result of NFB action.

Depends on whether the equilibrium involves the control input, and
on whether there is significant delay.

Are we getting anywhere?

cheers, Ian

"Stewart Pinkerton" wrote

aka feedback..............

Only in the loose sense, not in the sense a control engineer should
use for the purpose of designing an amplifier. Check most of my
posts in this thread for more info.

Incidentally, your view of the difference between a tetrode and a
triode is misleadingly simplistic. The screen does not merely
break
the "loop", but is an active element in its own right.

Once you do something with it, of course it is. Otherwise, merely
a
screen preventing the gain reduction mechanism from occurring.
Save
your pomposity for your unfortunate students.
--

Me? Pompous?

So now we have an imaginary screen defeating imaginary nfb in an
imaginary manner.

As if adding a screen to a triode isn't doing "something with it".

If to avoid pomposity I should bring myself down to your level, I'd
rather stay pompous.

Cute trick to get that nfb round your bjt, though Patrick's
proper shot himself in the foot there.

cheers, Ian