[The Phantom]

On Fri, 29 Jun 2007 08:52:19 +0100, wrote:

Ian Iveson wrote:

Other Ian said

I don't see that but maybe I am looking at the wrong part of the
data sheet.
The part I see under 'Characteristics and Typical Operation' has a
column
with screen and plate volts both 100V and a gm of 3.9mA/V. These are
the
voltages my AVO tester is set up with for meauding the 6AU6 in
pentode
mode. The plate current is 5mA and the screen current is 2.1mA,
about 30%
of cathode current - OK that's higher than my 20% figure but not
hugely.
This gives a triode equivalent gm of 3.9 x 7.1/5 = 5.5mA/V, still
rather
lower than the 7 I was measuring.

I can see in the right hand column where the plate is at 250V and
the screen
at 150 that the gm rises to 5.2. The triode equivalent under these
conditions works out at about 7.2mA/V which is much closer to what I
measured but with rather different voltages.

Firstly, 30% is much greater than 20% in my book. 50% greater, in
fact.


Depends on its effect.

Second, the issue is gm not current, ie rate of change of current with
voltage, dIa/dVgc.


Er, I thought it was about how gm is split in the ratio of the plate and
screen currents - so its about current AND gm.

My inspection of the curves was made at 150V because the given pentode
curves are with the screen at 150V

Screen gm at that voltage is about 1.2mA/V, whereas anode gm is 2.8.
That is over 40%. When does a difference become "huge"?


When it makes a significant change to the answer. And no, I am not going to
define significant ;-)


As you get closer to the knee of the curves, the proportion increases
further.

All I am saying is that your meter may be showing the right ratio of
values, although it may not be showing the correct absolute values. I
have an AVO CT 160 that is fairly accurate in the middle of its
ranges, but rather poor at the extremes. However, it has always given
usably accurate comparisons, in that the error is reliable and
favourably distributed.


OK

There is much greater provision for calibration on yours, but AFAIK
the fundaments of its operation are equally bizarre, using AC on
screen and anode.


Yes it is odd isn't it. They even have (had) a patent on the technique.

Could it be a variation on the technique shown on page 103 of RDH4?


Ian
cheers, Ian1

[[email protected]]

The Phantom wrote:

On Fri, 29 Jun 2007 08:52:19 +0100, wrote:

Ian Iveson wrote:

Other Ian said

I don't see that but maybe I am looking at the wrong part of the
data sheet.
The part I see under 'Characteristics and Typical Operation' has a
column
with screen and plate volts both 100V and a gm of 3.9mA/V. These are
the
voltages my AVO tester is set up with for meauding the 6AU6 in
pentode
mode. The plate current is 5mA and the screen current is 2.1mA,
about 30%
of cathode current - OK that's higher than my 20% figure but not
hugely.
This gives a triode equivalent gm of 3.9 x 7.1/5 = 5.5mA/V, still
rather
lower than the 7 I was measuring.

I can see in the right hand column where the plate is at 250V and
the screen
at 150 that the gm rises to 5.2. The triode equivalent under these
conditions works out at about 7.2mA/V which is much closer to what I
measured but with rather different voltages.

Firstly, 30% is much greater than 20% in my book. 50% greater, in
fact.


Depends on its effect.

Second, the issue is gm not current, ie rate of change of current with
voltage, dIa/dVgc.


Er, I thought it was about how gm is split in the ratio of the plate and
screen currents - so its about current AND gm.

My inspection of the curves was made at 150V because the given pentode
curves are with the screen at 150V

Screen gm at that voltage is about 1.2mA/V, whereas anode gm is 2.8.
That is over 40%. When does a difference become "huge"?


When it makes a significant change to the answer. And no, I am not going
to define significant ;-)


As you get closer to the knee of the curves, the proportion increases
further.

All I am saying is that your meter may be showing the right ratio of
values, although it may not be showing the correct absolute values. I
have an AVO CT 160 that is fairly accurate in the middle of its
ranges, but rather poor at the extremes. However, it has always given
usably accurate comparisons, in that the error is reliable and
favourably distributed.


OK

There is much greater provision for calibration on yours, but AFAIK
the fundaments of its operation are equally bizarre, using AC on
screen and anode.


Yes it is odd isn't it. They even have (had) a patent on the technique.

Could it be a variation on the technique shown on page 103 of RDH4?


Probably that represents its starting point. It uses a rectified but not
smoothed 50Hz signal for testing and after nulling out dc components,
operting a switch gives a direct meter reading of gm rather than having to
calculate is as in the RDH4 method. I have a description of the method used
somewhere. if I can find it I'll post it to the group.

Cheers
Ian (TB)

[Ian Iveson]

Ian said

Er, I thought it was about how gm is split in the ratio of the plate
and
screen currents - so its about current AND gm.

But gm is *not* always split in the same ratio as that of screen
currents. That is an approximation that becomes less true as you
approach the knee, AFAICS from the curves.

Anyway, an interesting question is *why* the 6AU6 has so much screen
current. What is it about the geometry of the valve that determines
the ratio of screen and anode currents and gms?

Yes it is odd isn't it. They even have (had) a patent on the
technique.

There is a totally classic page of English Engineer's AVO-speak that
explains a constant used to correct for the ensuing error. You might
expect a conversion from rms to average, and a factor of 2 because of
the half-wave operation, but the number they use isn't quite what you
would expect, AFAIR.

cheers, Ian1

[[email protected]]

Ian Iveson wrote:

Ian said

Er, I thought it was about how gm is split in the ratio of the plate
and
screen currents - so its about current AND gm.

But gm is *not* always split in the same ratio as that of screen
currents. That is an approximation that becomes less true as you
approach the knee, AFAICS from the curves.


Indeed, but I don't intend to work in that region because gm is lower and
distortion higher.

Anyway, an interesting question is *why* the 6AU6 has so much screen
current. What is it about the geometry of the valve that determines
the ratio of screen and anode currents and gms?


That is an intersting question. Do you know the answer?

Yes it is odd isn't it. They even have (had) a patent on the
technique.

There is a totally classic page of English Engineer's AVO-speak that
explains a constant used to correct for the ensuing error. You might
expect a conversion from rms to average, and a factor of 2 because of
the half-wave operation, but the number they use isn't quite what you
would expect, AFAIR.

AVO speak is exactly right. The manual I have says:

"It was eventually found that a complete co-relation between the two sets of
condition was held when the grid voltage took the form of a sinusoidal
waveform with the positive half cycle suppressed (in other words rectified
but completely unsmoothed AC) and the following relationships were
maintained:

Va (rms) = 1.1 Va indicated dc
Vg2(rms) = 1.1 Vg2 indicated dc
Vg1(mean unsmoothed) = 0.51 Vg1 indicated dc
Ia(mean dc) = 0.5 indicated Ia"

No explanation whatsoever as to why these particular values hold.

Ian

cheers, Ian1

[Ian Iveson]

Ian said:

AVO speak is exactly right. The manual I have says:

"It was eventually found that a complete co-relation between the two
sets of
condition was held when the grid voltage took the form of a
sinusoidal
waveform with the positive half cycle suppressed (in other words
rectified
but completely unsmoothed AC) and the following relationships were
maintained:

Va (rms) = 1.1 Va indicated dc
Vg2(rms) = 1.1 Vg2 indicated dc
Vg1(mean unsmoothed) = 0.51 Vg1 indicated dc
Ia(mean dc) = 0.5 indicated Ia"

No explanation whatsoever as to why these particular values hold.

Right, thanks, that's the page. It reads like an explanation, but
isn't. But we might be missing something obvious.

Just in case anyone is following this and might have an answer...the
"rectification" of the screen and anode supplies is done by the valve
under test...ie the valve is subjected to the whole AC cycle. The only
rectifier in the tester is used for the grid and bridge circuit. Hence
it avoids the need for a high power rectifier. That's how it is with
the CT160 anyway.

Indeed, but I don't intend to work in that region because gm is
lower and
distortion higher.

Wise decision. But the 80V test was quite close.

What is it about the geometry of the valve that determines
the ratio of screen and anode currents and gms?

That is an intersting question. Do you know the answer?

No. I almost don't dare ask :-)

The knee is at a remarkably low voltage, and high current, and there's
no kink. A power valve like that would be an instant candidate for
pentode operation with cathode feedback, in my amps perhaps, to
replace the much gentler 6CH6. It would be interesting to try pp 6AU6
as a driver or headphone amp with a transformer: perhaps 30% cathode
winding and cross-coupled screens to 30% taps to maintain constant
Vks.

But perhaps there is a necessary link between the knee position and
the high screen current. Perhaps the screen is relatively close to the
grid. That might also explain why small differences in the details of
construction between different examples lead to relatively large
variations in characteristics.

I seriously don't dare ask why sharp cut-off pentodes are, er, sharp
cut-off, and in what particular circumstances it is advantageous for
them to be so. I'm tired of getting my ankles savaged.

cheers, Ian1

[Patrick Turner]

wrote:

Patrick Turner wrote:



Ian Iveson wrote:

Other Ian said

I don't see that but maybe I am looking at the wrong part of the
data sheet.
The part I see under 'Characteristics and Typical Operation' has a
column
with screen and plate volts both 100V and a gm of 3.9mA/V. These are
the
voltages my AVO tester is set up with for meauding the 6AU6 in
pentode
mode. The plate current is 5mA and the screen current is 2.1mA,
about 30%
of cathode current - OK that's higher than my 20% figure but not
hugely.
This gives a triode equivalent gm of 3.9 x 7.1/5 = 5.5mA/V, still
rather
lower than the 7 I was measuring.

I can see in the right hand column where the plate is at 250V and
the screen
at 150 that the gm rises to 5.2. The triode equivalent under these
conditions works out at about 7.2mA/V which is much closer to what I
measured but with rather different voltages.

Firstly, 30% is much greater than 20% in my book. 50% greater, in
fact.

My data book for 6AU6 has the following data,
Ig2 = 4.3mA, Eg2 = 150V, when Ia = 10.6mA, Ea 250V, gm = 5.2mA/V,
And
Ig2 = 2.1mA, Eg2 = 100mA when Ia = 5.0mA, Ea = 100V, gm = 3.9mA/V

So over a broad range of voltages, Ig2 = 40% of Ia.

However, if you go to the RCA sheets with 6AU6 TRIODE curves, at Ia =
10mA and Ea = 250V
the µ = 38.6 and Ra = 9.1k, and gm = µ / Ra = 38.6 / 9,100 = 4.24 mA/V

Who was it who told me that triode gm was HIGHER than pentode gm????????


Actually is was page 34 of RDH4.

In theory this is right. RDH4 says the gm is shared with respect to the
proportion of Ig2 and Ia.

But if we consider the real world use of pentodes like the 6AU6,
then we'd expect gm in triode to be 40% due to screen, and 60% due to
anode,
if that's the ratio of Ig2 to Ia, and when in triode, the Ea = Eg2.

So you'd think it could be compared with a pentode set up where Ea = Eg2
in the idle condition.

And one would expect that the pentode gm would be a lot less the the
triode connection,
but anode voltage moves and screen voltage does not, and just exactly
what you will get depends on someone doing a test on a 6AU6 so that we
don't just have to take a tiny little
paragraph from RDH4 as gospel.

At the top of page 36 RDH4 mentions the 6AU6, and in pentode
you get Ra = 1.5M, gm = 4.45mA/V.
µ = gm x Ra = 0.00445 x 1,500,000 = 6,675.

Triode µ is given as 36, and THEREFORE screen µ = 6,675 / 36 = 185.

OK so far, and that's all RDH4 says about 6AU6 in this part of the book.

If therefore we say that gm of the screen = µ / Ra,
then gm g2 = 185 / 1,500,000 = 0.123mA/V, but this sounds way too low.

So what is the Ra of the 6AU6 when driven as a triode with fixed g1
voltage and drive to g2???

RDH4 doesn't fill in the gaps AT ALL. MUCH more could be said but isn't.

So to know what you really get you have to set up a tube and measure
its gain at low output voltages for negligibe distortions, say 10Vrms
will do,
and use an accurate volt meter for input and output voltage, and use two
RLs
with one twice the value of the other, and finally you can get two gain
equations to
work out Ra, µ and hence gm for any tube, triode or pentode or g2 drive
with g1 at say 0V.

To understand the tubes this way you have to be just as clever and
disciplined as the authors of RDH4.

It isn't very difficult, but it just takes time, and you have to build a
decent test rig.

Get away from the PC and the books, get out into your workshop, and
maybe you'll answer all your
own questions.

Patrick Turner.


Ian

[[email protected]]

Ian Iveson wrote:

Ian said:

AVO speak is exactly right. The manual I have says:

"It was eventually found that a complete co-relation between the two
sets of
condition was held when the grid voltage took the form of a
sinusoidal
waveform with the positive half cycle suppressed (in other words
rectified
but completely unsmoothed AC) and the following relationships were
maintained:

Va (rms) = 1.1 Va indicated dc
Vg2(rms) = 1.1 Vg2 indicated dc
Vg1(mean unsmoothed) = 0.51 Vg1 indicated dc
Ia(mean dc) = 0.5 indicated Ia"

No explanation whatsoever as to why these particular values hold.

Right, thanks, that's the page. It reads like an explanation, but
isn't. But we might be missing something obvious.

Just in case anyone is following this and might have an answer...the
"rectification" of the screen and anode supplies is done by the valve
under test...ie the valve is subjected to the whole AC cycle. The only
rectifier in the tester is used for the grid and bridge circuit. Hence
it avoids the need for a high power rectifier. That's how it is with
the CT160 anyway.


According to the circuit I have there is a rectifier in the screen
connection too.

Indeed, but I don't intend to work in that region because gm is
lower and
distortion higher.

Wise decision. But the 80V test was quite close.

What is it about the geometry of the valve that determines
the ratio of screen and anode currents and gms?

That is an intersting question. Do you know the answer?

No. I almost don't dare ask :-)

The knee is at a remarkably low voltage, and high current, and there's
no kink. A power valve like that would be an instant candidate for
pentode operation with cathode feedback, in my amps perhaps, to
replace the much gentler 6CH6. It would be interesting to try pp 6AU6
as a driver or headphone amp with a transformer: perhaps 30% cathode
winding and cross-coupled screens to 30% taps to maintain constant
Vks.


I think this has already been done. I have done lots of Googling for 6AU6
and I am pretty sure I came across one guitar amp and one headphones amp
based on the 6AU6.

But perhaps there is a necessary link between the knee position and
the high screen current. Perhaps the screen is relatively close to the
grid. That might also explain why small differences in the details of
construction between different examples lead to relatively large
variations in characteristics.


As far as my rudimentray knowledge goes, gm of a particular tube depends
primarily on two things. First the relative distances of the cathode, grid
and anode and, secondly, the spacing of the turns on the grid. I have
recently purchased over 50 6AU6 tubes from various sources and checked them
all on my AVO tester under the same conditions. I know there can be
considerable variation between tubes of the same type even from the same
manufacturer but this lot of 50 seem to fall into three groups. The largest
group ( about 35) show a gm of about 3.9mA/V. The second group (about 12)
shows a gm of over 5mA/V. The smallest group (of 3) are probably worn out
and give a gm reading of 2mA/V or less. Ignoring the duffers there are
clearly two typical gm readings. I have looked at the construction and as
yet cannot draw any conclusion as to group and construction other than the
higher gm tubes are mainly about a quarter of an inch taller than the
others.

I seriously don't dare ask why sharp cut-off pentodes are, er, sharp
cut-off, and in what particular circumstances it is advantageous for
them to be so. I'm tired of getting my ankles savaged.


They are sharp cut off because the grid turns are equally spaced so the grid
volts at which the electron stream gets cut off is the same everywhere so
the cut off occurs quite sharply. Variable mu or remote cut off tubes have
a grid wound with variable spacing between turns. So the closer turns reach
cut off first but only shut of part of the electron stream. Wider spaced
turns cut off at higher voltages. The overall effect is a much more gradual
cut off.

Ian (TB)

[[email protected]]

Patrick Turner wrote:



wrote:

Patrick Turner wrote:



Ian Iveson wrote:

Other Ian said

I don't see that but maybe I am looking at the wrong part of the
data sheet.
The part I see under 'Characteristics and Typical Operation' has a
column
with screen and plate volts both 100V and a gm of 3.9mA/V. These are
the
voltages my AVO tester is set up with for meauding the 6AU6 in
pentode
mode. The plate current is 5mA and the screen current is 2.1mA,
about 30%
of cathode current - OK that's higher than my 20% figure but not
hugely.
This gives a triode equivalent gm of 3.9 x 7.1/5 = 5.5mA/V, still
rather
lower than the 7 I was measuring.

I can see in the right hand column where the plate is at 250V and
the screen
at 150 that the gm rises to 5.2. The triode equivalent under these
conditions works out at about 7.2mA/V which is much closer to what I
measured but with rather different voltages.

Firstly, 30% is much greater than 20% in my book. 50% greater, in
fact.

My data book for 6AU6 has the following data,
Ig2 = 4.3mA, Eg2 = 150V, when Ia = 10.6mA, Ea 250V, gm = 5.2mA/V,
And
Ig2 = 2.1mA, Eg2 = 100mA when Ia = 5.0mA, Ea = 100V, gm = 3.9mA/V

So over a broad range of voltages, Ig2 = 40% of Ia.

However, if you go to the RCA sheets with 6AU6 TRIODE curves, at Ia =
10mA and Ea = 250V
the µ = 38.6 and Ra = 9.1k, and gm = µ / Ra = 38.6 / 9,100 = 4.24 mA/V

Who was it who told me that triode gm was HIGHER than pentode
gm????????


Actually is was page 34 of RDH4.

In theory this is right. RDH4 says the gm is shared with respect to the
proportion of Ig2 and Ia.

But if we consider the real world use of pentodes like the 6AU6,
then we'd expect gm in triode to be 40% due to screen, and 60% due to
anode,
if that's the ratio of Ig2 to Ia, and when in triode, the Ea = Eg2.

So you'd think it could be compared with a pentode set up where Ea = Eg2
in the idle condition.

And one would expect that the pentode gm would be a lot less the the
triode connection,
but anode voltage moves and screen voltage does not, and just exactly
what you will get depends on someone doing a test on a 6AU6 so that we
don't just have to take a tiny little
paragraph from RDH4 as gospel.


That will be me. I just got emails to say the last of the parts I ordered
have been despatched. So patrick, what is the simplest, reasonably accurate
way for me to measure the 6AU6 gm in triode mode?

At the top of page 36 RDH4 mentions the 6AU6, and in pentode
you get Ra = 1.5M, gm = 4.45mA/V.
µ = gm x Ra = 0.00445 x 1,500,000 = 6,675.

Triode µ is given as 36, and THEREFORE screen µ = 6,675 / 36 = 185.

OK so far, and that's all RDH4 says about 6AU6 in this part of the book.

If therefore we say that gm of the screen = µ / Ra,
then gm g2 = 185 / 1,500,000 = 0.123mA/V, but this sounds way too low.

So what is the Ra of the 6AU6 when driven as a triode with fixed g1
voltage and drive to g2???

RDH4 doesn't fill in the gaps AT ALL. MUCH more could be said but isn't.

So to know what you really get you have to set up a tube and measure
its gain at low output voltages for negligibe distortions, say 10Vrms
will do,
and use an accurate volt meter for input and output voltage, and use two
RLs
with one twice the value of the other, and finally you can get two gain
equations to
work out Ra, µ and hence gm for any tube, triode or pentode or g2 drive
with g1 at say 0V.


OK, you just answered my earlier question. So if I use 47K as one anode
resistor and a pair in series for the other, keep the cathode resistor
(bypassed) the same in both cases and measure the output with my VTVM we
should be good to go. I have a good quality stable oscillator so I can
easily set the input voltage to avoid distortion at the output.

To understand the tubes this way you have to be just as clever and
disciplined as the authors of RDH4.


A tall order.

It isn't very difficult, but it just takes time, and you have to build a
decent test rig.


Yup, I am.

Ian

[Patrick Turner]

wrote:

Patrick Turner wrote:



wrote:

Patrick Turner wrote:



Ian Iveson wrote:

Other Ian said

I don't see that but maybe I am looking at the wrong part of the
data sheet.
The part I see under 'Characteristics and Typical Operation' has a
column
with screen and plate volts both 100V and a gm of 3.9mA/V. These are
the
voltages my AVO tester is set up with for meauding the 6AU6 in
pentode
mode. The plate current is 5mA and the screen current is 2.1mA,
about 30%
of cathode current - OK that's higher than my 20% figure but not
hugely.
This gives a triode equivalent gm of 3.9 x 7.1/5 = 5.5mA/V, still
rather
lower than the 7 I was measuring.

I can see in the right hand column where the plate is at 250V and
the screen
at 150 that the gm rises to 5.2. The triode equivalent under these
conditions works out at about 7.2mA/V which is much closer to what I
measured but with rather different voltages.

Firstly, 30% is much greater than 20% in my book. 50% greater, in
fact.

My data book for 6AU6 has the following data,
Ig2 = 4.3mA, Eg2 = 150V, when Ia = 10.6mA, Ea 250V, gm = 5.2mA/V,
And
Ig2 = 2.1mA, Eg2 = 100mA when Ia = 5.0mA, Ea = 100V, gm = 3.9mA/V

So over a broad range of voltages, Ig2 = 40% of Ia.

However, if you go to the RCA sheets with 6AU6 TRIODE curves, at Ia =
10mA and Ea = 250V
the µ = 38.6 and Ra = 9.1k, and gm = µ / Ra = 38.6 / 9,100 = 4.24 mA/V

Who was it who told me that triode gm was HIGHER than pentode
gm????????


Actually is was page 34 of RDH4.

In theory this is right. RDH4 says the gm is shared with respect to the
proportion of Ig2 and Ia.

But if we consider the real world use of pentodes like the 6AU6,
then we'd expect gm in triode to be 40% due to screen, and 60% due to
anode,
if that's the ratio of Ig2 to Ia, and when in triode, the Ea = Eg2.

So you'd think it could be compared with a pentode set up where Ea = Eg2
in the idle condition.

And one would expect that the pentode gm would be a lot less the the
triode connection,
but anode voltage moves and screen voltage does not, and just exactly
what you will get depends on someone doing a test on a 6AU6 so that we
don't just have to take a tiny little
paragraph from RDH4 as gospel.


That will be me. I just got emails to say the last of the parts I ordered
have been despatched. So patrick, what is the simplest, reasonably accurate
way for me to measure the 6AU6 gm in triode mode?

Just set one up in common cathode with 39k load from a B+ of 250V and
adjust the bypassed
Rk so that Ea = 100V so that Ia will be 3.8mA.

Have a 470k grid bias R and RCA socket to accept a signal from your sig
gene.

Have 0.47uF from the anode to another RCA socket to take the anode
signal out to a CRO,
and to a point where you can measure the signal voltage.

The total RL load can easily be halved from the 39k carrying dc to 19.5k
by simply
adding another 39k from the earthy side of the 0.47uF and 0V.

Apply a 1kHz sine wave, adjust initially for 5Vrms, THD should be under
1% for all tests.

Keep the input level constant at whatever is needed for 5Vrms output
and measure the grid signal with only 39k present as the dc load. Gain,
A = Vout / Vin.

Without changing input signal, add the second 39k and measure the Vout,
and find the gain for 19.5k.

The rest you can work out because you will have two equations for A = µ
x RL / ( RL + Ra ),
one for each RL, and you know A and RL in both cases, and from agebraic
substituition
you can find µ and Ra easily. gm = µ / Ra.




At the top of page 36 RDH4 mentions the 6AU6, and in pentode
you get Ra = 1.5M, gm = 4.45mA/V.
µ = gm x Ra = 0.00445 x 1,500,000 = 6,675.

Triode µ is given as 36, and THEREFORE screen µ = 6,675 / 36 = 185.

OK so far, and that's all RDH4 says about 6AU6 in this part of the book.

If therefore we say that gm of the screen = µ / Ra,
then gm g2 = 185 / 1,500,000 = 0.123mA/V, but this sounds way too low.

So what is the Ra of the 6AU6 when driven as a triode with fixed g1
voltage and drive to g2???

RDH4 doesn't fill in the gaps AT ALL. MUCH more could be said but isn't.

So to know what you really get you have to set up a tube and measure
its gain at low output voltages for negligibe distortions, say 10Vrms
will do,
and use an accurate volt meter for input and output voltage, and use two
RLs
with one twice the value of the other, and finally you can get two gain
equations to
work out Ra, µ and hence gm for any tube, triode or pentode or g2 drive
with g1 at say 0V.


OK, you just answered my earlier question. So if I use 47K as one anode
resistor and a pair in series for the other, keep the cathode resistor
(bypassed) the same in both cases and measure the output with my VTVM we
should be good to go. I have a good quality stable oscillator so I can
easily set the input voltage to avoid distortion at the output.

For keeping the Ea constant while changing RL, use the method
of a constant dc RL and make the change to RL via the cap coupled load.
Its simple and easy.



To understand the tubes this way you have to be just as clever and
disciplined as the authors of RDH4.


A tall order.

No, the crew of guys who wrote RDH4 were just ordinary guys with
very ordinary technical training.

The RDH4 seems lofty, but was written for the average technical person,
not for the post graduate honours student at uni.


It isn't very difficult, but it just takes time, and you have to build a
decent test rig.


Yup, I am.

Good,

Patrick Turner.

Ian

[[email protected]]

Patrick Turner wrote:


Just set one up in common cathode with 39k load from a B+ of 250V and
adjust the bypassed
Rk so that Ea = 100V so that Ia will be 3.8mA.

Have a 470k grid bias R and RCA socket to accept a signal from your sig
gene.

Have 0.47uF from the anode to another RCA socket to take the anode
signal out to a CRO,
and to a point where you can measure the signal voltage.

The total RL load can easily be halved from the 39k carrying dc to 19.5k
by simply
adding another 39k from the earthy side of the 0.47uF and 0V.

Apply a 1kHz sine wave, adjust initially for 5Vrms, THD should be under
1% for all tests.

Keep the input level constant at whatever is needed for 5Vrms output
and measure the grid signal with only 39k present as the dc load. Gain,
A = Vout / Vin.

Without changing input signal, add the second 39k and measure the Vout,
and find the gain for 19.5k.

The rest you can work out because you will have two equations for A = µ
x RL / ( RL + Ra ),
one for each RL, and you know A and RL in both cases, and from agebraic
substituition
you can find µ and Ra easily. gm = µ / Ra.



Thanks Patrick. I think I'll add a switch to include/exclude the additional
39K resistor. My supply is currently 350V so I might use 47K which give Ia
of 5.3mA with Ea of 100V OR I might add a dropper and some extra smoothing
to bring Ebb down to 250V. That would be handy later when making noise
measurements. So a 27K series resistor and 47uF smoothing should do it.

Many thanks for your help so far.

Cheers

Ian (TB)

[Ian Iveson]

Ian wrote:

According to the circuit I have there is a rectifier in the screen
connection too.


Right, thanks, should have looked...it's been ages since I used it.
Just half wave for the screen, presumably, or it would get a battering
during the half cycle when the anode is -ve.


The knee is at a remarkably low voltage, and high current, and
there's
no kink. A power valve like that would be an instant candidate for
pentode operation with cathode feedback, in my amps perhaps, to
replace the much gentler 6CH6. It would be interesting to try pp
6AU6
as a driver or headphone amp with a transformer: perhaps 30%
cathode
winding and cross-coupled screens to 30% taps to maintain constant
Vks.

I think this has already been done. I have done lots of Googling for
6AU6
and I am pretty sure I came across one guitar amp and one headphones
amp
based on the 6AU6.

Pentode with cathode feedback?

But perhaps there is a necessary link between the knee position and
the high screen current. Perhaps the screen is relatively close to
the
grid. That might also explain why small differences in the details
of
construction between different examples lead to relatively large
variations in characteristics.

As far as my rudimentray knowledge goes, gm of a particular tube
depends
primarily on two things. First the relative distances of the
cathode, grid
and anode and, secondly, the spacing of the turns on the grid. I
have
recently purchased over 50 6AU6 tubes from various sources and
checked them
all on my AVO tester under the same conditions. I know there can be
considerable variation between tubes of the same type even from the
same
manufacturer but this lot of 50 seem to fall into three groups. The
largest
group ( about 35) show a gm of about 3.9mA/V. The second group
(about 12)
shows a gm of over 5mA/V. The smallest group (of 3) are probably
worn out
and give a gm reading of 2mA/V or less. Ignoring the duffers there
are
clearly two typical gm readings. I have looked at the construction
and as
yet cannot draw any conclusion as to group and construction other
than the
higher gm tubes are mainly about a quarter of an inch taller than
the
others.

Right. And cathode area, material, and temperature, presumably,
considering that gm changes with age. But I was particularly wondering
about the current or gm split between screen and anode.

Do the taller valves have longer innards?

I wonder if, in middle age, these valves become closer? Is it better
to design for particular examples of valves, at particular points in
their lives, or for the species? If the latter, then it might make
sense to use the published data rather than actual tested values.

I seriously don't dare ask why sharp cut-off pentodes are, er,
sharp
cut-off, and in what particular circumstances it is advantageous
for
them to be so. I'm tired of getting my ankles savaged.

They are sharp cut off because the grid turns are equally spaced so
the grid
volts at which the electron stream gets cut off is the same
everywhere so
the cut off occurs quite sharply. Variable mu or remote cut off
tubes have
a grid wound with variable spacing between turns. So the closer
turns reach
cut off first but only shut of part of the electron stream. Wider
spaced
turns cut off at higher voltages. The overall effect is a much more
gradual
cut off.


OK, thanks. How is sharpness desirable in some circuits, but not
others, I wonder?

cheers, Ian1

[[email protected]]

Ian Iveson wrote:

Ian wrote:

According to the circuit I have there is a rectifier in the screen
connection too.


Right, thanks, should have looked...it's been ages since I used it.
Just half wave for the screen, presumably, or it would get a battering
during the half cycle when the anode is -ve.


The knee is at a remarkably low voltage, and high current, and
there's
no kink. A power valve like that would be an instant candidate for
pentode operation with cathode feedback, in my amps perhaps, to
replace the much gentler 6CH6. It would be interesting to try pp
6AU6
as a driver or headphone amp with a transformer: perhaps 30%
cathode
winding and cross-coupled screens to 30% taps to maintain constant
Vks.

I think this has already been done. I have done lots of Googling for
6AU6
and I am pretty sure I came across one guitar amp and one headphones
amp
based on the 6AU6.

Pentode with cathode feedback?


Not sure about the guitar amp but I copied the headphone amp article. I used
a pentode connected 6AU6 with unbypassed cathode as a SE output with global
NFB to the input stage which was one half of a 12AU7

But perhaps there is a necessary link between the knee position and
the high screen current. Perhaps the screen is relatively close to
the
grid. That might also explain why small differences in the details
of
construction between different examples lead to relatively large
variations in characteristics.

As far as my rudimentray knowledge goes, gm of a particular tube
depends
primarily on two things. First the relative distances of the
cathode, grid
and anode and, secondly, the spacing of the turns on the grid. I
have
recently purchased over 50 6AU6 tubes from various sources and
checked them
all on my AVO tester under the same conditions. I know there can be
considerable variation between tubes of the same type even from the
same
manufacturer but this lot of 50 seem to fall into three groups. The
largest
group ( about 35) show a gm of about 3.9mA/V. The second group
(about 12)
shows a gm of over 5mA/V. The smallest group (of 3) are probably
worn out
and give a gm reading of 2mA/V or less. Ignoring the duffers there
are
clearly two typical gm readings. I have looked at the construction
and as
yet cannot draw any conclusion as to group and construction other
than the
higher gm tubes are mainly about a quarter of an inch taller than
the
others.

Right. And cathode area, material, and temperature, presumably,
considering that gm changes with age. But I was particularly wondering
about the current or gm split between screen and anode.

Do the taller valves have longer innards?


No they are pretty much the same, they just have an extra bit of metal like
a brolly over the top of the whole assembly.

I wonder if, in middle age, these valves become closer? Is it better
to design for particular examples of valves, at particular points in
their lives, or for the species? If the latter, then it might make
sense to use the published data rather than actual tested values.


Depends on the performance you want to achieve. For some applications it
seems it is necessary or a least a good idea to select tubes e.g first
stage of preamps, but elsewhere almost any example will suffice.

I seriously don't dare ask why sharp cut-off pentodes are, er,
sharp
cut-off, and in what particular circumstances it is advantageous
for
them to be so. I'm tired of getting my ankles savaged.

They are sharp cut off because the grid turns are equally spaced so
the grid
volts at which the electron stream gets cut off is the same
everywhere so
the cut off occurs quite sharply. Variable mu or remote cut off
tubes have
a grid wound with variable spacing between turns. So the closer
turns reach
cut off first but only shut of part of the electron stream. Wider
spaced
turns cut off at higher voltages. The overall effect is a much more
gradual
cut off.


OK, thanks. How is sharpness desirable in some circuits, but not
others, I wonder?



I suspect winding regular spaced grids is easiest and that was how it was
done at first. Later, when someone wanted an AGC circuit, a variable
spacing technique was devised.


Ian (TB)

[Chris Hornbeck]

On Sat, 30 Jun 2007 14:40:11 GMT, "Ian Iveson"
wrote:

OK, thanks. How is sharpness desirable in some circuits, but not
others, I wonder?

Sharpness implies lower second harmonic distortion. Maybe
suprisingly, this isn't something always desirable in tuned
RF amplifiers. The simultaneous gain reduction and input
signal handling increase with a simple change of bias of
"remote cutoff" valves is a handy combination for linear
amplifiers operating in a low dynamic range environment with
a large dynamic range of possible inputs.

Third harmonic distortion *is* important for RF amplifiers
though, because it causes crossmodulation. Variable-mu
valves can be made to have low third and high second
harmonic distortion.

Thanks, as always,

Chris Hornbeck
"Money doesn't buy happiness. But happiness isn't everything."
- Jean Seberg

[Patrick Turner]

wrote:

Patrick Turner wrote:


Just set one up in common cathode with 39k load from a B+ of 250V and
adjust the bypassed
Rk so that Ea = 100V so that Ia will be 3.8mA.

Have a 470k grid bias R and RCA socket to accept a signal from your sig
gene.

Have 0.47uF from the anode to another RCA socket to take the anode
signal out to a CRO,
and to a point where you can measure the signal voltage.

The total RL load can easily be halved from the 39k carrying dc to 19.5k
by simply
adding another 39k from the earthy side of the 0.47uF and 0V.

Apply a 1kHz sine wave, adjust initially for 5Vrms, THD should be under
1% for all tests.

Keep the input level constant at whatever is needed for 5Vrms output
and measure the grid signal with only 39k present as the dc load. Gain,
A = Vout / Vin.

Without changing input signal, add the second 39k and measure the Vout,
and find the gain for 19.5k.

The rest you can work out because you will have two equations for A = µ
x RL / ( RL + Ra ),
one for each RL, and you know A and RL in both cases, and from agebraic
substituition
you can find µ and Ra easily. gm = µ / Ra.


Thanks Patrick. I think I'll add a switch to include/exclude the additional
39K resistor. My supply is currently 350V so I might use 47K which give Ia
of 5.3mA with Ea of 100V OR I might add a dropper and some extra smoothing
to bring Ebb down to 250V. That would be handy later when making noise
measurements. So a 27K series resistor and 47uF smoothing should do it.

Many thanks for your help so far.

Cheers

Ian (TB)

Having switchable R for the cap coupled load makes it easier to make
quicker comparisons.


For testing pentodes with Ea = Eg2, perhaps you could have a zener diode
shunt regulated +100V
to supply the screen, ie, dropper R from B+ to +100V, and a 22uF to 0V
as well.
Then adjust Ea by adjusting Rk so maybe you need a wire wound pot of
500ohms in series with say 220 ohms for
Rk, and a 470uF bypass cap. Carbon track pots are notoriously unreliable
in dc flow conditions
so don't ever use on here!

The Rk pot can be shunted with say 500ohms to give a finer adjustment...


I have used a choke for testing power tubes because I have some around
of suitable L value
but its not necessary with a signal tube.

One can also set up a signal tube with a transistor CCS using one
MJE350.

But then you'll find the gain without any low value of RL connected will
be
enormous, and theoretically = µ = gm x Ra, and in the case of 6AU6 could
be 6,600.
But the stray capacitance of the leads to voltmeters and the R loads of
this measuring gear
will drastically reduce the gain you are trying to measure, because the
Rout = Ra of the pentode
which could be 1.5M so a 1.5M measuring load will halve the voltage at
the anode.

To avoid such losses, another triode needs to be set up with its grid
taken directly to the pentode
anode and a cathde R to 0V; another 6AU6 in triode as a CF will do fine
for this, and then the
gain with a CCS can actually be measured by measuring the signal at the
CF cathode output
which has Rout = 1 / gm approx = 250 ohms, and the meter R and CRO leads
et all won't affect the
measurements at 1 kHz.
Meaurements for gain and the anode Ea can be taken at the CF cathode,
because gain tube anode Ea will be
about 2V below the CF cathode, so they are close enough to being the
same for your purposes.

What you may find with a CCS dc supply is that adjusting the cathode
resistor to set the idle Ea voltage
in pentode with G2 set at 100V becomes a very fiddly tricky business
because an Ek variation of only 1.5mV can cause 10V variation in Ea.
I have tried all this and was frustrated at each step.
One way it better adjust Ea is to establish a shunt regulated -100V
supply rail,
and have the gain tube Rk taken down to the -100V, and THEN you will
find that slight adjustment in the
much higher value of Rk will allow you more control of setting EA and
witnessing the
extraordinarily high possible voltage gain of a pentode.
The gain tube cathode is still bypassed with the same 470uF to 0V, and
grid biased at 0V.

With a CCS set up as the only load, there will be a huge amount of noise
at the anode,
both from the noise of the input bias R being amplified thousands of
times,
and from hum from heaters.
So use clean DC on the heaters!!!
The input noise of the pentode also will be there, so you must monitor
with a CRO all the time to ensure
your voltage meters are not telling you bull****.


If you do ever SEE gain with a CCS = 6,600, and you only have 10Vrms
output,
then the input grid signal = 1.5mV only, and measuring this tiny drive
signal
is also VERY prone to noise and inaccuracies, so a resistance divider of
100:1
nneds to be made using 1,000 ohms and 10 ohms, and if you have 0.15Vrms
at the top of the divider,
there is 1.5mV at the top of the 10 ohms, and the 10 ohms has low noise.

Short neat wiring is imperative.

You can still have a cap coupled load to the anode of the pentode and
still use the CF
to BUFFER what you are testing and avoid the errors caused by impedance
of the measuring gear.

Where you do realise a gain of over 6,000 for a 6AU6 with CCS then when
you connect
say 39k cap coupled RL, gain will perhaps be only 170,
and connecting the RL load will seem to make the anode signal
dissappear.

Soon you will see how difficult it is to duplicate and witness the raw µ
of a tube like the 6AU6.
Maybe you'll think its pointless to try.
That was my final conclusion, and soon all i did was just test with two
RL
of about 1:2 ratio, knowing that high pentode µ is a fact, but there
isn't any need to see it for real.

A 6AU6 set up with a CCS load also acts queerly with any change of
signal because the slight THD
in the signal translates to a slight change in Ia so the Ea of the tube
under test
swings up and down with changes of Va. So there is not much practical
application of using
a signal pentode with pure CCS in any circuit, because of the dc drift.
It is however possible to build a high RL value approaching a CCS which
will give
high gain, say 1,000, and it involves making the MJE350 act as a low R
of say 50k at DC but about 250k at signal F.

It means a R divider between anode to base to B+, say 500k and 50k.
the 50k is bypassed with say 5uF, so at signal F the anode sees 500k,
but at dc the
divider control the base voltage controlling the emitter current, and
the effective RL at dc = 9 times the Re.If Re = 3.3k, then it
it is made to appear as 30k to the anode at dc.

You attitude should always be that what you are doing is never up to par
and never
very wonderful until you prove to yourself that it really is OK.
When you have done the calculations for a pair of RL and have attained µ
and Ra, insert the values
into equations for the gain and test if they work OK. So what you
measured for gain
should be able to be calculated in both RL cases.
If this works out, you have achieved the mission, if not, then its back
to wondering what you done wrong.

To know a vacuum tube you must strip her bare and wonder at her
qualities.
A pentode is like a reluctant girl, shy, and full of capricious
behaviours.

Patrick Turner.

[Ian Iveson]

Chris explained:

OK, thanks. How is sharpness desirable in some circuits, but not
others, I wonder?

Sharpness implies lower second harmonic distortion. Maybe
suprisingly, this isn't something always desirable in tuned
RF amplifiers. The simultaneous gain reduction and input
signal handling increase with a simple change of bias of
"remote cutoff" valves is a handy combination for linear
amplifiers operating in a low dynamic range environment with
a large dynamic range of possible inputs.

Third harmonic distortion *is* important for RF amplifiers
though, because it causes crossmodulation. Variable-mu
valves can be made to have low third and high second
harmonic distortion.

Aha. Interesting. Thanks for the clarity.

cheers, Ian1

[Chris Hornbeck]

On Sat, 30 Jun 2007 03:24:55 -0500, flipper wrote:

As I said, the canonical system has only one input.

Look at a tube datasheet and, in particular, Plate V and grid V.
That's 2.

Pay close attention, folks. Here is where the argument
takes its big twist.

The canonical model has an input that arrives at a summing point, from
which a forward function leads to the output. From the output, a
feedback function leads to the same single summing point.

Notice there can be one input only.

Not two, or three, or any other multiplicity.

Just one.

Guess we'll have to throw away all those opamps with a + and - input.

Mathematically, opamps have a single, differential, input
with two terminals.

The interesting (apparently only to me) bit is that your
model has an output which is also an input (anode). This element
differs from the classic model, and so IMO should be the
next point of discussion.

Thanks, as always,

Chris Hornbeck
"Money doesn't buy happiness. But happiness isn't everything."
- Jean Seberg

[Patrick Turner]

Chris Hornbeck wrote:

On Sat, 30 Jun 2007 03:24:55 -0500, flipper wrote:

As I said, the canonical system has only one input.

Look at a tube datasheet and, in particular, Plate V and grid V.
That's 2.

Pay close attention, folks. Here is where the argument
takes its big twist.

The canonical model has an input that arrives at a summing point, from
which a forward function leads to the output. From the output, a
feedback function leads to the same single summing point.

Notice there can be one input only.

Not two, or three, or any other multiplicity.

Just one.

Guess we'll have to throw away all those opamps with a + and - input.

Mathematically, opamps have a single, differential, input
with two terminals.

The interesting (apparently only to me) bit is that your
model has an output which is also an input (anode). This element
differs from the classic model, and so IMO should be the
next point of discussion.

Suppose you have a shunt NFB network between an anode and a grid with a
pair of resistors,
and with cathode grounded.
Two of the conventionally considered terminals, grid and cathode, are
input terminals.

The anode is the conventional output terminal but in a shunt NFB array
it sends a signal back to the grid
and thus contributes input to the grid. Thus indirectly, it is an input
terminal.

In a cathode follower, output is from a cathode, usually a low impedance
input terminal.
ALL the output signal is fed back so cathode inputs signal relative to
the grid signal.

Similarly a follower opamp does a similar job to the CF.

Classical Model? this is 2007, not Greece in 500BC.

But there always has to be two input terminals, at least a ground and
active.
A "classical" application of a differential amp is in nearly every tube
amp with global NFB
where the cathode has say a 1V sample of the output signal applied as
NFB at low Z and the
and the grid has 1.2V applied at high Z as input signal. Only the
difference
between the two, 0.2V, is amplified.

We need to differentiate between being too sloppy with terminology,
and to rigidly classical.
Diferrentiation is part of a balanced mind and ideas.

Patrick Turner.





Thanks, as always,

Chris Hornbeck
"Money doesn't buy happiness. But happiness isn't everything."
- Jean Seberg

[robert casey]

...Two voltage inputs, grid and plate, with a current output...

As I said, the canonical system has only one input.


Look at a tube datasheet and, in particular, Plate V and grid V.
That's 2.



And the amount of resulting plate current is the output. That's legit.
Now use a resistor on the plate to B+. The current via Ohm's law will
cause a voltage drop to happen on the plate, thus reducing the plate
voltage. Which in turn will affect the output current. Feedback!
(ducking, running for cover) :-)

[Chris Hornbeck]

On Tue, 03 Jul 2007 07:56:41 GMT, Patrick Turner
wrote:

Suppose you have a shunt NFB network between an anode and a grid with a
pair of resistors,
and with cathode grounded.
Two of the conventionally considered terminals, grid and cathode, are
input terminals.

The anode is the conventional output terminal but in a shunt NFB array
it sends a signal back to the grid
and thus contributes input to the grid. Thus indirectly, it is an input
terminal.

In a cathode follower, output is from a cathode, usually a low impedance
input terminal.
ALL the output signal is fed back so cathode inputs signal relative to
the grid signal.

But what does any of this have to do with the operation
of a vacuum valve?


Similarly a follower opamp does a similar job to the CF.

Therefore opamps have internal feedback?

Sorry, but I still find available arguments silly.

Thanks, as always,

Chris Hornbeck
"Money doesn't buy happiness. But happiness isn't everything."
- Jean Seberg

[Chris Hornbeck]

On Tue, 03 Jul 2007 17:48:00 -0500, flipper wrote:

The answer is yes and no, but not quite. Functionally, the plate
'output' is current while the plate 'input' (forgive the 'canonical
confusion', because we aren't to that point yet) is voltage. So the
'no' part is they're different signals.

That the two are on 'the same pin' (the 'yes' part) is a physical
aspect of the device but 'more than one' signal, and more than one
direction, on a single wire, or pin, is not something unique to the
triode as engineers often intentionally design such things. I.E.
multiplexing, which also includes time division (but not applicable to
this case)..

Point is, it is, as you rightly contended in the opamp case, the
'mathematical' that determines, not the physical. (nor that ST uses
the word 'input' on two pins).

That both plate current output and plate voltage input are on the
'same pin', by nature of the device, leads to the argument its
'internal', a claim I take partial dispute with because there's a
'missing piece' to finish the 'loop': something to convert the current
output to a voltage representation which is, then, applied to the
plate: I.E. the plate load. And that is external.

I'd wager that most engineers would have no trouble at all with an
integrated circuit constructed in a similar manner, I.E.pin
multiplexing with an 'internal' connection forming the feedback loop
and I imagine that's because they'd be happy envisioning 'lumped
circuits' inside the thing and, perhaps, as someone amusingly
postulated, 'confused' by 'magical field effects' but microwave
designers have no such mystification about designing, or
comprehending, distributed parameters and they are no less 'real' than
lumped models of the real field effects.

And then, lastly, we have the trivial case that no one seems to have
any problem whatsoever recognizing "cathode feedback" even though the
'output' and (feedback) 'input' (for that local loop) are on 'the same
pin', nor that the 'voltage' is created by an external load converting
output current to voltage, that the cathode is the summing junction,
and that the whole thing works because of 'mystical field effects'
just like all electronic devices do.

But what does any of this have to do with the operation
of a vacuum valve?

And contrary to the argument it 'adds nothing' I find it eminently
useful to, among other things, illuminating why a pentode
'mysteriously' behaves like a triode when so connected, as opposed to
what I consider un illuminating arm waving that it behaves like a
triode because it now is one and, well, that's just 'how triodes
behave' vs. 'how pentodes behave." To me, that is like asking why gray
clouds often produce rain while white ones generally don't where the
answer given is because that's what gray and white clouds do.

I hope none of that sounds 'ungrateful'. I'm just trying to explain
how I see it and why.

I'm sorry, but I still find the available arguments silly.

Thanks, as always,

Chris Hornbeck
"Money doesn't buy happiness. But happiness isn't everything."
- Jean Seberg

[Chris Hornbeck]

On Tue, 03 Jul 2007 23:07:08 -0500, flipper wrote:

But what does any of this have to do with the operation
of a vacuum valve?

Frankly, I don't know how to answer what the operation of a vacuum
valve has to do with the operation of a vacuum valve as this seems
rather self evident.

Maybe if you could explain how you come to the conclusion that cathode
feedback in a triode has nothing to do with how a triode operates I
could get a better idea of what you're trying to ask.

Cathode feedback, or any other external characteristic, has
*nothing* to do with the operation of the triode. Nothing.

Do the triode's characteristics differ in *any* way when
operated as a cathode follower, or indeed in *any* external
configuration? Are it's plate curves changed? Is there any
way that you could observe, while viewing only its three
(yeah, yeah, but you know what I mean) terminals that the
triode was being used as a cathode follower, or in any other
external configuration?

If your model of internal feedback is to be convincing, it
will need to free itself of external (and, frankly, shakey)
props.


I'm sorry, but I still find the available arguments silly.

I'm sorry but just declaring something 'silly', and nothing else,
isn't illuminating..

"Feedback" is a word with meaning established by folks smarter
than us and back before we were born. Anybody wishing to make
a special case for their pet theory will have to get used to a
certain level of scepticism. The word has an established
meaning in engineering and attempts to redefine it will get
short shrift. Can you make a convincing case for your model
that doesn't include any special exceptions?

Anything less just wouldn't be the Cowboy Way.

Thanks, as always,

Chris Hornbeck
"Money doesn't buy happiness. But happiness isn't everything."
- Jean Seberg

[Patrick Turner]

Chris Hornbeck wrote:

On Tue, 03 Jul 2007 07:56:41 GMT, Patrick Turner
wrote:

Suppose you have a shunt NFB network between an anode and a grid with a
pair of resistors,
and with cathode grounded.
Two of the conventionally considered terminals, grid and cathode, are
input terminals.

The anode is the conventional output terminal but in a shunt NFB array
it sends a signal back to the grid
and thus contributes input to the grid. Thus indirectly, it is an input
terminal.

In a cathode follower, output is from a cathode, usually a low impedance
input terminal.
ALL the output signal is fed back so cathode inputs signal relative to
the grid signal.

But what does any of this have to do with the operation
of a vacuum valve?

Lots, if you happened to notice the repeated descriptions
of vacuum valves ( tubes ) and ways of hooking them up....

Similarly a follower opamp does a similar job to the CF.

Therefore opamps have internal feedback?

Some indeed to have internal FB, but its not like that of a triode.


Sorry, but I still find available arguments silly.

Ah, so the unavailable arguments are the wise ones...

Patrick Turner.

Thanks, as always,

Chris Hornbeck
"Money doesn't buy happiness. But happiness isn't everything."
- Jean Seberg

[Patrick Turner]

Chris Hornbeck wrote:

On Tue, 03 Jul 2007 23:07:08 -0500, flipper wrote:

But what does any of this have to do with the operation
of a vacuum valve?

Frankly, I don't know how to answer what the operation of a vacuum
valve has to do with the operation of a vacuum valve as this seems
rather self evident.

Maybe if you could explain how you come to the conclusion that cathode
feedback in a triode has nothing to do with how a triode operates I
could get a better idea of what you're trying to ask.

Cathode feedback, or any other external characteristic, has
*nothing* to do with the operation of the triode. Nothing.

Its true that a tube or valve has no consciousness of the external
circuit around itself.

We do, and a distinction does need to be made with regard to
internal NFB and external.

It is by external observation that those of us with sufficient
consciousness
will conclude that triodes have internal NFB and are not just a
resistor.



Do the triode's characteristics differ in *any* way when
operated as a cathode follower, or indeed in *any* external
configuration? Are it's plate curves changed? Is there any
way that you could observe, while viewing only its three
(yeah, yeah, but you know what I mean) terminals that the
triode was being used as a cathode follower, or in any other
external configuration?

If your model of internal feedback is to be convincing, it
will need to free itself of external (and, frankly, shakey)
props.

Plenty of models have bee drawn in RDH4 of the changed Ra curves
for a CF or a tube with an unbypassed Rk, but this is to illustrate the
result
of exernal NFB connection.
No matter how you connect a triode, the internal FB remains, and
the equivalent basic circuit of low Z generator and series R
representing Ra all remain the same.



I'm sorry, but I still find the available arguments silly.

I'm sorry but just declaring something 'silly', and nothing else,
isn't illuminating..

"Feedback" is a word with meaning established by folks smarter
than us and back before we were born.

NFB wasn't known well at first by our ancestors.
Although very intelligent, none were much different to us
and we are as silly or wise as they ever were.

But NFB became something usable, I recall a Mr Black
began publishing aspects of NFB useage in 1928.
The first tubes were all triodes, and NFB wasn't a term thought much
about until
thery tried to amplify higher and higher frequencies, and then the
C between anode and grid meant the input impedance to a triode became
less as F rose, limiting frequencies much above a MHz.
AFAIK, the boffins of 1920 had a fair idea of the
summed joint actions of electrostatic fields due to grid and anode
voltage changes.
I suggest you bury your head in the shelves of books available for study
from the time vacuum tubes were first developed.

Tetrodes were invented to INTERRUPT the NFB electrostatic field effect
of the anode upon its own electron flow and to reduce the Cga
capacitance, ie,
reduce the Miller effect.



Anybody wishing to make
a special case for their pet theory will have to get used to a
certain level of scepticism. The word has an established
meaning in engineering and attempts to redefine it will get
short shrift. Can you make a convincing case for your model
that doesn't include any special exceptions?

Anything less just wouldn't be the Cowboy Way.

I for one don't need to make exceptions for the triode.
I take it for granted it has an internal NFB loop whether we like it or
not.
Internal triode NFB gives it very different properties to a pentode,
and the difference can be dramatically seen
when running a 6AU6 in either triode or pentode mode.

Even when running a triode such as 6AU6 as a triode in a CF,
the internal NFB operates to maintain that
Vk / Vgk = open loop gain = µ x RL ( RL + Ra ) where
Ra and µ as as previously discussed in posts about measuring them to
find gm,
and where RL = Rk and in fact = total dc and ac loads.

Were we to use the 6AU6 as a pentode CF with the screen also
changing its signal just as the cathode, then the open loop gain
will be much higher than the triode case.

But regardless of triode or pentode, the output resistance
of either pentode or trioded 6AU6 = 1/gm approximately.

A model of any type of circuit including a triode can be made
to include the triode NFB but most people don't ever bother;
they just consider the simple generator model with series Ra
and this model IS the result of the tube including its internal NFB.
A circuit with a pentode can be drawn similarly, only
that Ra is a lot higher, and µ a lot higher.

The accedemic importance of the internal NFB of triodes, or lack of it
in pentodes
fades in the presence of the importance given to
what we might do with such devices, ie, in ways to build amplifiers.

I find myself so busy each day with a soldering iron that I scarcely
give
internal NFB in triodes the slightest thought.
I am outcomes oriented, and do need to please people, so
I don't have some undefined vague quest to investigate the mysterious
behaviour
of tubes any more than I have already.

Patrick Turner.





Thanks, as always,

Chris Hornbeck
"Money doesn't buy happiness. But happiness isn't everything."
- Jean Seberg

[Chris Hornbeck]

On Wed, 04 Jul 2007 09:15:37 GMT, Patrick Turner
wrote:

Cathode feedback, or any other external characteristic, has
*nothing* to do with the operation of the triode. Nothing.

Its true that a tube or valve has no consciousness of the external
circuit around itself.

We do, and a distinction does need to be made with regard to
internal NFB and external.

Excellent. If you could somehow explain this to "flipper",
maybe we could move on. I've tried twice to start a post,
but have found myself overwhelmed. Seems an obvious point,
but maybe not...


It is by external observation that those of us with sufficient
consciousness
will conclude that triodes have internal NFB and are not just a
resistor.

This, of course, remains to be proven. We're arguing about
applicability of models, so sweeping generalizations are
somewhere short of convincing. Yet.

Thanks, as always,

Chris Hornbeck
"The air is always curved if you choose to see it."
-Steve McMullen reviewing the first Curved Air album

[Patrick Turner]

Chris Hornbeck wrote:

On Wed, 04 Jul 2007 09:15:37 GMT, Patrick Turner
wrote:

Cathode feedback, or any other external characteristic, has
*nothing* to do with the operation of the triode. Nothing.

Its true that a tube or valve has no consciousness of the external
circuit around itself.

We do, and a distinction does need to be made with regard to
internal NFB and external.

Excellent. If you could somehow explain this to "flipper",
maybe we could move on. I've tried twice to start a post,
but have found myself overwhelmed. Seems an obvious point,
but maybe not...

Try to be friendly. Don't worry, be happy.


Having established this state in yourself, consider that
ones intellectual adversaries need prolonged persuasions,
not prolonged revenges.
Dis-establish your anger which if acted upon as a first response will
yield to you a vale of tears for all concerned.
Treat even your adversaries with respect,
and not as they do unto you, and then hopefully
they'll explore a journey of learning with you, because there is
something in it for them.

But if they tell you to eat ****, and get stuffed,
and keep repeating their recalcitrance, let em have it both
barrels to illustrate the idiocy of they determined negative efforts.

They'll take even longer to heal, but it would not be because you didn't
give them a huge long chance
to mend their ways.
Show your eventual forgiveness and show that you'll talk again later if
they
manage something interesting to say to the group.

Most ppl come around, or at least leave you alone......

Meanwhile valves, or tubes have only spirits, and no brains.
The God Of Triodes meant it to be this way.
On at least 50,000 other planets in the very tiny part of the
universe which we are aware of, beings who have evolved to
farnarkle around with electronics have been allowed devices which
presently could be assumed to be very
different to what the GOT allowed us to findoutabout.

The 3 prong devices we have thus far managed to make act in ways which
can give very different circuit results yet they remain themselves,
and should we shrink ourselves down in size to be able to walk inside a
triode or a tiny transistor, and set up camp, we would see electrons
coming and going
along pathways of vacuum or substrate materials, and marvel at it all,
and notice the changes in voltages promoting the currents, and not be
aware
of where the heck the little device is located, perhaps in someone's
preamp,
or ad converter, or microwave oven.

Of course if one was equipped with a suitable small amp and speaker
while
standing on the riverbanks where electrons flow past, one might dip a
wire in,
and hear Motzart coming from the stream, and yes, we are probably in
someone's
preamp. If it sounded dirty, maybe its a driver transistor on a solid
state amp,
and the error signal is present to cancel the the output stage
distortions.
If you just hear a stream of clicks, its a lousy campsite; you are
inside digital gear :-(

The nicest site to camp would be in a triode with no external NFB loop
:-)
There we'd sample the signal and find it clean enough, almost always
we'd find it to be
music, and it'd be much more spacious,
and the weather would be nice and warm.

Having shrunk ourselves down in size to allow such observations
poses the greater problem of what to do with the excessive mass and bulk
we have dispensed with, E = mc squared comes to mind, but I don't like
violence.

Patrick Turner.






It is by external observation that those of us with sufficient
consciousness
will conclude that triodes have internal NFB and are not just a
resistor.

This, of course, remains to be proven. We're arguing about
applicability of models, so sweeping generalizations are
somewhere short of convincing. Yet.

Thanks, as always,

Chris Hornbeck
"The air is always curved if you choose to see it."
-Steve McMullen reviewing the first Curved Air album

[Chris Hornbeck]

On Thu, 05 Jul 2007 08:38:00 GMT, Patrick Turner
wrote:

Try to be friendly. Don't worry, be happy.

Lots of other interesting and surprisingly well edited
stuff snipped for bandwidth but largely recommended for content

Dunno why anyone would consider my comments as in any
way unfriendly or antagonistic. We're discussing models,
not religion. It's perfectly OK to change one's mind, just
as it's perfectly OK to be wrong. Personally, I'm wrong
*lot's*; any real world gig includes it as a daily regimen.
Only the President of the United States is never wrong,
and look how well that works. My comments are just my comments.



To sum up my current opinion about the best (that I've heard
to date) thoughts arguing for the inherent-feedback-in-
vacuum-valve-triodes(-but-not-in-any-other-active-device...):

1) Electrostatic field summing effects upon a virtual grid:

Why is there no measurable external effect? This is a ghost-
in-the-machine argument of the type that went out of use
in real science shortly after Newton. And even before that,
Occar's Razor should have been a warning sign.

2) Any argument relying on some analogy from some external
circuit configuration:

How can *any* external circuit configuration have *any* bearing
on the internal operation of a device when it's impossible
to measure *anything* about the external circuit configuration
when observing only the terminals of the device?

IOW, only the device itself can be observed, or measured. If
this is difficult to imagine as true, one need only make some
actual measurements.


Thanks, as always,

Chris Hornbeck
"The air is always curved if you choose to see it."
-Steve McMullen reviewing the first Curved Air album

[Chris Hornbeck]

On Thu, 05 Jul 2007 07:59:21 -0500, flipper wrote:

"flipper",

"Flipper" doesn't have any problem with it.

If "flipper" is your real name, I'm sorry to have enclosed
it in parenthesis. I post with my real name, but others
feel otherwise. That's their call. Quite unusual name;
what's the origin?


maybe we could move on. I've tried twice to start a post,
but have found myself overwhelmed. Seems an obvious point,
but maybe not...

Maybe if you'd be a bit more precise and clear as to what you are
agreeing with and what you're disputing because, as it is, you appear
to jump between 'feedback' and 'internal' and cathode and plate as if
they're all synonyms.

The (hijacked) thread topic is the old one of inherent-feedback-
in-triode-vacuum-valves. You've posted elaborate models pro-INFBiTVV
that rely on analogy to external circuits. Arguments by analogy
aren't automagically suspect; rather, the reliance on an *external*
analogy is fatally flawed.

A powerful way to think about the flaw is to imagine observing
a triode (or any other active device) with fast voltmeters on
all active terminals. Can we, using these fast voltmeters (or
oscilloscopes, or, in fact, *ANY* measuring device, observe the
triode acting any differently in *ANY* external circuit?

Because we cannot, any external circuit is "trivial". It cannot
contribute anything interesting to the internal model of the
triode, because it doesn't contribute anything *at all* to the
operation of the triode.

So, the actual internal operation of the triode remains to be
completely described. Not that surprising; the universe is
bazillions of years old and models get better but are still
somewhat iffy. Dark matter, anyone?


Thanks, as always,

Chris Hornbeck
"The air is always curved if you choose to see it."
-Steve McMullen reviewing the first Curved Air album

[Patrick Turner]

Chris Hornbeck wrote:

On Thu, 05 Jul 2007 08:38:00 GMT, Patrick Turner
wrote:

Try to be friendly. Don't worry, be happy.

Lots of other interesting and surprisingly well edited
stuff snipped for bandwidth but largely recommended for content

Dunno why anyone would consider my comments as in any
way unfriendly or antagonistic. We're discussing models,
not religion. It's perfectly OK to change one's mind, just
as it's perfectly OK to be wrong. Personally, I'm wrong
*lot's*; any real world gig includes it as a daily regimen.
Only the President of the United States is never wrong,
and look how well that works. My comments are just my comments.

To sum up my current opinion about the best (that I've heard
to date) thoughts arguing for the inherent-feedback-in-
vacuum-valve-triodes(-but-not-in-any-other-active-device...):

1) Electrostatic field summing effects upon a virtual grid:

Why is there no measurable external effect? This is a ghost-
in-the-machine argument of the type that went out of use
in real science shortly after Newton. And even before that,
Occar's Razor should have been a warning sign.

2) Any argument relying on some analogy from some external
circuit configuration:

How can *any* external circuit configuration have *any* bearing
on the internal operation of a device when it's impossible
to measure *anything* about the external circuit configuration
when observing only the terminals of the device?

IOW, only the device itself can be observed, or measured. If
this is difficult to imagine as true, one need only make some
actual measurements.

Thanks, as always,

Chris Hornbeck
"The air is always curved if you choose to see it."
-Steve McMullen reviewing the first Curved Air album

There are things in physics where if we measure, we
spoil what is being measured, and nothing is certain.
Quantum physics is like this, and the more they try to unravel
the atom to find out what particles are responsible for say, mass
and what exact particles are responsible for say time,
then the more difficult finding the truth seems to be.

But a if you were small enough, and had the right gear,
you could measure summed fields in a triode and deduce there is a NFB
network present.

Patrick Turner.