[[email protected]]

Well, I've been and gone and done it. Built a rig along the lines suggested
by Patrick and measured a couple of 6AU6 pentodes wired as triodes. The
basic set up has a 39K plate resistor fed from 250V. Looking at the 6AU6
data sheet in triode mode, to get about 3.8mA Ia needs about 1.5V grid bias
which means an Rk of 390 ohms which I bypassed with 4.7uF (because I had
one handy) - that gives a -3dB point just under 90Hz but as I am testing at
1KHz Rk is effectively bypassed at this frequency.

The 6AU6 is wired with plate, screen and suppressor grids wired together, a
470K grid resistor and a 100mV signal fed to the grid from an external
oscillator. A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement.

Seems to work well. Tried it on two tubes. Both give a grid bias around 1.5V
and a plate voltage near 100V as expected. And the results:

Tube 1:

G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

Tube 2:

G1 = 35.6
G2 = 29.6

which means

mu = 44.65
ra = 9.66K
gm = 4.62 mA/V

I have found an article about using the 6AU6 as a triode where the screen is
used as the plate and this has curves that show mu just under 40 at this Ia
so these mu figures a little higher seem about right considering the
different triode connection.

Next I'll try a few more tubes to verify the results then mod the circuit to
the screen as plate version and repeat the tests.

Cheers

Ian (TB)

[Patrick Turner]

wrote:

Well, I've been and gone and done it. Built a rig along the lines suggested
by Patrick and measured a couple of 6AU6 pentodes wired as triodes. The
basic set up has a 39K plate resistor fed from 250V. Looking at the 6AU6
data sheet in triode mode, to get about 3.8mA Ia needs about 1.5V grid bias
which means an Rk of 390 ohms which I bypassed with 4.7uF (because I had
one handy) - that gives a -3dB point just under 90Hz but as I am testing at
1KHz Rk is effectively bypassed at this frequency.

The 6AU6 is wired with plate, screen and suppressor grids wired together, a
470K grid resistor and a 100mV signal fed to the grid from an external
oscillator. A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement.

Seems to work well. Tried it on two tubes. Both give a grid bias around 1.5V
and a plate voltage near 100V as expected. And the results:

Tube 1:

G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

Tube 2:

G1 = 35.6
G2 = 29.6

which means

mu = 44.65
ra = 9.66K
gm = 4.62 mA/V

I have found an article about using the 6AU6 as a triode where the screen is
used as the plate and this has curves that show mu just under 40 at this Ia
so these mu figures a little higher seem about right considering the
different triode connection.

Next I'll try a few more tubes to verify the results then mod the circuit to
the screen as plate version and repeat the tests.

Cheers

Ian (TB)

Does (TB) mean Tube Boffin?

If not, maybe you are heading that way, from what you have said above
about testing triodes
for their parameters.

The method I suggested works eh? Maths were not too difficult now were
they?

I assume your "unloaded" meant you had just the 39k dc RL, and "loaded"
meant with
anm added 39k, for a total load of 19.5k.

When using the screen as the anode, and with anode grounded as a shield,
the screen dissipation needs to be low and well within its rated limits
lest its high temp
interferes with electron collection.

What results do you get with pentode connections?

Patrick Turner.

[Ian Iveson]

...A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement...

Seems a clause or two short of meaning. You mean .47uF from anode to
voltmeter? Is that a second 39k in series with the first?

Seems to work well. Tried it on two tubes. Both give a grid bias
around 1.5V
and a plate voltage near 100V as expected. And the results:

It's bound to seem to work well. How closely does it approximate to
exact values? Why don't you measure gm more directly, with an anode
sense resistor of perhaps 100ohms or less and a supply of around 100V?
Then you can compare easily with pentode connection by connecting the
screen to the supply. Are you short of accurate instrumentation?

I've lost the plot...a continuity problem...was hoping you would
determine the current and gm split between screen and anode, to answer
your original question.

G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

By what arithmetic, please?

cheers, Ian

[[email protected]]

Patrick Turner wrote:



wrote:

Well, I've been and gone and done it. Built a rig along the lines
suggested by Patrick and measured a couple of 6AU6 pentodes wired as
triodes. The basic set up has a 39K plate resistor fed from 250V. Looking
at the 6AU6 data sheet in triode mode, to get about 3.8mA Ia needs about
1.5V grid bias which means an Rk of 390 ohms which I bypassed with 4.7uF
(because I had one handy) - that gives a -3dB point just under 90Hz but
as I am testing at 1KHz Rk is effectively bypassed at this frequency.

The 6AU6 is wired with plate, screen and suppressor grids wired together,
a 470K grid resistor and a 100mV signal fed to the grid from an external
oscillator. A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement.

Seems to work well. Tried it on two tubes. Both give a grid bias around
1.5V and a plate voltage near 100V as expected. And the results:

Tube 1:

G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

Tube 2:

G1 = 35.6
G2 = 29.6

which means

mu = 44.65
ra = 9.66K
gm = 4.62 mA/V

I have found an article about using the 6AU6 as a triode where the screen
is used as the plate and this has curves that show mu just under 40 at
this Ia so these mu figures a little higher seem about right considering
the different triode connection.

Next I'll try a few more tubes to verify the results then mod the circuit
to the screen as plate version and repeat the tests.

Cheers

Ian (TB)

Does (TB) mean Tube Boffin?


LOL, but not a bad idea - sounds like a good name for a web page ;-)

If not, maybe you are heading that way, from what you have said above
about testing triodes
for their parameters.

The method I suggested works eh? Maths were not too difficult now were
they?

The maths was a bit awkward the first time but once I had derived the
equations I simply created a spreadsheet. You just plug in G1 and G2 and
out cone mu, ra and gm. If you are interested I start with ra:

ra = 0.5*RL*(G2-G1)/(G2-0.5*G1)

then you can work out mu form:

mu = G1(ra+RL)/RL or G2(ra+0.5*RL)/0.5*RL

and then gm from:

gm = mu/ra


I assume your "unloaded" meant you had just the 39k dc RL, and "loaded"
meant with
anm added 39k, for a total load of 19.5k.


Correct.

When using the screen as the anode, and with anode grounded as a shield,
the screen dissipation needs to be low and well within its rated limits
lest its high temp
interferes with electron collection.


Yup. 0.65W is the limit.

What results do you get with pentode connections?



That is experiment three, right after I try the triode with the screen as
plate. I'll let you know. First I am going to try a lot more tubes with the
current set up to get an idea of the spread of values.

Cheers

Ian (Tube Boffin ;-)

[[email protected]]

Ian Iveson wrote:

...A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement...

Seems a clause or two short of meaning. You mean .47uF from anode to
voltmeter? Is that a second 39k in series with the first?


39K from anode to Ebb. 0.47uF from anode to measuring point. Switchable 39K
from measuring point to ground.

Seems to work well. Tried it on two tubes. Both give a grid bias
around 1.5V
and a plate voltage near 100V as expected. And the results:

It's bound to seem to work well. How closely does it approximate to
exact values? Why don't you measure gm more directly, with an anode
sense resistor of perhaps 100ohms or less and a supply of around 100V?

Because:

a) you did not suggest this when I asked how to measure gm
b) I already measured gm directly with my AVO valve tester and got gm values
that seemed rather high.
c) I also want to know mu and ra and this method allows them all tovbe
measured.

Then you can compare easily with pentode connection by connecting the
screen to the supply. Are you short of accurate instrumentation?


I've lost the plot...a continuity problem...was hoping you would
determine the current and gm split between screen and anode, to answer
your original question.

My original question was prompted by the rather high gm readings I got on my
AVO tester with the 6AU6 wired as a triode. gm split between screen and
plate was insufficient to explain the high readings so Patrick suggested I
build a rig to measure gm et al which I have done.


G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

By what arithmetic, please?


G1 is with the 39K anode resistor load only. G2 is with the additional 39K
load to ground making the total RL 19K5.

So:

mu = G1(ra+39K)/39K = G2(ra+19.5K)/19.5K

It is easy to solve these for ra or mu from which gm can be calculated.

Ian (TB)

[The Phantom]

On Sat, 30 Jun 2007 20:06:22 +0100, wrote:

Well, I've been and gone and done it. Built a rig along the lines suggested
by Patrick and measured a couple of 6AU6 pentodes wired as triodes. The
basic set up has a 39K plate resistor fed from 250V. Looking at the 6AU6
data sheet in triode mode, to get about 3.8mA Ia needs about 1.5V grid bias
which means an Rk of 390 ohms which I bypassed with 4.7uF (because I had
one handy) - that gives a -3dB point just under 90Hz but as I am testing at
1KHz Rk is effectively bypassed at this frequency.

The 6AU6 is wired with plate, screen and suppressor grids wired together, a
470K grid resistor and a 100mV signal fed to the grid from an external
oscillator. A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement.

Seems to work well. Tried it on two tubes. Both give a grid bias around 1.5V
and a plate voltage near 100V as expected. And the results:

Tube 1:

G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

Tube 2:

G1 = 35.6
G2 = 29.6

which means

mu = 44.65
ra = 9.66K
gm = 4.62 mA/V

I have found an article about using the 6AU6 as a triode where the screen is
used as the plate and this has curves that show mu just under 40 at this Ia
so these mu figures a little higher seem about right considering the
different triode connection.

Next I'll try a few more tubes to verify the results then mod the circuit to
the screen as plate version and repeat the tests.

Cheers

Ian (TB)

It would be interesting if you would try measuring gm directly. With the single
39k load resistor in place, connect a 5 uF capacitor to the plate, and the other
end of the capacitor to an AC milliammeter, thence to ground; this provides an
AC short from plate to ground as the canonical gm definition requires. You
could use the AC current range of a DMM, or if you have some AC milliammeter
panel meters (make sure they're accurate at 1 kHz), you could use those. The
milliammeter reading divided by the 1 kHz AC voltage on G1 will be the gm of the
tube.

Then you can measure the separate gm's of plate only, or plate plus screen using
the suggestion I made in another post:

" Perhaps one could supply the screen through a high value resistor, and have a
large bypass capacitor to ground (with a cathode resistor or not, as the case
may be). Different DC voltages could be applied to plate and screen in pentode
mode, and a measurement of gm made with an audio frequency AC signal on grid 1
and an AC short on the plate; I think this is the usual method. Then the ground
end of the screen bypass capacitor could be moved to the plate and (plate plus
screen) gm could then be measured in triode mode, but still with different DC
voltages on plate and screen."

You could also measure the screen gm by putting the AC milliammeter in series
with the screen bypass capacitor (at the ground end for safety).

If you have two AC milliammeters, you could have a capacitor from each of
plate and screen connected to a grounded milliammeter, and measure the gm's
simultaneously. You could watch how the two gm's vary as you vary the screen
voltage relative to the plate voltage.

[Patrick Turner]

Ian Iveson wrote:

...A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement...

Seems a clause or two short of meaning. You mean .47uF from anode to
voltmeter? Is that a second 39k in series with the first?

Seems to work well. Tried it on two tubes. Both give a grid bias
around 1.5V
and a plate voltage near 100V as expected. And the results:

It's bound to seem to work well. How closely does it approximate to
exact values? Why don't you measure gm more directly, with an anode
sense resistor of perhaps 100ohms or less and a supply of around 100V?
Then you can compare easily with pentode connection by connecting the
screen to the supply. Are you short of accurate instrumentation?

I've lost the plot...a continuity problem...was hoping you would
determine the current and gm split between screen and anode, to answer
your original question.

Doing things my way he finds out Ra, µ and gm,
because gain, A = µ x RL / ( Ra + RL ).

µ = gm x Ra, so gm = µ / Ra

Your way gives only gm.

To calculate stage gain with an RL you NEED to know
at least TWO of the items out of 3, µ, Ra or gm.


Patrick Turner.



G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

By what arithmetic, please?

cheers, Ian

[Patrick Turner]

wrote:

Patrick Turner wrote:



wrote:

Well, I've been and gone and done it. Built a rig along the lines
suggested by Patrick and measured a couple of 6AU6 pentodes wired as
triodes. The basic set up has a 39K plate resistor fed from 250V. Looking
at the 6AU6 data sheet in triode mode, to get about 3.8mA Ia needs about
1.5V grid bias which means an Rk of 390 ohms which I bypassed with 4.7uF
(because I had one handy) - that gives a -3dB point just under 90Hz but
as I am testing at 1KHz Rk is effectively bypassed at this frequency.

The 6AU6 is wired with plate, screen and suppressor grids wired together,
a 470K grid resistor and a 100mV signal fed to the grid from an external
oscillator. A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement.

Seems to work well. Tried it on two tubes. Both give a grid bias around
1.5V and a plate voltage near 100V as expected. And the results:

Tube 1:

G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

Tube 2:

G1 = 35.6
G2 = 29.6

which means

mu = 44.65
ra = 9.66K
gm = 4.62 mA/V

I have found an article about using the 6AU6 as a triode where the screen
is used as the plate and this has curves that show mu just under 40 at
this Ia so these mu figures a little higher seem about right considering
the different triode connection.

Next I'll try a few more tubes to verify the results then mod the circuit
to the screen as plate version and repeat the tests.

Cheers

Ian (TB)

Does (TB) mean Tube Boffin?


LOL, but not a bad idea - sounds like a good name for a web page ;-)

If not, maybe you are heading that way, from what you have said above
about testing triodes
for their parameters.

The method I suggested works eh? Maths were not too difficult now were
they?

The maths was a bit awkward the first time but once I had derived the
equations I simply created a spreadsheet. You just plug in G1 and G2 and
out cone mu, ra and gm. If you are interested I start with ra:

ra = 0.5*RL*(G2-G1)/(G2-0.5*G1)

I wonder why you have a constant of 0.5 there.
But I assume its because you have gain for 39k = G1,
and gain for 19.5k = G2, and the 0.5 = 19.5 / 39.

So what you have done is combime the two full gain
equations of G1 = µ x RL / ( Ra + RL ).

So for one of the equations I would have derived µ = [ gain1 x ( Ra +
RL1 ) ] / RL1.

So for TWO values of gain for the same tube, same conditions, µ is the
same, so

G1 x ( Ra + 39 ) / 39 = G2 x ( Ra + 19.5 ) / 19.5 .

when the actual figures for Gain1 and Gain2, 39k and 19.5k are inserted
ther only unknown
is Ra which can be calculated.

But from the above,

G1 x ( Ra + 39 ) / 39 = 2 x G2 x ( Ra + 19.5 ) / 39.

Multiply both sides by 39, Therefore

( G1 x Ra ) + ( 39 x G1 ) = ( 2 x Ra x G2 ) + ( 39 x G2 ).

2Ra.G2 - Ra.G1 = 39.G1 - 39G2

Ra ( 2.G2 -G1 ) = 39 ( G1 - G2 ).

Ra = 39 ( G1 - G2 ) / ( 2.G2 - G1 )

Inserting your gain figures if 33.3 and 27.3,

Ra = 39 x ( 33.3 - 27.3 ) / ( 54.6 - 33.3 ),

= 39 x 6 / 21.3 = 10.98 kohms

You calculated 10.7k, and let's not argue about the 0.1k discrepancy....




then you can work out mu form:

mu = G1(ra+RL)/RL or G2(ra+0.5*RL)/0.5*RL

and then gm from:

gm = mu/ra


I assume your "unloaded" meant you had just the 39k dc RL, and "loaded"
meant with
anm added 39k, for a total load of 19.5k.

The other way to work out Ra is to measure the output voltage with 39k,
and calculate the load current.
Then measure the output voltage with 19.5k, and calculate load current.

Without any of the above rather complex formula,
Ra = ( Vout1 - Vout2 ) / ( Iout2 - Iout1 )

And if A = µ x RL / ( Ra + RL ),
then inserting the simply measured and calculated Ra into the
gain equation for any RL will give you µ, and hence gm.

The universally applicable gain equation can me also written as

A = gm x Ra x RL / ( Ra + RL )..

The two unknowns are gm and Ra rather than µ and Ra.


For pentodes this may be easier because the pentode µ is
often a figure of several thousand, and voltage measurements need to be
very accurate.

Patrick Turner.



Correct.

When using the screen as the anode, and with anode grounded as a shield,
the screen dissipation needs to be low and well within its rated limits
lest its high temp
interferes with electron collection.


Yup. 0.65W is the limit.

What results do you get with pentode connections?


That is experiment three, right after I try the triode with the screen as
plate. I'll let you know. First I am going to try a lot more tubes with the
current set up to get an idea of the spread of values.

Cheers

Ian (Tube Boffin ;-)

[[email protected]]

The Phantom wrote:

On Sat, 30 Jun 2007 20:06:22 +0100, wrote:

Well, I've been and gone and done it. Built a rig along the lines
suggested by Patrick and measured a couple of 6AU6 pentodes wired as
triodes. The basic set up has a 39K plate resistor fed from 250V. Looking
at the 6AU6 data sheet in triode mode, to get about 3.8mA Ia needs about
1.5V grid bias which means an Rk of 390 ohms which I bypassed with 4.7uF
(because I had one handy) - that gives a -3dB point just under 90Hz but as
I am testing at 1KHz Rk is effectively bypassed at this frequency.

The 6AU6 is wired with plate, screen and suppressor grids wired together,
a 470K grid resistor and a 100mV signal fed to the grid from an external
oscillator. A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement.

Seems to work well. Tried it on two tubes. Both give a grid bias around
1.5V and a plate voltage near 100V as expected. And the results:

Tube 1:

G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

Tube 2:

G1 = 35.6
G2 = 29.6

which means

mu = 44.65
ra = 9.66K
gm = 4.62 mA/V

I have found an article about using the 6AU6 as a triode where the screen
is used as the plate and this has curves that show mu just under 40 at
this Ia so these mu figures a little higher seem about right considering
the different triode connection.

Next I'll try a few more tubes to verify the results then mod the circuit
to the screen as plate version and repeat the tests.

Cheers

Ian (TB)

It would be interesting if you would try measuring gm directly. With the
single 39k load resistor in place, connect a 5 uF capacitor to the plate,
and the other end of the capacitor to an AC milliammeter, thence to
ground; this provides an
AC short from plate to ground as the canonical gm definition requires.
You could use the AC current range of a DMM, or if you have some AC
milliammeter
panel meters (make sure they're accurate at 1 kHz), you could use those.
The milliammeter reading divided by the 1 kHz AC voltage on G1 will be the
gm of the tube.


That's quite an easy mod. I'll try that.

Then you can measure the separate gm's of plate only, or plate plus screen
using the suggestion I made in another post:

" Perhaps one could supply the screen through a high value resistor, and
have a large bypass capacitor to ground (with a cathode resistor or not,
as the case
may be). Different DC voltages could be applied to plate and screen in
pentode mode, and a measurement of gm made with an audio frequency AC
signal on grid 1
and an AC short on the plate; I think this is the usual method. Then the
ground end of the screen bypass capacitor could be moved to the plate and
(plate plus screen) gm could then be measured in triode mode, but still
with different DC voltages on plate and screen."

You could also measure the screen gm by putting the AC milliammeter in
series
with the screen bypass capacitor (at the ground end for safety).

If you have two AC milliammeters, you could have a capacitor from each
of
plate and screen connected to a grounded milliammeter, and measure the
gm's
simultaneously. You could watch how the two gm's vary as you vary the
screen voltage relative to the plate voltage.


Interesting. I have saved this post for later reference. At present I am
interested principally in its triode connected parameters.

Cheers

Ian (TB)

[[email protected]]

Patrick Turner wrote:



wrote:

Patrick Turner wrote:



wrote:

Well, I've been and gone and done it. Built a rig along the lines
suggested by Patrick and measured a couple of 6AU6 pentodes wired as
triodes. The basic set up has a 39K plate resistor fed from 250V.
Looking at the 6AU6 data sheet in triode mode, to get about 3.8mA Ia
needs about 1.5V grid bias which means an Rk of 390 ohms which I
bypassed with 4.7uF (because I had one handy) - that gives a -3dB
point just under 90Hz but as I am testing at 1KHz Rk is effectively
bypassed at this frequency.

The 6AU6 is wired with plate, screen and suppressor grids wired
together, a 470K grid resistor and a 100mV signal fed to the grid from
an external oscillator. A 0.47uF cap to the anode is used to measure
stage gain and a switch adds a 39K load to this for the second
measurement.

Seems to work well. Tried it on two tubes. Both give a grid bias
around 1.5V and a plate voltage near 100V as expected. And the
results:

Tube 1:

G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

Tube 2:

G1 = 35.6
G2 = 29.6

which means

mu = 44.65
ra = 9.66K
gm = 4.62 mA/V

I have found an article about using the 6AU6 as a triode where the
screen is used as the plate and this has curves that show mu just
under 40 at this Ia so these mu figures a little higher seem about
right considering the different triode connection.

Next I'll try a few more tubes to verify the results then mod the
circuit to the screen as plate version and repeat the tests.

Cheers

Ian (TB)

Does (TB) mean Tube Boffin?


LOL, but not a bad idea - sounds like a good name for a web page ;-)

If not, maybe you are heading that way, from what you have said above
about testing triodes
for their parameters.

The method I suggested works eh? Maths were not too difficult now were
they?

The maths was a bit awkward the first time but once I had derived the
equations I simply created a spreadsheet. You just plug in G1 and G2 and
out cone mu, ra and gm. If you are interested I start with ra:

ra = 0.5*RL*(G2-G1)/(G2-0.5*G1)

I wonder why you have a constant of 0.5 there.
But I assume its because you have gain for 39k = G1,
and gain for 19.5k = G2, and the 0.5 = 19.5 / 39.

So what you have done is combime the two full gain
equations of G1 = µ x RL / ( Ra + RL ).

So for one of the equations I would have derived µ = [ gain1 x ( Ra +
RL1 ) ] / RL1.

So for TWO values of gain for the same tube, same conditions, µ is the
same, so

G1 x ( Ra + 39 ) / 39 = G2 x ( Ra + 19.5 ) / 19.5 .

when the actual figures for Gain1 and Gain2, 39k and 19.5k are inserted
ther only unknown
is Ra which can be calculated.

But from the above,

G1 x ( Ra + 39 ) / 39 = 2 x G2 x ( Ra + 19.5 ) / 39.

Multiply both sides by 39, Therefore

( G1 x Ra ) + ( 39 x G1 ) = ( 2 x Ra x G2 ) + ( 39 x G2 ).

2Ra.G2 - Ra.G1 = 39.G1 - 39G2

Ra ( 2.G2 -G1 ) = 39 ( G1 - G2 ).

Ra = 39 ( G1 - G2 ) / ( 2.G2 - G1 )

Inserting your gain figures if 33.3 and 27.3,

Ra = 39 x ( 33.3 - 27.3 ) / ( 54.6 - 33.3 ),

= 39 x 6 / 21.3 = 10.98 kohms

You calculated 10.7k, and let's not argue about the 0.1k discrepancy....



I just took another look at the spreadsheet and I realised I had put RL =
38K not 39K (the spreadsheet works for any value of RL). Correcting that
value gives identical results to your calcs. Thanks for finding the error.



then you can work out mu form:

mu = G1(ra+RL)/RL or G2(ra+0.5*RL)/0.5*RL

and then gm from:

gm = mu/ra


I assume your "unloaded" meant you had just the 39k dc RL, and "loaded"
meant with
anm added 39k, for a total load of 19.5k.

The other way to work out Ra is to measure the output voltage with 39k,
and calculate the load current.
Then measure the output voltage with 19.5k, and calculate load current.

Without any of the above rather complex formula,
Ra = ( Vout1 - Vout2 ) / ( Iout2 - Iout1 )


And
Iout1 = Vout1/39
Iout2 = 2*Vout2/39

So Ra = (Vout1 - Vout2) / (2*Vout2 - Vout1)/39

= Ra = 39(Vout1 - Vout2) /(2*Vout2 - Vout1)

Substituting Vout1 = 3.33 and Vout2 = 2.73 (because these are the actual
measured voltages and the G1 and G2 values assume 100mV input) we get:

Ra = 39*0.6/2.13 = 10.98K

The good thing about this is that it is *independent* of the input voltage.


Ian (TB)

[Ian Iveson]

Ian wrote [below]:

OK, but you set out to measure gm, I thought. Your present method
approximates to several parameters over a range rather than isolating
one at a particular spot. There is unlikely to be a lot of difference,
but it would be interesting to know exactly how much. I always feel
uneasy about the kind of measurement that requires assumptions about
what you are measuring, which you then take to be vindicated by the
results. Descartes springs to mind.

As others have pointed out, a direct measure of gm requires that Vak
and Vsk remain constant. It is not difficult to think of any number of
ways of doing it, including the AVO method, except without the
chicanery and malcalibration.

I measure Ra directly by connecting a generator probe in series with a
current sense resistor to the anode in circuit and noting the ensuing
AC voltage and current...same as any other measurement of resistance.
I do the same with amplifiers to measure the output impedance. Dunno
why anyone does any different.

mu can be calculated coz it's the hardest to measure for most folk I
guess.

I really would appreciate it if you could at least find how much of
that 4mA/V is due to the anode, and how much to the screen. Either my
method or, er, The Phantom's...but why complicate things?

I reckoned a split of 1.2 : 2.8, if you remember.

If you characterise a valve precisely, you could use it to calibrate
your AVO. Calibration should be done using a valve somewhere around
the middle of whatever range you generally expect to measure, so the
6AU6 may not be the best choice. Can't remember what AVO recommends.

cheers, Ian1


...A 0.47uF cap to the anode is used to measure stage gain and a
switch adds a 39K load to this for the second measurement...

Seems a clause or two short of meaning. You mean .47uF from anode to
voltmeter? Is that a second 39k in series with the first?

Seems to work well. Tried it on two tubes. Both give a grid bias
around 1.5V
and a plate voltage near 100V as expected. And the results:

It's bound to seem to work well. How closely does it approximate to
exact values? Why don't you measure gm more directly, with an anode
sense resistor of perhaps 100ohms or less and a supply of around
100V?
Then you can compare easily with pentode connection by connecting
the
screen to the supply. Are you short of accurate instrumentation?

I've lost the plot...a continuity problem...was hoping you would
determine the current and gm split between screen and anode, to
answer
your original question.

Doing things my way he finds out Ra, µ and gm,
because gain, A = µ x RL / ( Ra + RL ).

µ = gm x Ra, so gm = µ / Ra

Your way gives only gm.

To calculate stage gain with an RL you NEED to know
at least TWO of the items out of 3, µ, Ra or gm.


Patrick Turner.



G1 (unloaded) 33.3
G2 (loaded) 27.3

which means

mu = 42.68
ra = 10.7K
gm = 3.99 mA/V

By what arithmetic, please?

cheers, Ian

[Ian Iveson]

Ian wrote:

Ra = 39(Vout1 - Vout2) /(2*Vout2 - Vout1)

Substituting Vout1 = 3.33 and Vout2 = 2.73 (because these are the
actual
measured voltages and the G1 and G2 values assume 100mV input) we
get:

Ra = 39*0.6/2.13 = 10.98K

The good thing about this is that it is *independent* of the input
voltage.

Or the bad thing...

Ian1

[[email protected]]

Ian Iveson wrote:

Ian wrote [below]:

OK, but you set out to measure gm, I thought. Your present method
approximates to several parameters over a range rather than isolating
one at a particular spot. There is unlikely to be a lot of difference,
but it would be interesting to know exactly how much. I always feel
uneasy about the kind of measurement that requires assumptions about
what you are measuring, which you then take to be vindicated by the
results. Descartes springs to mind.


I agree. However, I do not need to make precise measurements to determine if
the AVO measurement was wrong. Despite the inaccuracies of the method it is
clear that gm in this mode is nowhere near the over 7mA/V measured on the
AVO. It also looks to have established that the particular tube samples
that measured high gm values in pentode mode measure, in triode mode, the
same as all other types (I need to test a few more samples to be sure of
this conclusion).


As others have pointed out, a direct measure of gm requires that Vak
and Vsk remain constant. It is not difficult to think of any number of
ways of doing it, including the AVO method, except without the
chicanery and malcalibration.


Indeed, but it requires at least one stabilised and preferably variable HT
supply which I do not have. I am happy for you to make measurements this
way and report back.

I measure Ra directly by connecting a generator probe

What is a generator probe??

in series with a
current sense resistor to the anode in circuit and noting the ensuing
AC voltage and current...same as any other measurement of resistance.
I do the same with amplifiers to measure the output impedance. Dunno
why anyone does any different.

mu can be calculated coz it's the hardest to measure for most folk I
guess.

I really would appreciate it if you could at least find how much of
that 4mA/V is due to the anode, and how much to the screen. Either my
method or, er, The Phantom's...but why complicate things?


Indeed. Why not do it yourself?


Cheers

Ian (TB)

[Ian Iveson]

I haven't got a 6AU6, obviously. That's why I made such an effort to
help you.

Bye

Ian
wrote in message
...
Ian Iveson wrote:

Ian wrote [below]:

OK, but you set out to measure gm, I thought. Your present method
approximates to several parameters over a range rather than
isolating
one at a particular spot. There is unlikely to be a lot of
difference,
but it would be interesting to know exactly how much. I always feel
uneasy about the kind of measurement that requires assumptions
about
what you are measuring, which you then take to be vindicated by the
results. Descartes springs to mind.


I agree. However, I do not need to make precise measurements to
determine if
the AVO measurement was wrong. Despite the inaccuracies of the
method it is
clear that gm in this mode is nowhere near the over 7mA/V measured
on the
AVO. It also looks to have established that the particular tube
samples
that measured high gm values in pentode mode measure, in triode
mode, the
same as all other types (I need to test a few more samples to be
sure of
this conclusion).


As others have pointed out, a direct measure of gm requires that
Vak
and Vsk remain constant. It is not difficult to think of any number
of
ways of doing it, including the AVO method, except without the
chicanery and malcalibration.


Indeed, but it requires at least one stabilised and preferably
variable HT
supply which I do not have. I am happy for you to make measurements
this
way and report back.

I measure Ra directly by connecting a generator probe

What is a generator probe??

in series with a
current sense resistor to the anode in circuit and noting the
ensuing
AC voltage and current...same as any other measurement of
resistance.
I do the same with amplifiers to measure the output impedance.
Dunno
why anyone does any different.

mu can be calculated coz it's the hardest to measure for most folk
I
guess.

I really would appreciate it if you could at least find how much of
that 4mA/V is due to the anode, and how much to the screen. Either
my
method or, er, The Phantom's...but why complicate things?


Indeed. Why not do it yourself?


Cheers

Ian (TB)