[[email protected]]

What happens to Gm, Rp, and mu when two identical triode sections are
connected together in parallel (for a given bias current & plate
voltage)?

Thanks,

Sean B

[Scott Dorsey]

wrote:
What happens to Gm, Rp, and mu when two identical triode sections are
connected together in parallel (for a given bias current & plate
voltage)?

This sounds like a homework question.

Think about the tube as a variable resistor whose value is controlled by
the grid voltage, and which has zero grid current. If you parallel two
such devices, what happens? Twice the current should flow, but do you
think anything else is going to happen? For the mu to change, the slope
of the curve would have to change. What could cause that?
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Paul Stamler]

wrote in message
...
What happens to Gm, Rp, and mu when two identical triode sections are
connected together in parallel (for a given bias current & plate
voltage)?

Do you mean that the bias current through each section is the same? In other
words, if one section was biased to 2mA, would the two in parallel be
conducting 4mA?

Also, what are you doing with the plate resistor(s)? Paralleling them too?

Peace,
Paul

[[email protected][_2_]]

Sorry about the multiple posts, my first one wasn't popping up after a
few minutes, so i tried again.

To answer Paul's question, think of having the ability to switch in
and out a second triode in parallel, all other conditions being the
same. Consider one or no plate resistor, not two.

So if I read Scott's post correctly, Gm doubles, Rp halves, and Mu
stays the same?

SB

[Paul Stamler]

wrote in message
...
Sorry about the multiple posts, my first one wasn't popping up after a
few minutes, so i tried again.

To answer Paul's question, think of having the ability to switch in
and out a second triode in parallel, all other conditions being the
same. Consider one or no plate resistor, not two.

So if I read Scott's post correctly, Gm doubles, Rp halves, and Mu
stays the same?

Yup. Which gives you a 50% lower output impedance and 3dB lower noise
contribution from the tube circuit. The tradeoff is that you have twice the
input capacitance, which may or may not be important in a given application.

And of course twice the cost and current consumption, both for filament and
plate circuits. And twice the space and heat.

Peace,
Paul

[Scott Dorsey]

wrote:

To answer Paul's question, think of having the ability to switch in
and out a second triode in parallel, all other conditions being the
same. Consider one or no plate resistor, not two.

I'm not sure why this would be useful.

So if I read Scott's post correctly, Gm doubles, Rp halves, and Mu
stays the same?

Right. And because Gm doubles, your output current doubles too
which is a nice thing in a cathode follower circuit.

If the two halves are not properly matched, though, you can get difference
currents between the two sections. There are cases where this is bad for
distortion numbers.

For a good example of when paralling sections is useful, take a look at
the Futterman amplifiers, with a whacking load of 6080 tubes in parallel
to bring the output impedance down as low as possible.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Scott Dorsey]

Paul Stamler wrote:
wrote in message
...
Sorry about the multiple posts, my first one wasn't popping up after a
few minutes, so i tried again.

To answer Paul's question, think of having the ability to switch in
and out a second triode in parallel, all other conditions being the
same. Consider one or no plate resistor, not two.

So if I read Scott's post correctly, Gm doubles, Rp halves, and Mu
stays the same?

Yup. Which gives you a 50% lower output impedance and 3dB lower noise
contribution from the tube circuit. The tradeoff is that you have twice the
input capacitance, which may or may not be important in a given application.

It gives you 3 dB lower thermal noise! It does not always give you lower
flicker noise and in the case of radiation-induced noise and hum from
filament leakage it may be no lower at all.

Incidentally this is a hum-reduction trick that has been used with the
12AX7... if you run 12V across the filaments they are effectively out of
phase with one another. So if the filament leakage is the same on both
halves, when you parallel the two halves the hum is cancelled out.

And of course twice the cost and current consumption, both for filament and
plate circuits. And twice the space and heat.

Sheesh, if we were worried about space and heat we'd use a 2N2222 instead.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Paul Stamler]

"Scott Dorsey" wrote in message
...
Paul Stamler wrote:
wrote in message
...
So if I read Scott's post correctly, Gm doubles, Rp halves, and Mu
stays the same?

Yup. Which gives you a 50% lower output impedance and 3dB lower noise
contribution from the tube circuit. The tradeoff is that you have twice
the
input capacitance, which may or may not be important in a given
application.

It gives you 3 dB lower thermal noise! It does not always give you lower
flicker noise and in the case of radiation-induced noise and hum from
filament leakage it may be no lower at all.

True. And in a well-designed circuit, the active device's noise is
significantly lower than the source noise, and so contributes little to the
overall noise. But if you're trying to squeeze the last little fraction of a
dB from the noise figure in a particular circuit, parallel devices may help.

Peace,
Paul

[Don Pearce]

Paul Stamler wrote:
"Scott Dorsey" wrote in message
...
Paul Stamler wrote:
wrote in message
...
So if I read Scott's post correctly, Gm doubles, Rp halves, and Mu
stays the same?
Yup. Which gives you a 50% lower output impedance and 3dB lower noise
contribution from the tube circuit. The tradeoff is that you have twice
the
input capacitance, which may or may not be important in a given
application.
It gives you 3 dB lower thermal noise! It does not always give you lower
flicker noise and in the case of radiation-induced noise and hum from
filament leakage it may be no lower at all.

True. And in a well-designed circuit, the active device's noise is
significantly lower than the source noise, and so contributes little to the
overall noise. But if you're trying to squeeze the last little fraction of a
dB from the noise figure in a particular circuit, parallel devices may help.

Peace,
Paul



Paralleling devices doesn't actually reduce noise; what it does is alter
the balance between current and voltage noise. What this does is change
the optimum source impedance downwards, so if you are driving from a low
impedance, noise will improve, while if the source impedance is higher,
it may well get worse.

d

[Scott Dorsey]

Don Pearce wrote:

Paralleling devices doesn't actually reduce noise; what it does is alter
the balance between current and voltage noise. What this does is change
the optimum source impedance downwards, so if you are driving from a low
impedance, noise will improve, while if the source impedance is higher,
it may well get worse.

Right. The thing is, in the tube audio world, the source impedance is
_always_ lower than the input impedance of the stage.... and since the
input impedance of the stage is set by the leak resistor rather than the
tube input capacitance, the resistor noise is sometimes a significant
issue.

In the RF world where the input capacitance of the tube means that it
sometimes takes appreciable current to drive it, or in the bipolar
transistor world, that's not a good assumption to make. But for audio
tubes, it's a normal one.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Mark]

On Oct 28, 3:27*pm, Don Pearce wrote:
Paul Stamler wrote:
"Scott Dorsey" wrote in message
...
Paul Stamler wrote:
wrote in message
....
So if I read Scott's post correctly, Gm doubles, Rp halves, and Mu
stays the same?
Yup. Which gives you a 50% lower output impedance and 3dB lower noise
contribution from the tube circuit. The tradeoff is that you have twice
the
input capacitance, which may or may not be important in a given
application.
It gives you 3 dB lower thermal noise! *It does not always give you lower
flicker noise and in the case of radiation-induced noise and hum from
filament leakage it may be no lower at all.

True. And in a well-designed circuit, the active device's noise is
significantly lower than the source noise, and so contributes little to the
overall noise. But if you're trying to squeeze the last little fraction of a
dB from the noise figure in a particular circuit, parallel devices may help.

Peace,
Paul

Paralleling devices doesn't actually reduce noise; what it does is alter
the balance between current and voltage noise. What this does is change
the optimum source impedance downwards, so if you are driving from a low
impedance, noise will improve, while if the source impedance is higher,
it may well get worse.

d- Hide quoted text -

- Show quoted text -

in some cases conbining multiple devices DOES ACTUALLY improve the S/N
ratio..... this is becasue the desired signal from the multiple
devices is correlated and adds constructively as +6 dB, but the noise
in each device is random and uncorrelated so adds only randomly for +3
dB. So you get a 3 dB advantage (in theory) for each doubling of the
number of devices that are combined.

Mark

[Paul Stamler]

"Don Pearce" wrote in message
...

True. And in a well-designed circuit, the active device's noise is
significantly lower than the source noise, and so contributes little to
the
overall noise. But if you're trying to squeeze the last little fraction
of a
dB from the noise figure in a particular circuit, parallel devices may
help.


Paralleling devices doesn't actually reduce noise; what it does is alter
the balance between current and voltage noise. What this does is change
the optimum source impedance downwards, so if you are driving from a low
impedance, noise will improve, while if the source impedance is higher,
it may well get worse.

You're thinking of bipolar devices. FETs and tubes (see the title of this
thread) have virtually no current noise. Parallel them, and the noise goes
down, but the capacitance goes up.

Peace,
Paul

[[email protected][_2_]]

I have one more tube question. What is the actual input impedance of
a triode's grid, say a 12AU7, at low frequency, say 20 Hz?

Sean B

[Anahata]

On Wed, 29 Oct 2008 08:29:45 -0700, wrote:

I have one more tube question. What is the actual input impedance of a
triode's grid, say a 12AU7, at low frequency, say 20 Hz?

Close to infinity.
As long as you don't make it positive w.r.t. cathode, no current flows.

--
Anahata
==//== 01638 720444
http://www.treewind.co.uk ==//== http://www.myspace.com/maryanahata

[Scott Dorsey]

wrote:
I have one more tube question. What is the actual input impedance of
a triode's grid, say a 12AU7, at low frequency, say 20 Hz?

Upwards of a hundred gigohms, depending on how much contamination is
on the inside of the tube envelope.

On a good clean tube, the Miller capacitance sets the actual input
impedance. On some junk from the Shugang plant with a fingerprint on
the mica, the leakage can sometimes be significant.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Paul Stamler]

wrote in message
...
I have one more tube question. What is the actual input impedance of
a triode's grid, say a 12AU7, at low frequency, say 20 Hz?

Whatever the grid leak resistor is -- that's the resistor going from the
grid to ground. Unless you're in the (very nonlinear) region where you're
actually drawing grid current, but under normal circumstances you don't want
to be there.

Peace,
Paul