Re: Pentode-Triode Sound



Mark Aitchison wrote:

I (badly) wrote:
Patrick Turner wrote:

What are the physics of the situation as you understand them?

The key is: what forces on electrons - the current flow depends on the
number of electrons making the exodus from cathode region to the anode
per second, which depends on an electric field. If the anode is
screened it is reasonable to say the grid voltage is what is defining
the electric field in the important region where electrons are
"bubbling" at the cathode surface - whether they return, opposed by the
negative grid, is determined mostly by the grid (we can begin by
assuming), and sure enough the plate curves are pretty horizontal for
many pentodes at low current levels so long as the anode voltage is high
enough. So they are, to a first approximation, current devices
controlled by a grid voltage. In real life there is an influence from
the anode voltage on the electric field strength in the grid-cathode
region, and so the plate curves aren't perfectly horizontal, especially
at low anode voltages and highish currents....

Wow, that first sentence was terrible. Actually the whole message was. I
didn't proof-read it. Sorry. Pretty un-understandable (is that
"derstandable" then?). Here's another crack at it:

The key is: what forces are acting on the electrons that we hope might
travel from cathode to anode? Such details determine the device's
characteristics; every twist and bump in the device's curves, and so
what distortion components result, what "sonic qualities" we might
expect, and the boring-but-essential stuff like getting the biasing
conditions correct, all depend on bulk flows of electrons in the maze
they have to run to the anode. The electrons "bubbling" at the
cathode's surface (due to thermal agitation) may or may not be repelled
from their exodus to the anode based on the electric field strength
along the way, and won't have much chance of getting even as far as the
first grid if the combined electric field strengths from the grid and
the anode (and other elements) is very negative. [Physics lingo: the
resultant force vector for the electrons that have left the cathode due
to thermally-induced velocities, is mostly due to the electric field
that comes from the superposition of the fields due to the grids and
anode etc, so the acceleration depends on (largely) the close grid
voltage difference from the cathode, so only those electrons with very
high initial velocities - statistically few - will make it past a very
negative grid - at which time they have a good chance of getting to the
anode.] If the anode [and g2 if it exists] is held at a constant voltage
(as it is in most Ia vs Vg1 graphs for triodes and pentodes) we have a
(nonlinear) function of one variable and the curve is rather
parabola-like no matter which type of valve we use. So if the load
resistance is very low, triodes and pentode sound pretty much the same.

But in the real world, loads are never uniformly low.
Speaker loads vary enormously, so there must be a mechanism
to keep the signal source resistance low, either use a triode with its
internal NFB
or use a pentode with external loop NFB.
But even if the external NFB results in pentode Ra being = triode Ra
without NFB,
the two amps won't sound the same because of the spectral difference in
the THD.
So its a bit like comparing lemons to oranges, but some would say more
like lemons to strawberries.

As the load becomes lower, the load line drawn across anode Ra data
curves becomes
more vertical, and we see maximal 2H with a triode.
voltage FB becomes increasingly impossible as Rl falls towards zero,
so to linearize the current outcome we must use some form or current
NFB.
There is no current NFB within a triode; its a voltage thing.



If the anode voltage is allowed to vary (due to decent-sized resistors
in the anode circuit we kinda need to have in real world amplifiers!)
then pentodes and triodes behave differently; the difference being in
pentodes the effect of the anode voltage changes being screened from the
grid-cathode region to a large extent by a (presumably constant) nearer g2.

The *voltages* on elements together determine the rate of electron flow,
i.e. *current out*, so it is essentially a current-output device.

But all tubes do not have to be a current output device.
They can operate as pure voltage output devices, when fitted with an
anode
load that is a CCS where current CHANGE IS IMPOSSIBLE.
But the triode still has a FB network which controls the voltage output,
and the internal NFB is maximally applied when no current change is
allowed.
but the pentode has a near complete suppression of the NFB
opp, its like you have a shunt NFB network with R1 and R2, and the
presence of a screen
makes the R2 a very much larger R than it is in the triode.


But
at least one of those voltages, the anode voltage, depends on the
current, so the output voltage depends on several variables, one of
which is the output voltage itself.

But tubes operate fine without current change occuring as pure voltage
gain devices.
There is NO dependance on current change within the tube to produce a
NFB effect.
The NFB loop is the same loop regardless of I change.
Whatever loop of NFB exists in a triode or pentode just exists, some
equivalent R1 and R2 in a shunt arrangement, and what changes when load
is reduced is that
voltage gain reduces, and the so does the amount of applied NFB,
as applied NFB = 20 log (( A / [ 1 + { A x ß }] )) . A is gain without
NFB,
so as A becomes low, so does the NFB.

So a pentode connected to 10 ohms as a load has gain little different
to when its connected as a triode, where the addition of internal NFB
has almost no effect
on voltage gain.

That has two implications: one is
this is reminiscent of feedback systems, the other is that it stops
being a pure current source, but moves closer and closer to being a
voltage source as the anode voltage drops (or current rises) so much the
screen effect is reduced. At low currents and not-too-low voltages for
almost all pentodes the plate curve is pretty flat, and the current
source model is very good. But pentode and triode curves never get close
to being good voltage sources, i.e. the lines never get close to
straight vertical, so a voltage source is not a good model, just a
helpful approximation with we are happy taking gradients of curves at a
point and using them for small-signal calculations.

Well, a triode with Ra = 800 ohms ( 300B ) feeding a load of 5k has
an Ra a lot lower than the load, hence we conventionally
regard the triode Ra as being a voltage source, not current source,
which would be the case of you have a KT88 in beam tetrode with Ra = 18k
driving a load of 3k; here the Ra is 6 times greater than RL,
so the KT88 in beam tet mode is deemed to be a current source.



If we start with the (good approximation) of a current source, what do
we need to do to get the characteristics of a triode accurate over
non-trivial regions of the graph? Negative feedback. Simply biffing a
linear resistor in the equation doesn't do it. Also, negative feedback
circuits (be they g2-to-anode strapping, or cathode winding feedback in
Quad or McIntosh output stages) give rise to characteristic curves for
pentodes that look darned close to triodes.

Any pentode set up as a cathode follower will effectively
have Ra lines that look like a triode.

Patrick Turner.

Mark A
.

Relevant Pages

  • Re: On the absence of feedback in Triodes
    ... The "built in" triode NFB is a pair of high impedance paths, ... >>> voltage and then apply that voltage back onto the plate no feedback ... But voltage gain varies depending on load and NFB present as either the ... >>> The anode is the summing junction.. ...
    (rec.audio.tubes)
  • Re: a simple experiment
    ... >> applied between the grid and cathode, ... The output is a voltage ... >> between anode and cathode, ... > Lets do a simple experiment to confirm if there is feedback in a triode. ...
    (rec.audio.tubes)
  • Re: On the absence of feedback in Triodes
    ... >>FB is a voltage effect. ... Oh but the FB *is* internal or else the triode would have gain ... > that will feedback into the anode. ... its the net electrostatic field acting on the electrons after the ...
    (rec.audio.tubes)
  • Re: Pentode-Triode Sound
    ... If the anode is screened it is reasonable to say the grid voltage is what is defining the electric field in the important region where electrons are "bubbling" at the cathode surface - whether they return, opposed by the negative grid, is determined mostly by the grid, and sure enough the plate curves are pretty horizontal for many pentodes at low current levels so long as the anode voltage is high enough. ...
    (rec.audio.tubes)
  • Re: There is no feedback in a triode
    ... I forgot the Miller effect this time round. ... If the load has too low a value, little anode voltage change occurs, so there ... the applied VOLTAGE NFB is maximal when no current change occurs in the triode. ...
    (rec.audio.tubes)