Re: Pentode-Triode Sound
Westley, my old gabbas, you've been on RAT long enough to
Not like the archetypal crony to fly his true colours. Are you
standing on a box?
Or a pair connected in push-pull to minimize the 2H would make a nice
Yes and indeed variable µ tubes WERE USED in some of the best sounding
Your logic is just plain wrong because there's nothing about feedback
that requires it be 'the same' if done one way vs another.
As I said before, if you'd spend 2 cent's worth of thinking about
what's already been given to you, instead of trying to invent ways
around it, you'd get it.
As soon as you stop making personal remarks we can have a civilised
Your propensity to flat ignore and snip out everything proves
and auto level microphone amps around.
I would think having a 5 band compressor would allow low IMD of the HF
when the bass levels are gain adjusted by the bias variations....
No it absolutely is not. Variation in output due to load does not
require feedback in the sense we usually use the term. Quite the
This states the obvious. you can either have or not have NFB, and to get
voltage change at an anode, you don't need NFB.
Precisely, so the fact that the anode current varies with plate voltage does
NOT prove NFB exists in a triode.
reverse, in fact.
And what is that Ian?
Pentode and beam tetrode have varying 2H and 3H and other H according to
whatever load value is being used.
2H is like a triode when RL is low, and the 2H phase relative to
the fundemental is like the triode's.
As RL is raised, 2H gets lower, reaches a null, then increases again as
is raised above the value for minimum 2H.
But at the high RL values, the 2H phase relative to the 2H below the RL
for least 2H is reversed,
and careful load line plotting will reveal this behaviour.
Triodes don't suffer the phase reversal of their 2H as RL rises; 2H
to a minima.
Yes, but you have ignored my point about the relative levels of 3H and 2H in
OK, in pentodes and beam tubes the 2H varies releative to 3H which also
output and so the 3H : 2H ratio is variable.
Triodes always have mostly 2H, with 3H usually well below the 2H levels,
so much so
that an SE triode might have 5% 2H and 0.5% 3H.
But when two triodes are in class A PP, there is very tiny amount of 2H
and about 1% 3H at the clipping level.
The various classes of operation change the mix of the two most common
There is more about it in the books you should be reading.
I realise that, but I now also realise that just about everyone contributing
to this thread except me is talking about power amplifiers. I was not
thinking particularly about power amplifiers, in fact I was considering
small signal pre amps initially. Start with the simple case, understand
that and work up to the more difficult one is my philosophy.
But anyway, usually pentodes and beam tubes have a lot of distortion at
clipping power quoted for the load, and much more than a triode, and the
mix is more complex, so you must apply 20dB to at least get the Rout
down to acceptable values.
And behold, the second order IMD products begin to be suppressed well
but 10dB NFB around a pentode amp only has a mild effect on distortions.
3 times more than 10dB is 20 dB.
The dynamic Dn of a pentode amp can make transients more impressive;
each drum beat
causes a spray of extra IMD...
But in lousy SE pentode SE amps such as a lone 6V6 in a mantle radio
with no NFB,
the speaker has rising response until it cuts off at about 8kHz, and
the Dn products compete with the attenuated real HF products in the
music or speach.
( The RF parts of old AM radios rarely produced more than 3kHz of AF
Hence we have unique "old radio" sound, with artificial boost to the HF
Guitar amps get by on a similar principle.
All devices are basically ridled with non linear behaviours without any
FB of any kind.
BJTs have linear current in/out curves, but make lousy voltage amps
huge amounts of both current and voltage NFB.
J-fets and mosfets have attrocious voltage linearity, but Dn is mainly
still you need I and V NFB.
Tubes are more linear than SS devices, but are useless without NFB.
Every piece of gear used in the analog recording chain from mic to
cutting head had lots of NFB.
I doubt one single recording any company ever made with a full bandwidth
after 1955 when some fabulous recording were made without loop NFB.
Companies using triode based gear used less loop FB than the pentode
based products full of EF86, EL84, KT66 etc.
All devices depend on correctional facilities available.
The self regulating triode is the most forgivable criminal,
and he controls his urge to act uglililly in public.
The pentode or a beam tetrode requires a stern and firm handed warden,
Senior Loopaldo Neegartiff Feedbak, who is a fair minded giant.
But even he cannot ever eradicate all spurious behaviours.
Dept of Custodial Singing Lessons.
Well, Ian, just what is the effect of the screen in a UL amp?
If you'd read a couple of my posts on the subject you would have seen
the advice I gave about breadboarding a 6AU6 or 6SH7 as a simple commom
cathode gain stage
then using a 1/2 6SN7 as a follower in a µ follower so you get enormous
open loop pentode gain which external shunt loop FB can reduce to line
level gain of about 8.
This can all be compared simple 6SN7 topology.
I agree you should start with the simplest amp and work up.
Its a rewarding pastime.
The tube's internals don't rearrange themselves one way or the other.
Still works the same.
I realise that, but several respondents simply raised complications due to
high power levels that I felt did not comtribute to my understanding of the
question at hand.
I just used UL because it's exceedingly simply and direct with an
uncomplicated continuum from screen feedback to 'triode connection',
not that it seems to have helped any.
No, I did not find that it did.
How does the screen manage to reduce odd number harmonics?
I am not sure but UL is a special case where even harmonics are already
cancelled by the basic push pull topology. This confuses the issue rather
than helps resolve it.
UL is not married to PP and you can do SE UL just as easily. Simply
pick yourself up a Hammond 1627 series SE UL OPT. UL tap at 40%.
Instead consider an EF86 connected as a triode. Can you connect it as a
pentode with feedback to obtain the identical characteristics? I don't
think you can.
So the characteristics would not be the same then because the THD spectra
Of course you can, because you have. That's what 'triode connection'
does: applies 100% feedback because you've got 100% of the plate
signal going into the screen by the act of connecting them together.
I know you can connect it as a triode - the point is wether the
characteristics are identical to its pentode connection with external NFB.
That *IS* "its pentode connection with external NFB," set to 100%.
That's the point, it isn't.
Your supposed 'point' is just plain wrong.
Think of the UL tap. Tap the OPT at 40% for 'sweet spot', as Hammond
calls it, UL or move the tap to 50%, then to 75%, then 90%.... where
is it at 100%? On the plate, I.E. 'triode connected'.
How does the screen make the pentode achieve Dn spectra approaching
But he's right, and its tough to have to hear it.
Enough has been said. Time is ripe for serious experimental action.
Typing BS into a PC has all been done adequately well.
I do not see any way around actually soldering something together and
doing all the hard work of measuring and analysing what you have built.
I spend thousands of hours in my freezing cold or baking hot workshop
to understand tubes a bit before I came online here.
We all expect you will put in a great big effort like we did.
You have our help, but you gotta work at things.
I don't see how any amount of work in the workshop, at least what we
tube amp fanciers think of as workshop, is going to resolve the original
question? IIRC the original question/disagreement had to do with
whether or not triodes have internal negative feedback. While it is
obvious to me that they do, some are fussy about what constitutes an
"input" that it appears to me that if triodes don't have internal
negative feedback, then negative feedback doesn't exist anywhere and is
a pure fantasy.
A pentode can be used so the screen is the anode for an ac signal
Pda is limited to a few watts in a power tube, but Ea could remain at a
or operate into a low value load; its voltage makes little difference to
triode operation if the screen is allowed to move.
This operation has no value in power amps, but for low noise operation
as a triode, the screen is used as the anode, and the Ia is quite low,
and the anode is grounded.
Gain and Ra are triode like, and like you get if the anode and screen
are tied together.
The screen voltage has the same electrostatic effects inside the tube as
an anode placed
where the screen is placed.
This post tells us all you can see 3 metres into the fog in front of
There is a fence post there, watch out.
If the fog ever lifts, you will see further, and see the other fence
and you'll see wires connecting the fence posts together,
and then you'll see the fence goes around a farm called Negative
Maybe if we are lucky, you'll see there are interactions
between the animals on the farm; there's Anode, a large hairy beast, and
he pulls the plough,
Cathode, a smelly small hot creature, he likes to root around in the
Earth, and Grid,
a small lively animal which hardly eats much, but controls the lot.
Then Farmer B+ brings the food for them all,
and the Electrostatic fields provide the very sustenance for all the
creatures on the farm while they
plod around on the farm.....
And that's your beddie bye story for tonight, sweet dreams.
The notion of internal feedback is neither necessary nor useful in
explaining either the triode or the tetrode. It may feel and sound OK
to think of it like that, but it does not bear analysis, and obscures
whatever the truth might be.
Great, place your head in the sand. I think you are Egyptian, Ian.
Methinks this is all splitting hairs.
One may theorise all day long, and never really be able to conclude
If your experimental work is just done, instead of talking about doing
you'd have something real to decide about.
Most good tube preamps have less than 0.02% THD at operating levels
if you have a gain pot followed by a gain tube that feeds the power amp.
If you have the cd player feeding a gain tube and place the volume pot
gain tube, the cd player signal will be amplifed from say 1Vrms to 8Vrms
and THD will be much higher, but I can say that it will still sound
In my preamp I use I can switch out the gain tube and just have cd
player ( or phono amp output )
feeding the volume control, then a cathode follower buffer without any
So levels are not amplified at all and thd remains below 0.02%.
Initially I enjoyed building just about every type of preamp and power
circuit ever invented, but after time I would draw up what I wanted and
just do it,
and the results are at my website.
I much prefer using a CCS dc supply to a gain triode, this can be 6CG7,
with very mild shunt NFB, gain max of 7 times, so 15dB, this feeds a
then I have a cathode follower with CCS cathode load.
I've never used a pentode yet. Don't let me stop you.
I also like the µ follower with 2 triodes.
Anyway, there are plenty of good sounding schematics at my site to play
Its easy to set up a trioded KT90 in a circuit designed for a 300B
if the socket, heater voltage, and bias is adjusted
for the same 32 watts of idle Pda.
I bet nobody could tell the difference.
You are in de nile!
YOU are denying a truth. YOU are the one to obscure the truth,
to tell us not to enjoy its beauty and presence.
You feel the NFB isn't there, you like the simple sound of your mind
saying it isn't there.
We don't mind the analysis.
As for strict definition, a triode is not a feedback system because
the output is not referred to the input. You may posit some internal
Wrong, it IS refered to the input.
subsystem in which that is true, but that becomes a trivial exercise
because there is no delay.
Please speak the English of your country Ian.
You've drifted in gibberish.
We can't find a translator for Gibberish.
The people of Gibber have a lousy time running their nation state
and cannot conduct family affairs or run businesses.
A triode with an anode load and a resistance at its grid becomes a
feedback system via the miller capacitance, which is real,
non-trivial, and significant, since we must take into account the
delay it introduces.
Which is why we use the disclaimer, "as if". The old books talk about
the action of screen nad suppressor grids until they are blue in the
face and apparently to little avail. I am a poor typist and don't care
to copy by hand (and, now that the CapShare scanner has apparently
been quietly removed from existence,don't have a scanner.)
Read the books. It's in there.
Oh and also don't forget the mountain of pro-NFB-in-tiodes evidence on
Is that the best you can do? It is completely erroneous.
Unfortunately for Ian Iveson, there are very few sources of Web based or
book information that would support
his idea than no NFB exists in a triode.
That is silly. You don't expect to references to something that does not
exist. None of the major texts mention it simply becuase it does not exist.
And they knew all about NFB back then; they had whole chapters on the
Actually it isn't. What you are mistaking for feedback is the simple result
of anode resistance. The reason the books don't call it feedback is because
More gibberish. And its also a red-herring.
At AF, the Miller effect is quite negligible
because it depends on capacitances to function, and signal triode Cag
are very small, and have no effect at AF unless the source R driving a
is very high.
What you fail to accept or understand is the net interactive
of both grid and anode voltages upon Ia, and where Ia remains constant
when the triode becomes a pure voltage operated device with maximal
No, feedback needs to be to the input. Feedback to the screen grid is not.
Such electrostatic effects are utterly distinct from the Miller
effect which does generate current flow even at LF although it is
In a triode strapped pentode, the screen, G2, is an active input port
control of anode current, or anode voltage only in the case where
the anode has a CCS load.
When you begin serious analysis of a pentode such as an old 6SH7,
set up the tube as a pentode with G2 biased at say +100v, but taken to a
variable low impedance signal supply.
Set up the G1 with 0V bias, and have a cathode biasing R & C network.
This will allow you to measure anode gain as a pentode, fixed G2 supply,
or triode gain, with anode connected to G2,
and also to measure gain with G1 shunted to 0V, and driven by G2.
Pure pentode gain is very high.
Pure G2 drive is very low, and for all tubes,
voltage gain A = µ x RL / ( RL + Ra )
So for any condition we can use say a load of 100k, or 50k, and get two
different gain figures
measured at say 5vrms in a 6SH7/6AU6, and thus work out the Ra, and the
using simple schoolboy algebra.
From knowing µ and Ra, you can work out the transconductance, gm, for
the G1 triode case,
pentode case, or for G2 drive with fixed G1 supply.
In a triode strapped pentode you will find that there is a mathematical
between the gm of the G2, and the G1 that leads to further understanding
of the intimate
mutual effct both G1 and G2 have upon the electron current in the tube.
The anode voltage has almost no effect whatsoever, but just tags along
for the ride
and absorbing most of the the electrons because most heading north after
passing the G1
also slip past G2, because the grid wires of G2 are in the same line as
G1 grid wires,
or as ppl say the G2 wires are aligned to be in the G1 wires' shadow.
Once you realise that the anode is signal is the result of TWO
electrostatic field effects
having a real and non virtual summed effect on the anode current, then
you'll understand the FB action.
Ordinary triode strapping has lower gain because the screen signal is
sending an opposite phase signal
electric field to the space charge of electrons around the cathode.
And its the same signal as at the anode, because the anode and G2 happen
to be tied to each other.
The G1 signal may move +1V, and have a large effect on Ia, and the G2
signal may move
-100V just as the anode moves, and because the G2 is MUCH further away
from the cathode space charge,
the G2 voltage does not have as large an effect as the V change at G1,
which controls matters mostly.
The G1 will have high gm, the G2 low gm, as you should have calculated
from the experiment above.
But G2 action is enough to reduce the pure pentode gain a lot, from say
100 to say 25, with say a 50k load.
The summing effect of the TWO changing electrostatic fields in a triode
is like having a shunt NFB resistance
network between anode and input, but thinking of it this way involves an
imaginey model for the tube with a
virtual input point.
But in areal triode, no such virtual or imaginery input point exists,
and no resistances
are involved, just electrostatic fields, no different to those you
with a plastic comb on a dry jumper when you rub it, excet
that the comb charge can be thousands of volts, but the G1 or G2
can be from 0V to say +200V.
I cannot stress strongly enough that if people do not do the
experiments, record all they do
like a properly done school science experiment, they will learn nothing
real about all this.
)V xearth,rmas sc
relationship between rapsuestablish the either triode orallowed to 2 G
Well, looks like you don't get it like Ian Iveson.
Read some MORE books. I would suggest Terman's 1937 Radio Engineering.
I have read lots of them and not one talks about inherent NFB in a tride.
Name one respectable book that does.
Or study the thoughts at
This is not an easy tome to understand because its full of math.
But you don't need math or Stockman to lead to concluding NFB exists
as a reality in every triode.
I am very good at math. This article is full of errors.
The low Ra of triodes is the direct result of the NFB.
No it isn't. If triodes were pentodes with NFB then their distortion spectra
would be identical, just altered on overall amplitude by the NFB, but the
relative amplitudes would be the same. As their spectra are in fact quite
different one can only conclude that triodes are not like pentodes with
One can't conclude any such thing from the spectra. This statement
underscores your complete lack of understanding of triode operation and
the negative feedback within. Adding feedback to a circuit or device
Mostly agreed. There's an additional subtlety in this case, which Patrick
pointed out. The subtlety is that the feedback is applied through a
mechanism that is itself nonlinear. This is relatively rare. In fact its so
rare that I managed to do most of my post-graduate work in engineering and
control systems without it being mentioned.
does not simply alter the overall amplitude of the distortion spectra,
it interacts with the nonlinearity in the device or circuit to not only
completely change the relative amplitudes of the spectral components,
but can even add new spectral components that weren't present in the
output of the circuit or device without NFB applied.
I suggest you check out pages 308 and 309 of the radiotron Designers
Handbook (1960 edition) and readily available on line for the math that
demonstrates my point. Its conclusion on the effect of voltage negative
feedback on harmonic distortion is:
"That is, the magnitudes of *all* the harmonics (and of course the
intermodulation products) introduced by the amplifier are reduced by
negative voltage feedback in the same proportion that the gain is reduced."
The point you are missing about the distortion spectra is that the
introduction of negative feedback introduces new distortion products and
that changes the relative spectral amplitudes. Your analysis assumes
How am I missing that? The maths is correct is it not?
that the feedback signal itself does not produce any new distortion when
passing through the amplifier.
The math makes no such assumption.
The RDH makes a correct statement. However that correct statement means that
the relative amplitudes of the distortion spectra will remain the same, only
if the reduction in gain at all frequencies is identically the same. A
triode amplifier will tend to have flatter response before loop feedback is
applied, because the plate resistance of the tubes that are changed to be
triodes is lower. Thus there will be differences in the reduction in gain at
This is true, and in general the gain will be lower at higher frequencies
which mean the higher harmonics will be reduced less. This again means that
the triode internal NFB theory fails because pentodes have relatively more
third harmonic distortion than second.
This is quoted out of context.
On page 309, the Bible goes on to state 3 conditions under which the
statement above is correct.
Nearly all amplifiers operate outside the conditions stated
for full equal reductions of harmonic content.
Within the audio band those conditions are true.
And RDH4 does not explore to what extent new additional higher numbered
harmonics of frequencies above the open loop
spectra are produced by the NFB.
Are you saying the math in RDH is WRONG? or INCOMPLETE? If so please point
to to a reference where it is corrected.
The RDH4 is incomplete, it does not consider new harmonics are generated
by the feedback, see the third assumption on page 309.
Yes, a lot depends on 'how hard' the valve is driven. In preamps the open
loop distortion is small to begin with because the signal levels are
relatively small so the third assumption is not an issue. In power amps
things are different - however, that should be immaterial if the 'triode is
You have taken a paragraph out of context. I was talking of Stockman.
There is no delay in his account, so on that count it is at best a
trivial application of control systems theory. On other counts it is
also obviously only notional.
No your delay does not invalidate my argument. Just because all
feedback control systems incorporate delay does not mean that all
things which incorporate delay are feedback control systems.
Elementary logic: the difference between a necessary and a sufficient
There is always delay between the input and output of a real system. I
haven't argued that it is necessarily true, simply that it is true. It
is a matter of fact. I don't know why. God would know, perhaps. Why
ask me when you just said in another post that I have no brains?
Perhaps brains aren't necessary. Maybe god just knows without needing
to mull it over.
In mechanical systems it is necessary that an extant thing has
extension, and as a matter of fact all materials have elasticity and
mass and some element of damping. Consequently every system made from
real materials will incorporate some delay. In electronics, extension,
capacitance, inductance and resistance are equally inescapable. Can
you cite a system in which there is no delay? Perhaps without such
delay there would be no time in the world. It's hard to speculate
without a brain.
But all of these concern the meaning of "delay" that we rather take
for granted in audio, and in control system theory in general: group
delay. We are generally safe in the assumption that our outputs will
start at the same time as our inputs, and only need to worry about the
fact that they take longer to happen, and so we get phase shift.
Delay is the raison d'etre of feedback control theory. If there is no
delay, then you are left with only the common time-invariant "feedback
equation" which can be derived in one line of simple algebra. Nothing
else applies. Bode, Laplace, Nyquist, et al....the whole body of
analysis returns nothing.
A delay arising from the (presumably light-speed?) travel of
electrostatic fields within a valve would be absolute AFAIK, in the
sense that the output would start some time after the beginning of the
input. That makes it rather a different animal from that arising from
combinations of reactance and resistance. Something we don't normally
need to bother with, fortunately, since I don't find it so easy to fit
such delay into systems theory.
What kind of delay is yours? Is it considered to be a propagation
delay, the time it takes to establish a standing wave perhaps? In that
case I guess it would be group delay. Or is there considered to be an
absolute gap in time between the departure and the arrival of
electrons? How are the velocities of electrons distributed across the
population at a particular point in time? Perhaps it can be seen
either way. I am not a physicist, and have never worked in an area of
engineering where such small time intervals are significant.
In any case, it is not what Stockman is on about, and has no bearing
on whether a triode has real internal feedback or not. Delay happens
in both open and closed loop systems.
Anyone know how a maser works? That might cast some light on the
Incidentally, I am using this term "absolute delay" in the hope that
the meaning is obvious. Is there an accepted term in electronics
My question about the necessity of delay for negative feedback to exist
was just an aside. My point was that there is delay in the triode,
which invalidates the argument that you put forward that one reason
negative feedback couldn't exist in a triode was because there was no
delay in a triode.
a pentode with NFB' contention is true.
There is much the RDH4 hasn't said because of the space and time
constraints on its team or authors.
But many bright minds came along after RDH4 to shed a much brighter
upon the subject, but their combined wisdom was never covened into ONE
because the world was a changing for where old traditions were being
and solid state had become so invasive to the status quo, and so rapid
that writing one really good about electronics was pointlesss, because
by the time
the book was published, many techniques were obsolete.
And the sheer amount of electronic information became so vast....
electronics was total modern for ,
In short, even if the feedback path is linear, there's a tendency for the
error components to simulataneously be reduced and also to pick up more high
order components. The usual effect is that all orders of distortion are
reduced, but there is a shift of relative amounts to higher orders.
Since you claim to be good at math, try putting feedback around a simple
square law nonlinearity and see what happens to the distortion spectrum.
That's a good first exercise. The second exercisemight be to apply the
feedback through a path that is itself highly nonlinear.
Pass, it has become so convoluted it has pretty much become meaningless. ;-)
Compression should be about reducing the amplitudes during the wave
but it seems it would be impossible to do
it without creating lots of 3H. I find it hard to believe real time
compression does not add a lot of distortion, unless there is a phase
circuit that detects the wave coming which needs to be reduced in
then adjusts the bias and hence gain in advance of the of wave passing
through the compressor.
This way nothing is limited, its just threshold amplitude reduced, and
probably is a tool against noise, except drum beats get attenuated as
People say they here the compressor "pumping" when the bias varies.....
Of course compression is all BS really, and should NEVER be needed,
because we are
interested in the recording of the real event, and manipulation of any
kind after recording it
is not in the interests of real hi-fi.
But the pop music industry would never agree with me, and I couldn't
Most pop artistes are wasting their time on stage, and sound like tom
with a cracker up their arse...
It is your right to believe anything you like. I guess I am a born again
sceptic; worse, I want to UNDERSTAND what the gods are playing at. At the
moment I don't understand what goes on in a triode.
The basics are simple. Current flow is governed by the sum of two
electrostatic fields, the control grid and plate. Both are 'inputs'.
And since the 'output' is internally connected to one of them, in
particular the plate, any voltage signal created by the current into
the load is feed back via that electrostatic field being summed with
the grid electrostatic field.
But what you are saying here does not explain that triodes or any other
types of devices can be pure voltage
operated gain devices which require zero current change at their output
terminal in order to
sustain a predictable voltage change.
We see evidence of NFB in a triode WITHOUT ANY ANODE CURRENT CHANGE.
How do you factor this fact into your reasonings?
You must have posted this question before you read the other message
because there is no voltage change without current flow.
To criticise a statement accurately, you need to refer and quote the
"other message" exactly. Which one was it? I make a lot of posts,
and here you are saying something that is incomprehensible to many in
the group imho.
Once a dc idle current is established in a triode connected with an
which is a constant current source, voltage changes occur which indicate
of intrnal NFB. Voltage and field related phonemena occur
according to laws and equations without current change within them.
Using the the screen as the control grid with fixed g1 voltage gives a
low µ high Ra triode, almost useless in audio amps.
Fairly linear though, and some have done it. Gain is about 2 or 3.
Applied NFB could be fed to G1, but you could never apply all of the Va
because never enough VG2 could be applied to create gain.
Only a small % of Va could be applied to G1 and then you'd need
2 or 3 times more G2 drive, as much as Va, and creating the drive
creates more THD, and it all becomes a silly exercize.
Cathode feedback windings from the OPT are much better, with fixed EG2.
know it is never this simple!
The fly in your ointment is, as always, global negative
feedback, which can homogenize anything. That's great for
those who eat preprocessed, pre-sliced, predigested cheese
on tasteless white bread. But it isn't why you spent the money
on a tube amp.
We'll assume that your amp is playing in Class A at all times,
which may or may not mean modest volume, and that your
system is well matched. Assuming on top of that there there is
NFB in the order of 20dB, you will hear almost no difference.
Assuming 12dB of NFB, which is then added to the native NFB
of the tube when it is trioded, I still don't think you will hear much;
Patrick in his long screed has about a dozen other things that
might or might not make a difference.
However, assuming Class A operation and only 6dB of NFB in
the amp, chances are that you will hear that the triode amp is
smoother, warmer, more pleasing, easier to listen to for a long t
ime. If you could manage to remove the NFB altogether (it might
mean snipping two or four wires or disconnecting some
components on a PCB and fitting jumpers instead -- you gotta
put it back the way it was before you operate it in pentode again)
you will definitely hear a difference. At these modest volumes
we're talking about, average listening volume, you will find the
pentode amp a little unrefined, you will find a UL or CFB amp
more refined, but your dissatisfaction with them will grow over
time; they will make you edgy. Notice that we're talking about
a good amp here, not one in which you can consciously hear
distortion in any mode of operation. We're in fact talking of a
subliminal effect, where in the pentode amp the lower level
(absolute) mix of higher harmonics after NFB impacts on
your enjoyment in a way that you will find difficult to pinpoint. I
n fact, as Patrick may have been trying to say, the triode version
with low or zero negative feedback cannot be made as t
heoretically clean on a distortion meter as the high-NFB
There is no point in arguing about this with those who still cling to
the idea, Ian, as you will see if you look back over the "triode
Only God could be Complete, or Perfect.
His Creations and his Destructions have reasons we cannot understand,
and no book can enlighten us fully, not even the Bible.
It is our Human Condition to be alone with whatever information exists.
We are mere men, and so were our fathers who wrote RDH4.
Forgive them their shortcomings, as they forgave us when they
brought us up.
I cannot point to everything you need to find.
You must rely on your ability to search high and low for knowledge,
and on your ability to work out the math, not ours.
I am pleased you admit your are not perfect and that you could possibly be
Ian, you are standing higher, on the roof top, and shouting BS to the
Being perfect requires lots of people to lie about you.
One can only ever really be partially correct, and partially wrong.
Newton gave us laws of motion which seemed SO RIGHT.
but you could never get a rocket to visit planets of the solar system
just relying on Momentum = 1/2 m x V squared et all.
He was right within his limited views, and for our little world, Newton
was pretty right,
if all we want to do is pull a train with a locomotive.
Is this really true, that it requires more than Newtonian Physics to
send a rocket anywhere in the solar system? I don't know one way or the
other, I never thought about it and just assumed that Newton was all we
needed for this simple task, so I am curious.
Nah. We went to the moon with slide rules. It ain't
My question related to your statement that Newtonian physics wasn't
adequate for interplanetary travel within the solar system, and now even
for travel to earth's moon. I will put the question to another group
that may be able to direct me to the answer with respect to navigating
to the moon and back.
The closest we civilians come to relativistic speeds
is the electrons in our CRT's, which achieve relativistic
mass gains of maybe 5%. Navigation within the solar
system is Newtonian to a very large number of decimal
You must be kidding. Not only 'spooky action at a distance' stuff but
things 'magically' changing mass?
Relative mass changes.
If it ain't got the mass it had when you last saw it, perhaps its got a
lot of relative
Add the two together and there has been no real change.
Chances are you'll never witness any mass changes in anything because
anything fast enough to swap mass for
energy will go past you so quick you won't notice it.
Perhaps it blows you to bits.
I did not raise any questions one way or the other about the need for
computers to implement the navigation equations, be they Newtonian or
We really did go to the moon with slide rules. You
can't make this stuff up.
Much thanks, as always,
I just thought getting a space shot to "slingshot around"
other planets to eventually arrive at Saturns rings and get a close up
took more math than could be worked out with a slide rule within our
I suggested relativistic effects be taken into account, even if they be
A moon shot might be very easy, just step on the gas and aim towards the
moon, and watch
it get bigger out the window.
But maybe the slide rules were special...
Much thanks, as always,
Then along came Einstein, and he had a few visions about the universe,
and he knew a good mate with much better
math ability to work out how bits of the universe related to each other,
along with energies
of many varieties. Some things he discovered existed only in theory, to
be verified much later by observations.
Einstein made a few blunders too. Women flocked to him because he must
have been a great
I heard he gave up playing the violin at 70. There are some things
you can't stay doing when they turn out wrong.
As the known universe becomes far bigger the more we look, it would seem
we cannot ever comprehend the infinite amount of information describing
because we have only finite brains, only marginally better than a dog.
As we disect atoms, we get Russian Dolls, ever more smaller bits and
pieces and they last
for shorter and shorter amounts of time, and we begin to see we know
less, rather than more.
There would always be something a quantum physicist may say about a
amoung us no quantum guy is to be heard...
Newton would have been fascinated were he around now, not to mention
having Leonardo Da Vinci
on the NFB problem. I'd much like to have Leonardo, Isaac, and Albert
in for dinner one evening.
I'd probably never follow what the heck they were talking about.
But when Liz Hurley visits me (for dinner), she doesn't have a clue
but she likes a big pentode.
Patrick Turner, triodolically addicted.
feedback" debates over the past couple of years.
Just as feedback theory can be used to arrive at ohms law for a
resistor, so it can be used in deriving the triode function. It is a
trivial application of the term "feedback", is merely notional, and
understanding is better off without it.
I've read them and you've missed every point everyone has presented,
as well as the plain fact of it.
Then perhaps you would do me the favour of spending a little more time
trying to explain it to me and answering my queries, as Patrick does,
rather than indulging in personal remarks.
I've already posted the governing equations as well as numerous
The equations are simply fact, but all you do is snip them out and
The equations are NOT fact. They are simply a way to represent what is going
If you want to quibble with the definition of 'fact' then so be it.
But they are as much of a 'fact' as can be established by science.
on. The equations are no more real than the the equivalent circuit. No one
So is language. Although, unlike 'the spoken word', mathematics is
I having not much luck with an ESL kit I am building for a client.
The first one gives really abysmal sensitivity, I estimate
12dB below my 88dB/W/M dynamics, and the ESL just won't go as loud as
do when you do have a watt being used without making strange noises and
The theory says they should go louder, and I have to see what I am doing
You've sure got your work cut out for ya. Sensitivity falls
off with the square of spacing distance, and everything is
tighter than Andronicus.
This wouldn't be a kit with the name David L. attached, by
any chance? Comes from the States? Sure hope not, by local
I will not be disclosing the name of the kit provider because its
work that is at fault, not his, and I need to establish the truth,
and without unecessary angst, but nevertheless leave good advice to
others considering building any ESL kit.
Maybe I have to re-membrane te panel i have just done; so be it.
There are some things one has to attempt several times before you really
The subject of ESL building really needs a new thread title, and I may
include a page about it at my site
in the fullness of time.
I doubt anyone else is presently engaged in ESL building.
Much thanks, as always,
I'm no making the second of a pair, and maybe I get better luck, and it
will tell me what I done wrong with the first one.
If it is the same, and I have not done anything wrong, then its a dark
alley I will be then...
ESLs are kind of unique. Back when, a friend had a pair of Quads. He
said they took forever to "loosen up" but then, he learned to love
them. Years later, a new signal generator with a lot of gin on a long
cold winter's night, he tried a 20 Hz square wave ... bright blue
flashes appeared and then silence ...
dark alleys are real, but not the end, just a boring bit in the mix.
If you deliver matching performance in both channels, the customer can
wait for them to 'loosen up'. Others have ...
I have poured much time reading all i can about membrane coatings,
classic pattern ESL, inverted ESL, step up transformers, glues and
so it now is not all that hard to re-membrane a panel in the kits I
the surround slips off, there are 12 nylon bolts to undo,
then the membrane that could be dodgy is sanded off the spacers and
another tensioned up and glued on,
then next day all is re-assembled. Its skilled work, but rather easy for
used to being practical. I am not kidding. But one slight drop of
anything sharp near the bare membrane
will puncture it, and you must start again.
Many of my friends, aquaintances and customers
have almost zero technical ability, and they'd find a way to stuff it
But even without looking, seems like I have, despite my care.
I'll use thinner coating on the next panel; I think I used too much,
and resistance is too low, or varying too much.
One sure learns by doing......
And I am not using 20Hz square waves after drinking lots of gin.
I am using pink noise though, and my dynamics can handle it without
producing very loud sound, but not these ESL, they start arcing and the
at only 3Vrms input, so i had to make a high pass filter at the input
with a 150uF NPcap. This becomes resonant with the low amount of step up
inductance at about 70Hz, so the C and input then need daming R across
the C and
primary of about 47ohms to damp the Fo, and stop a low Z at 70Hz.
Ppl forget that a secondary of an ESL step up tranny may have enormous
no of Henrys
of inductance, but because the voltage ratio is about 1:100, at the LV
The simplest CCS is done with a pentode with large cathode resistance.
Say you have a 6AU6 pentode set up with Ea = 150V, Ia = 5mA, then Eg2 =
100V, but well bypassed to the
cathode, and then have 20k Rk taken to -100V, then the anode input
= Ra + ( [µ + 1] x Rk ).
Ra is about 500k, and gm about 0.004A/V, so µ = 2,000.
So Ra efective = 500k + ( 2,001 x 20k ) = 500k + 40,020k = 40,520k =
Using a pair of seriesed 6AU6 so that one pentode's 500k of Ra is the
other's Rk will
make the Ra of the top pentode gi-normous.
The screen supply resistance of IM is in parallel with the Rk, but makes
little difference to the
One single MJE340/MJE350 can easily make a simple CCS with real R > 20M
Using a pair makes a CCS whose real R is very hard to measure.
500H is transformed down 1/10,000 times to 50mH only.
I have had to carefully monitor all i am doing, but I did manage a near
100Hz to 15kHz, but sensitivity is VERY low, and cranking up the EHT
didn't seem to help much
except cause some sputtery discharge noises in the bass panels.
At the moment the treble strip is the real problem which should be
giving a much higher level
above 1 khz, but it ain't, and I suspect way i applied the coating is
And If i do manage to lift the treble panel output 12dB as it needs,
the bass-mid panel series R can be reduced to ge that level up to match.
This will leave the low bass sagging below 200Hz,and nesesitate a decent
full range bass speaker,
something my customer probably doesn't appreciate yet.
Martin Logan 'Summit' model use a dynamic bass box, so they don't have
fooling ppl into believing there is more bass there than there actually
ML quote 92dB/2.83V at 1M, and I wish to see something like this from
just in the same way any diyer can get 50 watts from two 6L6 just like
The mid band Z of the ESL kit is quite high at 12 ohms at 1khz, as
but its only 1.5ohms at 20kHz, like many other ESL.
Amps do not find this difficult to drive because there is such a tiny
amount of energy by 20kHz.
I think a ribbon would be harder to make, and making one's own dynamic
drivers is almost
impossible at home...
Oh, and IMHO, speakers don't magically "loosen up", or run in very much.
If they measure apallingly when new, in 5 years they will be worse or
only marginally better.
They should at least measure well for response, and then you have a
chance they will sound well.
OK, listeners may 'loosen up'. Whatever, I am enjoying old tubes in a
new amp through old ehorns. Perhaps a bit more this week than last,
but, I was sicker last week ;)
Yes they are because they are simply the language of first principles
and the laws of physics, not notational conventions to 'conceptually
simplify' things into 'familiar terms' for ease of use.
That is also why I went back to Maxwell's original work before 'tubes'
is going to suggest that a triode REALLY consists of a voltage source in
series with a resistor even though that is commonly how it is represented.
That's right. Because those are 'models' to 'simplify' and put the
thing into 'familiar terms'.
In the same way, just because a triode's characteristics might vaguely
suggest it could be considered as a pentode with feedback does not mean
that is the case in reality. However, if the math that describes a triode
There is nothing "vague" about Maxwell's determination of
Veff=(Vg+Vp/u) from the raw electrostatic fields, irrespective of any
'device'. It's simply laws of physics.
But there is something vague about the Maxwell equation, Veff=(Vg+Vp/u).
What and how id Veff measured, and what and how is µ and how is it
All equations are meaningless jargon unless the basic definitions
accompany the terms used in them.
So could you please explain the derivation of Maxwell's equation, and
define all the conditions please,
to show you know what your'e talking about, and to inform the group.?
Use 1/2 a 6SN7 as an example basis.
I don't think going through a Maxwell derivation would be at all 'un
vague' and I'm not mathematician enough to un vague it.
However, there might be an easy way to visualize it by considering
light. Take a sheet of paper and two light sources at different
distances from the paper. It's easy enough to observe that the
'effective' light on the paper varies with the light source's distance
from the paper (demonstrating similarity to the u term) and that the
'effective' light at the paper, when both sources are on, is the sum
of the light from each source.
Granted, not quite 'identical' to the field equation but it's a
similar principle and 'illuminating'... twa thump thump ;)
Adding onto my previous post, maybe this link is clearer that my
And there's nothing "vague" about the same laws of physics operating
in electrostatic beam deflection, with electrons wandering between the
moon and earth, or a triode where Ip=K(Vg+Vp/u)^1.5
I don't have a problem with this formula, and a curved line can be drawn
showing increase in Ia for the increase in Ea, and the line describes
the Ra of the tube.
Pentode Ra lines and triode Ra lines are very different.
If you mean the typical tube curves, screen is held constant for the
Well, he just said the comparison was between a triode connected pentode
which would have triode type spectra, which is right afaik,
and the same tube pentode connected, but with a resistance NFB network.
The pentode with external loop FB with resistances would give different
spectra to plain triode connection.
The pentode spectra to begin with can vary enormously.
The RL can be chosen for almost zero 2H, and mainly all odd H
or it can be low RL for high 2H, some odds, or CCS, also high 2H but
but with 2H phase reversals. The screen dc voltage level affects
The high THD of a pentode with low RL or high RL are merely
reduced with external loop FB, and some high order introduced as well at
and in fact due to screen currents et all, pentode spectra with FB can
be quite varied depending
on local cathode current FB, and having a shunt NFB path as well to get
down to triode levels.
However, many pentodes have been used in circuits for mic amps and other
with just about all with NFB to reduce N&D, and keep gains where they
and great recordings were made, so arguing about triode or pentode sound
isn't all that productive.
I wish someone would breadboard something and measure it carefully,
and present a detailed post instead of publically theorizing,
but really not giving us any real idea of any real performances.
f I wasn't in the middle of a moving house that is just what I would do. As
soon as I have my need workshop set up i WILL.
I did a test on a 2N3904, and presented lots of cold hard facts,
so why can't one of you guys go test a 6AU6??????
But for datasheets that include it triode connected the curve looks
like a 'triode'.
Like, say, a 6GK6
Or a 6V6
Or, since he's complained about 'power' examples, a 6J7 RF pentode
Or a 6AU6
Or that the pentode equation substitutes Vs for Vp because Vs and Vg
are then the primary governing potentials generating the electrostatic
fields in a pentode.
as a pentode with feedback produces results that accurately model observed
behaviour then it is of value whether or not the assumed underlying
physical mechanism is correct. My attempts to compare the spectra of a
pentode with feedback to those of a triode demonstrate that this simple NFB
model is insufficient to describe what actually happens.
Because you're trying to make two physically different objects become
'identical' and that's just as fallacious as claiming a 6SL7 is not a
triode because it isn't 'identical' to a 6SN7.
To elaborate further, with a different example, 20dB of NFB around
amplifier A may not necessarily result in 'identical' spectral content
as 20dB of NFB around amplifier B. If it did then no one would care
about amplifier topologies, or the components used in them, or
anything else, as we'd simply wrap 20dB around them and, viola,
'identical' amplifiers. But it doesn't accomplish that even though
it's still 20dB of NFB in both cases, and the 'device' is a 'linear
amplifier' in both cases, with the same principles at work. And it
would be fallacious to argue one is NFB but the other is not, or that
one is a 'linear amplifier' but the other is not, because your bench
tests don't get 'identical' results from the two. Instead, you'd look
to see what's 'different' between the 'linear amplifiers', but not
argue one isn't a 'linear amplifier' or that one isn't using NFB.
Already done. You simply ignored it.
NFB around two different amps, or devices, does not always produce
'identical spectra' but no one argues it isn't NFB unless, for some
odd reason, the device is a triode.
Now, your bench tests might be useful to show, if its true, that one
can't, by simply connecting screen to plate, make a particular pentode
'identical' to a particular triode for any number of reasons and one
reason might be that connecting screen to plate makes a 'resultant
plate' that is different than having just one physical plate, or that
it's simply difficult to ascertain 'which' triode it would be 'like',
but that does not, in any way, show a 'contradiction' of it being the
same principles at work, as shown by the electrostatic equations, any
more than the amplifier example does. It might make a 'bad triode' or
a 'triode like no other' but it's still a 'triode' in the generic
sense of what a 'triode' is.
OK, so a trioded pentode cannot ever be exactly the same as a
triode without a screen. So what, they both behave like triodes, and
Well, I'm positing that as a possibility, not that I 'know' it for a
similar spectra and we don't need to go looking for exact likenessess.
Right, except I can also imagine there might be some spectral
differences if the previous posit is true.
I'd also note that the 'resultant plate' in a triode connected pentode
has a suppressor grid 'inside' it and that might cause a different
result than a single physical plate. I don't know. But the 'fact', to
the best available degree of scientific certainty, of the underlying
first principles electrostatic field equations remains.
Supressors are normally tied to the anode as well as the screen in
Oh good. When I asked that question a while back I didn't get a
But in beam tubes the beam forming plates which have the similar
function of the suppressor,
these are always tied to the cathode, so in effect you get beam triodes
when beam tetrodes are triode connected, and they make very decent
triodes when compared to
2A3, or 300B. In EL34, the suppressor has a pin 1 connection, but in
Exactly, whatever the truth is regard to the presence or absence of NFB
internal to triodes, Ian is using some very mysterious logic here, at
least it is mysterious to me.
There are many ways local NFB could be applied around a pentode,
just connect it as a cathode follower, use an OPT winding in the cathode
run a shunt NFB with two R, use an R divider to connect a floating
winding of an input transformer,
use a signal nulling circuit to apply only an error signal to the
or use ultralinear taps.
None will provide spectra that are the same as triode.
Methinks The goodness factor in the local external loop NFB is in the
ears of the listener.
We just had a guy hear say he just tried 3 different brand name EL34 in
his ST70 and heard no differences;
and this seems like heresy because some folks believe not all EL34 sound
Plenty of times have occurred where gathered listeners could not detect
between using a tube amp and solid state, despite a change in THD
spectra of at least 50 times,
and change of Rout of about 20 times.
On other occasions, its been chalk and cheese.
Not all sonic differences are immediately explainable.
Plenty of times I have shown that there isn't any sonic difference
between a UL amp
such as a Leak TL12 in triode with 12dB global NFB or in UL with 16dB
so that Rout and bandwidth remain very near constant.
Its nice to know all about such distortion issues, and to keep
but its the sound that matters at the end of the day.
Al in his Tube Garden just plugs and plays.......
Its nice to play with meters, but no need to have an obsession with
That's good to hear because I plan on triode connecting 10GK6s in the
amp I'm working the chassis for.
EL84 the suppressor is connected to the cathode.
Suppressors usually have low gm and µ and little effect within the tube.
Yes, agreed, but should anyone include in their imaginary abilities that
NFB shunt feedback comprising arms Z1 and Z2 that were slightly non
impedances, as indeed they really are in a real triode, then the unique
of the triode would be seen, as indeed it is.
Suppose you have a typical 6AU6 set up for open loop gain
of say 500, and you get say 5% THD at say 30Vrms output.
Then suppose you have a cathode follower output buffer diectly coupled
then have a
shunt FB network that can be adjusted to change the amount of NFB.
While keeping the voltage output at 30Vrms,
What is the spectra when open loop gain, OLG, has been reduced to 250?
What about at OLG = 100?, 50?, 10?
How does the pentode with external FB differ from simple triode
When you have done some real work in your real workshop, let us all know
what you have learnt.
I am busy doing just that Patrick, which is one reason why I have posted
less just lately.
Some don't though, like the 6DT6, used as a pentode and as a quadrature
for AF detection from the FM IF in TV sets. The gm of G3 was substantial
copmpared to G1.
If you want to know more, experiment with various ways of connecting G3,
and measure changes to THD spectra.
Then don't say it doesn;t.
Ther's no such thing as a 'perfect' CCS load and if there's no current
change there's NO VOLTAGE CHANGE because all the little electrons
making the field stay in the same place, which means the field hasn't
changed, and the voltage hasn't changed.
If there's no Ia change then there's no voltage change. It is simply a
No, it won't. It'll be 'almost' constant.
Try a book on basic electromagnetics. Triodes use the same electrons
as the rest of the universe. And if they don't move the field remains
I have checked every 'flipper' post I can find in this thread and not one of
them includes the governing equations.
No it's because I don't just believe everything I am told.
You must have had a hard time 'believing' Newton too.
Newton is OK as far as he goes - but he was not entirely accurate.
Actually, he was, within the realm of his observations. It's only when
extended beyond that scope that 'refinement' was needed.
Indeed, and that is precisely the opposite situation with valves. I suggest
you read 'Thermionic Valves - their theory and design' by A.H.W. Beck then
perhaps you will understand just how difficult it is to apply the theory to
From now one, every time I see a post from flipper, I will unavoidably
picture a kind of merman with a discretely placed flower.
But the fact still remains that if you want to dispute electromagnetic
theory, as is currently accepted and taught, then the burden is upon
you to 'discover' a 'new truth', as Einstein did. And even then it's
unlikely to alter the applicability to triodes because it's been
verified as far as that realm of observations go.
But, then, we're being rather silly anyway since everything you're
doing in 'electronics' is out the window if you don't 'believe' basic
Oh, I believe it, I just know that theory is harder to apply to real devices
than you might realise. Again I refer you to the book above.
Indeed you shouldn't believe all you are told, or all you read.
One takes time to form one's beliefs, and when your'e old, you become
full of beliefs,
and some are even still valid, but others may not be.
Didn't say "in this thread." I said posted. It's in the other one
you've participated in with "and negative feedback in triodes"
appended to the name of this one.
Particularly as those who tout the notion tend to argue that a tetrode
is merely a triode with the "internal feedback" removed. This leaves
comprehension of the tetrode obscured by fantasy.
Read the archives, there is a heck of a lot there.
I think its time I went to my workshop and did something useful.
I have some interesting ESL speakers to make and get working today...
pentode and especially the UL or CFB types, which can be
made stunningly clean. But that's an irrelevance. It isn't the
meter that is the arbiter here but your ear.
To cut to the chase, never mind all the other complications
Patrick has brought to the table (he'll soon get onto damping
factors at even greater length), there are already too many
complications here at a much more fundamental level because
you didn't take control of the NFB in your amp before you started
It is of the subtleties of the electronic interactions, so rudely
called complications above, which make tube sound endlessly
fascinating -- and sometimes so frustrating you want to scream.
To the moon and back using rocketry and gravity slingshot power is
simple (OK, not so simple) ballistics. The speeds do not begin to
approach relativistic levels. Jules Verne calculated it pretty
accurately-by-implication in his short novel, and the only thing wrong
with his "physics" is that had his projectile actually approached
escape-velocity + allowances for atmospheric friction, his passengers
would have been a thin smear on its base even with their water
cushion. That was well over 100 years ago, long before relativity was
even a gleam in anyone's eye. I believe he even had escape velocity
nailed as well. It has been years since I last read it, but there was
a table similar to this:
At the end of it.
Within the solar system it is a little more complicated and requires
in-flight corrections, but it is still just ballistics. Get to a
significant percentage of light-speed and you will finally require a
higher level of physics... maybe. Much of it depends on where you are
and where you are going and how close you are to large gravity
anomalies such as black holes.
I canot recall where I read that relativistic effects begin to
have an effect even with moonshots and very slow moving Saturn V
But I an't noexpert on this.
What do higher authority say, perhaps NASA, and I don't think Jules
was up to date...
As it happens, the software that NASA uses for its long-term space-
shots (including the latest-in-the-news "New Horizons" spacecraft) was
developed 'right here' in Exton, Pennsylvania by Analytical Graphics,
Inc. It uses Newtonian Physics, not relativistic physics as the speeds
(even at 83,200kph) are just too slow for those effects to manifest.
The Company describes their system as "celestial pinball".
You may see animations of their work showing the New Horizons mission
Verne did get it right as to speed, direction, location and pure
ballistics. What he missed are the effects of near-instant
acceleration, the full effect of atmospheric friction and similar
details. He had no way of knowing of the radiation belts and other
niceties. he also did not consider the need for mid-course corrections
and so forth. But not bad at all for well over 100 years ago, and the
science he used is still all that is required with the addition of a
few more variables.
The ability for listeners to subjectively discern NFB artifacts is not
clear after 74 years of arguments,
so tonight's discussions will not provide any certainty.
But I would say its up to diyers, audiophiles, tinkerers to decide what
with regard to NFB, and maybe they either agree with Andre, or myself,
or with neither of us,
or have an idea that is elsewhere.
I must raise the issue of damping factor as A has predicted I would, and
ask the group what
is best, a pentode amp with Rout = 40 ohms, as well as the THD = 0.5%
and IMD at 5% at 2 watts,
or is the music to sound better, less tiring, etc, with Rout = 1ohm, and
THD = 0.05% and IMD = 0.5% at 2 watts?
In about 1999, I demonstrated a 5050 UL PP stereo amp to the Audiophile
Society in NSW.
This had 2 x GE6550A tubes per channel and 6CG7 drivers. About 16dB of
GNFB was used.
On the same day, another guy demonstrated his Forced Symmetry circuit
with a quad of 6550 per channel
but in triode, and a claimed total of 6dB NFB, enough to halve the raw
but do little to quell any THD & IMD.
The 30 club members present were not overly impressed with the triode
amp and in every
way the UL amp with 1/2 the bottles was said to sound smoother at the
Arguments continued between the no NFB camp and the pro camps for some
time, and later when I
had another demo with SEUL amps with 13Ei tubes again with about 16dB
During the coffee break I overheard a guy say to his mate, "well, this
proves you CAN use NFB with an SE amp."
It is a fashionable idea that SEamps must not use any NFB. Bollocks. It
ain't necessarily so.
However, please remember this is only my opinion, and any here in the
group are free to find out for
Everyone liked those SEUL amps; there was sea of contended smiles.
They said the midrange was quite especially outstanding, but in fact it
was all true full range sound,
free of the artifacts of an ill concieved SE amp without enough power
poor technical performance with high Rout.
There were 30 ppl present, the venue is the upstairs section of an old
and much louder levels than anyone would use at home were used.
All quite OK with only 25 watts/channel.
The 13Ei is a perfectly HORRID non linear tube used as a pure bean
I applied 66% of the anode signal from a tap on the OPT to wrench the
towards triode op but with a higher power output nearing beam output
Rout was still too high.
GNFB did the rest of the job to reduce Rout to around 0.6 ohms at that
UL taps are NFB from anode to the screen, and triode is just maximum
unless you had an "extension of the anode winding" to derive a signal
than the anode signal which is then fed back to the screen.
I see nothing to recommend this idea, because of further power output
reductions lower than triode.
The triode connection is a form of NFB.
There IS NFB within a triode even without a screen.
Time and time again ppl tell me the dynamics and pizzazz are bettered
I just refuse to use too much NFB. I make sure the open loop bandwidth
of the OPT is at least equal or greater than the closed loop BW.
Devices have an adequate amount of class A power.
If your amp has more NFB than mine, and measures slightly better, I am
terribly interested, because mine sound well enough.
What I have seen occur often is customer preference for a tube amp
with say a reasonable 0.05% THD at 2 watts maybe 0.2% at 40 watts,
with NFB over a solid state amp with 0.002% at 100 watts, and such low
2 watts its almost un-measurable.
To my mind the tube amp artifacts in a FB amp don't matter at all.
Please don't believe me.
Find out with your own experiments.
"wonderfully well written and reasoned information
for the tube audio constructor"
John Broskie TubeCAD & GlassWare
"an unbelievably comprehensive web site
containing vital gems of wisdom"
Stuart Perry Hi-Fi News & Record Review