[Eeyore]

That video on YouTube got me thinking about them and why they might have
different characteristics to indirected heated triodes.

Clearly the heater/cathode has a voltage gradient across it. Turn the idea
'upside down' and it's equivalent to a grid with a non-equipotential effect over
its length. The most similar effect I can think of would be a variable mu valve
where the grid spacing varies.

Anyway, it occurs to me that a DHT will have a characterisitic similar to the
sum ( or composite ) of several triodes with displaced grid potentials. This may
possibly make it more linear. I haven't looked into it that far yet but I'm sure
it could be readily modelled.

Graham

[Patrick Turner]

Eeyore wrote:

That video on YouTube got me thinking about them and why they might have
different characteristics to indirected heated triodes.

Clearly the heater/cathode has a voltage gradient across it. Turn the idea
'upside down' and it's equivalent to a grid with a non-equipotential effect over
its length. The most similar effect I can think of would be a variable mu valve
where the grid spacing varies.

Anyway, it occurs to me that a DHT will have a characterisitic similar to the
sum ( or composite ) of several triodes with displaced grid potentials. This may
possibly make it more linear. I haven't looked into it that far yet but I'm sure
it could be readily modelled.

Graham

You are thinking that the DHT has a substantial voltage end to end
of the cathode with respect to grid bias, ie, Eg-k.
but many DHT have ac for heating the cathode, so what is the situation
then?
With a carefully balanced application of the mains F heating, the DHT is
supposed to remain hum free.

And what of where the DHT cathode is a zig-zag pattern so that the
electron space charge between the g and k remains about constant?

Anyway, the issues of cathode gradient have about all been addessed by
makers
during the last 100 years, some of their conclusions have of course been
forgotten
in the lurch toward multi element solid state gear.
I doubt you will find that just because a triode
is a DHT it will be any more linear than a IHT such as 6SL7, 12AX7, or
12AY7.

I suggest you test some samples of DHT and IHT using the same
linearity checking gear for all samples. The triode data curves given
for
many old triodes are often rather incorrect and do not agree with real
world
measurements and the distortion calculated from load line analysis
is often a percent or two different to what one may measure, and in fact
the measurements
often give lower THD than fooling around with a ruler and a set of
curves.
This simply means that many of the curves given for the tube data that
exists in the old
books is AN APPROXIMATE INDICATION ONLY of about what to expect from
a given tube type.
It also all means that the gear used for gaining enough dots on a page
to make Ra curves
may in fact have not been very linear and before the data was published
it was all tidied up
by draftsmen/women with a steady hand drawing curves.
Examining many data sheets shows that some tube data have implausible
looking curves
with inexplicable differences in curve shapes where none would actually
ever occur.

So the real world test of various types of triodes using the same gear
is the only
way to really find out anything about general comparisons of linearity
for the DHT and IHT which followed.

My bet is that the DHT are not any more linear than IHT with similar
basic Ra, Gm, and µ for the same Ia.

Some will argue DHT just sound better.

They are welcome to say that, but have any AB tests been done to support
the claims?
Or is it just nostalgic worshipping of a bygone age when life was
simpler, purer, better...
( unless one needed a good dentist or a decent pay packet. )

I have never heard a bad sounding triode amp unless too much was asked
of it.

Patrick Turner.

[Eeyore]

Patrick Turner wrote:

Eeyore wrote:

That video on YouTube got me thinking about them and why they might have
different characteristics to indirected heated triodes.

Clearly the heater/cathode has a voltage gradient across it. Turn the idea
'upside down' and it's equivalent to a grid with a non-equipotential effect over
its length. The most similar effect I can think of would be a variable mu valve
where the grid spacing varies.

Anyway, it occurs to me that a DHT will have a characterisitic similar to the
sum ( or composite ) of several triodes with displaced grid potentials. This may
possibly make it more linear. I haven't looked into it that far yet but I'm sure
it could be readily modelled.

Graham

You are thinking that the DHT has a substantial voltage end to end
of the cathode with respect to grid bias, ie, Eg-k.

Correct.


but many DHT have ac for heating the cathode, so what is the situation
then?

Effectively it does something similar but the effect has a line frequency modulation
involved.


With a carefully balanced application of the mains F heating, the DHT is
supposed to remain hum free.

I follow. Is there some objection to DC heating ?


And what of where the DHT cathode is a zig-zag pattern so that the
electron space charge between the g and k remains about constant?

You'd have to involve some serious modelling to establish the precise effect but once
again it's likely to be a composite.


Anyway, the issues of cathode gradient have about all been addessed by
makers during the last 100 years, some of their conclusions have of course been
forgotten in the lurch toward multi element solid state gear.

I dare say the considerations we're discussing weren't exactly their primary concerns
back then.

Graham

[Eeyore]

Patrick Turner wrote:

Some will argue DHT just sound better.

On what basis ? Quasi-Religious belief ?

I am at least interested in understanding the underlying physical principles and these
will clearly have the answer.

Graham

[[email protected]]

Eeyore wrote:
With a carefully balanced application of the mains F heating, the DHT is
supposed to remain hum free.

I follow. Is there some objection to DC heating ?

None at all, if you live on a farm without electricity and only have
batteries to play your radio.

In fact, by clever arrangement of series DC filaments in directly
heated tubes you can get the grid bias for one or more tubes "for
free".

I dare say the considerations we're discussing weren't exactly their primary concerns
back then.

Well, they address all your concerns, making them rather trivial. See
for example RDH4, pages 6, 80, and 560. Your OP was entirely correct in
that it all is supposed to average out in the end with AC DHT in the
approximations used in class A.

They certainly did have more primary concerns, mostly not very relevant
today, in the 50's and earlier texts where they talk about A battery
lifetime and filament performance and emission as the A battery ran
down, and some really clever tricks for doing grid bias and AVC with
directly heated DC triodes back when they were common on battery
radios.

Tim.

[Arny Krueger]

"Eeyore" wrote in
message

That video on YouTube got me thinking about them and why
they might have different characteristics to indirected
heated triodes.

For one thing, they are simply different tubes. I see no literature that
suggests that a #ZZ tube is a DHT version of the 6C4, or something like it.

Clearly the heater/cathode has a voltage gradient across it.

Not clear at all. There are many ways to make DHTs, and this is most
aparrent when you start looking at transmitting tubes, where the parts are
generally large enough that you can see them more clearly. The filament can
be a strand that zigs from top to bottom several times. It can be concentric
coils. It can be lots of things.

Turn the idea 'upside down' and it's equivalent to a
grid with a non-equipotential effect over its length.

Only in the simpler cases.

The most similar effect I can think of would be a variable mu
valve where the grid spacing varies.

IOW, a tube that is designed to be more nonlinear, but with extended
cut-offs.

[Patrick Turner]

Eeyore wrote:

Patrick Turner wrote:

Eeyore wrote:

That video on YouTube got me thinking about them and why they might have
different characteristics to indirected heated triodes.

Clearly the heater/cathode has a voltage gradient across it. Turn the idea
'upside down' and it's equivalent to a grid with a non-equipotential effect over
its length. The most similar effect I can think of would be a variable mu valve
where the grid spacing varies.

Anyway, it occurs to me that a DHT will have a characterisitic similar to the
sum ( or composite ) of several triodes with displaced grid potentials. This may
possibly make it more linear. I haven't looked into it that far yet but I'm sure
it could be readily modelled.

Graham

You are thinking that the DHT has a substantial voltage end to end
of the cathode with respect to grid bias, ie, Eg-k.

Correct.

but many DHT have ac for heating the cathode, so what is the situation
then?

Effectively it does something similar but the effect has a line frequency modulation
involved.

Yes, but the effect is minimised with a balanced supply of AC to heat
the cathode.

With a carefully balanced application of the mains F heating, the DHT is
supposed to remain hum free.

I follow. Is there some objection to DC heating ?

In a 300B, the typical grid bias is -70V.
If 5Vdc is applied to the cathode to heat it the
dc voltage across the cathode has a negligible effect giving rise to
more emission at the more positive end of the cathode.

So dc applied is OK.

And what of where the DHT cathode is a zig-zag pattern so that the
electron space charge between the g and k remains about constant?

You'd have to involve some serious modelling to establish the precise effect but once
again it's likely to be a composite.

Anyway, the issues of cathode gradient have about all been addessed by
makers during the last 100 years, some of their conclusions have of course been
forgotten in the lurch toward multi element solid state gear.

I dare say the considerations we're discussing weren't exactly their primary concerns
back then.

Graham

W.E. just wanted simple easy operation for their theatre amps which
were leased out, not sold, so the tubes had to be reliable.
And the circuitry was kept simple because the 300B didn't need loop FB.
And most DHT were operated with ac applied from a floating winding with
CT
because making dc supplies for low voltage at high current was difficult
and was avoided
in 1930.

Although most theatres had sensitive horn speakers, a pair of 300B were
adequate.
I don't recall folks complaining about hum...

Patrick Turner.

[Patrick Turner]

Eeyore wrote:

Patrick Turner wrote:

Some will argue DHT just sound better.

On what basis ? Quasi-Religious belief ?

Most audiophiles I know do not follow religion other than be cathode
followers.
Most agree with ppl like Richard Dawkins, and are never seen in a church
on sunday.
The are seen occasionally soldering things on sunday, and winding coils
for crossovers.

I am at least interested in understanding the underlying physical principles and these
will clearly have the answer.

Well sometimes what we measure does not explain the sound we hear.

Its pointless being obsessive about THD for example, especially
when its below 0.05% and all mainly 2H.
But an arrangement of source, preamp, power amp and speakers
will usually sound different to a different set of componenents.
Especially in a different room as well.

But the numbers are important though, and I like amps and all the other
bits to make
low amounts of THD and IMD, and its the IMD that so easily does the
damage.

I don't recall hearing too many tube amps that were much improved by
lowering the THD from
0.1% to 0.01%, let alone 0.001%, a difficult feat for a tube power amp.
And an SS amp won't sound better just because it can get down to 0.001%
THD.
But nor does it suffer from being constructed so that 0.001% is typical
at listening levels;
the absense of an OPT allows more FB so it can measure better. Big deal.
Some people say they
are cold and clinical and lifeless, and one needs triodes to present the
real thing that was heard in the studio or concert hall recording
session.
I don't argue, I just build the amps for them; I try to get the numbers
right while maintaining a semblance of
simplicity. After awhile one finds out what seems to really work the
best, but
many varieties of tube circuits work well.

Patrick Turner.


Graham

[Andre Jute]

Eeyore wrote:
That video on YouTube got me thinking

LOL.

about them and why they might have
different characteristics to indirected heated triodes.

They don't, unless you want to admit that some tubes sound different
from others in the same circuit. Do you?

Clearly the heater/cathode has a voltage gradient across it. Turn the idea
'upside down' and it's equivalent to a grid with a non-equipotential effect over
its length. The most similar effect I can think of would be a variable mu valve
where the grid spacing varies.

There's something fallacious about this but let's see where it leads
you:

Anyway, it occurs to me that a DHT will have a characterisitic similar to the
sum ( or composite ) of several triodes with displaced grid potentials. This may
possibly make it more linear.

Oh dear. You don't have the brains to live in ivory tower, Poopie.
Climb down off it, inspect the sand at the base. That's the stuff
amplification devices are made of, tubes and transistors both. In the
particular case of DHTs, no assistance from you is required to make
them more linear. Sit down now before you go to my netsite because,
given the season, we don't want even you to fall down dead of shock.
Now hold onto your chair while you study the linearity of a pair of my
Western Electric 300, the actual transfer curves taken off actual
tubes, not generic averages.
http://members.lycos.co.uk/fiultra/WE300BindivSPEC.jpg
See those long, long, long, ruler-straight lines? That's linearity.
(And without any bandaid of negative feedback applied, of course.) DHTs
need no help from you. But, since you will no doubt insist, there is
more information for you to work with at
http://members.lycos.co.uk/fiultra/KISS%20190.htm
and in particular pay attention to
http://members.lycos.co.uk/fiultra/112KISScurves.jpg
for the principle of determining a linear operating range that does not
need NFB and to
http://members.lycos.co.uk/fiultra/1...tingpoints.jpg
for a specific example. Standard generic curves are also available from
the same illustrations page cited above (I thought I'd give you the
piccies first, see how kind I am, and not frighten you off with actual
text and math you might balk at). The meaning of it the universe,
music, linearity, glee and DHTs is explained a level up, starting at
http://members.lycos.co.uk/fiultra/JUTE%20ON%20AMPS.htm
and continuing in particular from the KISS index page he
http://members.lycos.co.uk/fiultra/T...mp%20INDEX.htm

I haven't looked into it that far yet but I'm sure
it could be readily modelled.

It's clear you haven't looked into it at all. You're merely rambling
ignorantly. But don't let us discourage you from modelling DHTs. You
will learn lots. It will keep you off the streets.

Graham

However, you would be far better occupied modelling and trying to
improve devices that really need your help, anybody's help, because
they are so wretchedly illinear: transistors.

I understand: you're feeling unwanted. Now you're trying to say
something relevant. But you don't have the brains or the background.
Try hitting the RDH sixteen hours a day from now until next March and
you might, when you have digested it all, say something that isn't dead
wrong. At least you will be prepared to understand in a superficial way
the references I have given you above. On the same site you will find
http://members.lycos.co.uk/fiultra/J...MPS%20RDH.html
which is my famous guide to what you really need to understand in the
RDH before you can say anything relevant to tubies. (See, I'm much more
complaisant than Patrick, who wants you to learn the whole thing off by
heart before we let you into the club.)

Andre Jute
The trouble with most people is not what they don't know, but what they
know for certain that isn't true. ---Mark Twain

[Arny Krueger]

"Patrick Turner" wrote in message


Well sometimes what we measure does not explain the sound
we hear.

Expand on that. Isolated measurements mean nothing. I measure an object and
it is 1/3 the length of my ruler. I say, "It is 33%".

My point is that no isolated measurement has any meaning without context.
Doesn't matter what it is. Your comment is therefore completely trivial and
meaningless.

Its pointless being obsessive about THD for example,
especially when its below 0.05% and all mainly 2H.

Now you contradicted what you just said. You've in effect said that 0.05% 2H
is less than that which is audibly significant. Of course, you added
context. You've proven my point, which is that what we measure can explain
what we hear if we add appropriate context.

But an arrangement of source, preamp, power amp and
speakers
will usually sound different to a different set of
componenents. Especially in a different room as well.

So what?

But the numbers are important though, and I like amps and
all the other bits to make
low amounts of THD and IMD, and its the IMD that so
easily does the damage.

Missing context: IMD and THD are both the results of nonlinearity. It is
difficult to contrive things so that one does not come without the other.

I don't recall hearing too many tube amps that were much
improved by lowering the THD from
0.1% to 0.01%, let alone 0.001%, a difficult feat for a
tube power amp.

If you study the literature and do some experiments, you will find that some
place around 0.1% nonlinear distortion, further improvements in linearity
tend to be moot, no matter what sort of listening test you try to do.

And an SS amp won't sound better just
because it can get down to 0.001% THD.

Agreed.

But nor does it suffer from being constructed so that
0.001% is typical at listening levels;

Not necessarily.

the absense of an OPT allows more FB so it can measure
better. Big deal.

Getting rid of the OPT has other benefits. Heck, even tubed amps can benefit
from getting rid of the OPT.

Some people say they
are cold and clinical and lifeless, and one needs triodes
to present the real thing that was heard in the studio or
concert hall recording session.

Some people say the darndest things, especially if they think it will buy
them some influence over others.

I don't argue, I just build the amps for them; I try to
get the numbers right while maintaining a semblance of
simplicity. After awhile one finds out what seems to
really work the best, but
many varieties of tube circuits work well.

It is usually a mistake to overdefine the solution. Asking that an amp does
not audibly change its input signal under any reasonble operating condition
should suffice.

[Peter Wieck]

Andrew Jute McCoy babbled:

A bunch of crap proving that it did not read the actual question as posed.... Nor does it understand the first thing about tubes, but merely parrots what it reads elsewhere (and sometimes with hilarious consequences based on its ignorance).

The closest thing you got to 'getting it' was when you suggested that
different tubes will sound differently in the same circuit. No Sh*t.
And two of the 'same' tubes may also sound different based on their
peculiar condition at the moment of use. Hence the vast amounts of
mythology built around 'matching' and so forth. Patrick Turner hit it
dead-on when he discussed the issues based on AC or DC on the filament,
and how greater experience and better technology does not necessarily
mean better performance in actual use.

Tubes are sloppy devices, with parameters wide enough for the QE2 to
pass through with great ease relative to solid-state devices. However,
that allows relatively simple circuits with due allowances for that
sloppiness to perform splendidly, something that SS devices have not
achieved to-date and are unlikely ever to do so (based on simplicity,
not actual performance as-designed). A tube power amp of say.... 35
watts in PP can get away with perhaps 20 electrical parts in the
circuit from the power transformer through the output transformer. I
doubt a SS amp could do that with less than 50 parts, or more. And if a
triode tube is seen as the functional equivalent of a single
transistor, the contrast in complexity becomes even greater (ignore
size).

Not everyone here is gifted with a mind as rigidly focused and
all-knowing as yours. Permit these people to be curious as to them
there are still things in this world to be discovered. To something
such as yourself which has absorbed all worthwhile knowledge such as
there is none more worthwhile to learn, this state of incompleteness
may be annoying and fit only for ridicule. But for some it is a way of
learning.

You remind me of that typical bombastic psuedo-artist/kitchen table
engineer type: One that knows more and more about less and less until
it knows everything about nothing.

Peter Wieck
Wyncote, PA

[Doug]

Peter Wieck wrote:
Andrew Jute McCoy babbled:

A bunch of crap proving that it did not read the actual question as posed.... Nor does it understand the first thing about tubes, but merely parrots what it reads elsewhere (and sometimes with hilarious consequences based on its ignorance).

The closest thing you got to 'getting it' was when you suggested that
different tubes will sound differently in the same circuit. No Sh*t.
And two of the 'same' tubes may also sound different based on their
peculiar condition at the moment of use. Hence the vast amounts of
mythology built around 'matching' and so forth. Patrick Turner hit it
dead-on when he discussed the issues based on AC or DC on the filament,
and how greater experience and better technology does not necessarily
mean better performance in actual use.

Tubes are sloppy devices, with parameters wide enough for the QE2 to
pass through with great ease relative to solid-state devices. However,
that allows relatively simple circuits with due allowances for that
sloppiness to perform splendidly, something that SS devices have not
achieved to-date and are unlikely ever to do so (based on simplicity,
not actual performance as-designed). A tube power amp of say.... 35
watts in PP can get away with perhaps 20 electrical parts in the
circuit from the power transformer through the output transformer. I
doubt a SS amp could do that with less than 50 parts, or more. And if a
triode tube is seen as the functional equivalent of a single
transistor, the contrast in complexity becomes even greater (ignore
size).

Not everyone here is gifted with a mind as rigidly focused and
all-knowing as yours. Permit these people to be curious as to them
there are still things in this world to be discovered. To something
such as yourself which has absorbed all worthwhile knowledge such as
there is none more worthwhile to learn, this state of incompleteness
may be annoying and fit only for ridicule. But for some it is a way of
learning.

You remind me of that typical bombastic psuedo-artist/kitchen table
engineer type: One that knows more and more about less and less until
it knows everything about nothing.

Peter Wieck
Wyncote, PA


Peter:

Although tubes manufactured under a given part number do exhibit some
spread in their parameters (gm, rp, u etc.), they exhibit nowhere near
the parameter spead inherent in bipolar transistors or FETs. Just take
a look in any transistor data book to observe the spreads, for example,
in hfe for bipolar devices or in Vp and Idss for FETs. Great care must
be taken in fact in the design of solid state circuits to accomodate
these variations so that all final product in a production run will
meet its specifications.

Among the measures that are routinely taken in solid state designs to
accomodate these variations are the application of considerable
feedback, be it local, global or both. Bipolar devices in fact
exhibit such a non-linear transfer characterisitic that they are
unviable for audio amplification without these measures being taken.

On the other hand, if you were to take as an example a sample of US
built 12AX7s and examine their characteristic curves on a curve tracer
you would be very surprised at how closely their curves match. I do
not know how well their curves would match against Russian or Chinese
product, or against that of Telefunken, Amperex, Mullard etc.

It is certainly possible to build a solid state amplifier with a
similar component count to a tube counterpart. Many bipolar designs
from the 70s were very simple. These designs however relied heavily on
gobs of global negative feedback to linearize them, and in my humble
opinion the sound suffered as a consequence.

Regards: Doug Bannard

[robert casey]

Eeyore wrote:


Clearly the heater/cathode has a voltage gradient across it. Turn the idea
'upside down' and it's equivalent to a grid with a non-equipotential effect over
its length. The most similar effect I can think of would be a variable mu valve
where the grid spacing varies.


Some DHT tubes specify which end of the filament is to get the positive
end of the filament supply. 7 pin miniature "Battery Radio" tubes were
like this, as well as some sub-mini types. Idea being that the grid was
built to have the same global characteristic, but differing
characteristic from one end of the filament to the other. This could
easily be done if the filament was a single strand strung from the top
to the bottom of the tube.

If AC is used on a DHT, does the effect of the uneven voltage across the
filament exactly cancel out over the course of an AC waveform cycle?
Maybe there's a V^2 term in the physics. That would cause distortion
products that would cycle up and down at 120Hz (100Hz in Europe and such
places).

[Peter Wieck]

Doug wrote:

Although tubes manufactured under a given part number do exhibit some
spread in their parameters (gm, rp, u etc.), they exhibit nowhere near
the parameter spead inherent in bipolar transistors or FETs.

Sure. But bipolar transistors and FETs are not hardly the functional
equivalent of a standard transistor as stated in the premise. Pick one
of the most common output transistors of the early SS era the
2N3055.... That is about the rough (power) equivalent of any-of-several
output tubes of similar vintage and may crudely be compared to a triode
as if it is used in a low-power/low-distortion application; it will
give about as much as a typical triode-connected 6CA7/EL34 or 807 of
the period. As I am unfamiliar with the 300B *personally* excepting
observing them in Other People's equipment, I will not write to that
application other than to pick it as a "typical high performance
triode". These days, the 3055 may be replaced with any number of other
similar transistors from the 3771-2-3 and on up. But the difference
between any two 3055s off the line would be far less than any given two
OEM 300Bs. What is made with that number these days is a different
beast entirely, as they are by any measure a boutique tube and not even
a little bit a production tube as they were 60+ years ago.

So, you state clearly that simple SS circuits can be achieved with
similar part-counts. But that they suffer as a consequence. We agree
absolutely on that point. But the fundamental differences between a
given tube (let's stick with triodes for these purposes) and any given
transistor that is (very)roughly equivalent to a triode are pretty
large.

Tubes will "test" 'as new' if above a certain set of parameters. Over
time, they will continue to function even well below these parameters
but the degredation will be reasonably predictable and take place over
time and based on use. Transistors will operate "as new" until they
fail. Then they will fail all-at-once. They may *never* fail within a
reasonable time measurement, tubes will *always* fail.

Again, generally, transistors display a cliff-effect at forward bias.
Tubes generally do not.

There are many more similarly obvious differences.

Again, tubes have parameters that as compared to *similarly
functioning* transistors are wide enough for the QE2 to pass through at
speed. There are exceptions to every generalization of course.

Curve for the 2N3055 and its complement may be found at:

http://www.kbt-dc-supplies.com/image...5DataSheet.pdf

Curves for the WE300B may be found at:

http://www.westernelectric.com/spec_sheets/300B.pdf

Peter Wieck
Wyncote, PA

[Chris Hornbeck]

On Wed, 20 Dec 2006 09:58:42 -0500, "Arny Krueger"
wrote:

There are many ways to make DHTs, and this is most
apparent when you start looking at transmitting tubes, where the parts are
generally large enough that you can see them more clearly.

I believe that this is the largest effect. Large physical
size alone gives an easier path to linearity. High voltages
give an easier path to linearity. High voltages "cause"
cathode poisoning. Cathode poisoning is less of an issue
with white-hot "cathodes" than with oxide-coated cathodes.

IOW, I believe that the linearity "magic" of the big triodes
is mostly a historical artifact. At the time that they were
designed linearity was considered important. Oxide-coated
cathodes were something WE could do (at 500 volts, anyway),
but were challenging even for them.

A few decades later, oxide-coated cathodes became the norm
for audio power tubes, but linearity wasn't considered
important. Operation at 500 volts or slightly less became
an important design consideration.

OTOH, the directness of direct-heating of filamentary triodes
is an unmitigated pain in the ass to folks building amplifiers
with them. Hum can be dealt with many ways; search Google
in this newsgroup for things like ultra-sonic heating, etc.
The difficulties of distributing loading between anode and
"cathode" in advanced designs is a real bitch, approaching
insurmountable.


But some folks *still* consider their virtues to outweigh
all of the disadvantages. That's actually my position, that
intrinsic linearity matters *that* much.

All good fortune,

Chris Hornbeck
"Too soon oldt; too late schmardt."

[Chris Hornbeck]

On 20 Dec 2006 17:51:32 -0800, "Peter Wieck" wrote:

Again, tubes have parameters that as compared to *similarly
functioning* transistors are wide enough for the QE2 to pass through at
speed. There are exceptions to every generalization of course.

No idea what you mean here, but taking the most obvious
reading, this is very incorrect.

Surely you're not saying that tolerances for semiconductors
are lower than tolerances for electron valves? That would
be wacky.

Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

[Andre Jute]

Chris Hornbeck wrote:

But some folks *still* consider their virtues to outweigh
all of the disadvantages. That's actually my position, that
intrinsic linearity matters *that* much.

Silence is everything. I well remember being abused for recklessly
using up my tubes when I admitted to running 300B at 32W (80mA and 480V
delivered to the plate for an operating point of 400V with cathode
bias); back then the average was 55-60mA and around 300V, with daring
experimenters running 300B all the way up to 65mA. Those guys had
gotten into a mindset, perhaps from going blind working with little
preamp tubes, perhaps from not putting their minds in gear, of thinking
that just above the curvature on the Ep-Ia-Eg transfer, and sometimes
in the curvature at the extensions, was good enough, The slightest
rectification effect when signal was applied would screw their sound
over the entire transfer width. But some of them wouldn't or couldn't
see that the way to *guarantee* stunning linearity even from famously
linear tubes like 300B and the 845 and 211 (and later Svet's late and
much lamented late SV-572-3 and -10) was to run them at high voltage
and high current into flat (i.e. high primary impedance) loads, thus
sacrificing the extra power of the high current and high voltage for
exceptional linearity. When you can get linearity like that, you can
afford to sneer at feedback as the solution of cheapskates and people
who haven't thought about what it is they want from their sound or how
much they are willing to pay for it.

Interestingly, the cost turned out to be not that much at all. DHT are
generally sturdier than most people give them credit for. I ran one set
of 300B for 14000 hours at 80mA, using them to break in Lowther
drivers. Your average audiophile, by definition a paranoid, would have
disposed of them after a couple of thousand hours on the mere suspicion
that they were no longer giving their finest.

One final, important, aspect of linearity that you didn't touch on.
Beyond a not very low level, it is not the amount of noise that is
important but its composition. An agreeable composition of noise falls
naturally with a Class A1 amp, which is of course the class of
operation most DHT fall into naturally, almost axiomatically, when
applied for hi-fi. Patrick was saying that he never heard a triode amp
with an ugly sound; I would go a smidgen further and say that it is
awfully difficult to build a really bad amp with triodes and especially
with DHT, simply because they have so much natural goodness going for
them (1).

Chris Hornbeck
"Too soon oldt; too late schmardt."

Looks like the Amish version of the NY Jewish saying: The wisdom of age
is liking giving a bald man a comb.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"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


(1) Of course I phrased that last clause to give undesirables apoplexy.

[Andre Jute]

Chris Hornbeck wrote:
On 20 Dec 2006 17:51:32 -0800, "Peter Wieck" wrote:

Again, tubes have parameters that as compared to *similarly
functioning* transistors are wide enough for the QE2 to pass through at
speed. There are exceptions to every generalization of course.

No idea what you mean here, but taking the most obvious
reading, this is very incorrect.

Surely you're not saying that tolerances for semiconductors
are lower than tolerances for electron valves? That would
be wacky.

Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

I noticed Doug trying to educate Worthless, with presumably little
effect if you now have to take up the standard. But you're both wasting
your breath. It seems to me extremely doubtful that even Worthless can
be this ignorant. Check out the sequence of posts in this subthread
which I started. It seems clear Worthless went over the top into
fantasy-land and outright untruths in his constant attempts to top me.
Making a life of commenting on what I say and do was long since proved
a wasting disease for too many others; anyone's life must be very dull
who thinks that mine is exciting enough to dedicate so much time to as
Worthless Wiecky does.

Still, I love men tilting at windmills; it is all the proof necessary
that there is hope for humanity. And while I'm on about Sancho Panza's
master, a super Spanish writer I discovered: Carlos Ruiz Zafon, acute
on the O; his The Shadow of the Wind is a must-read, a picaresque,
several love stories, several thrillers, a lamentation for the state of
literature, a coming-of-age novel, a history of Franco's Spain, an
evocation of Barcelona (my second-favorite Spanish city) in the 1950s,
and much, much besides, all on an unputdownable rollercoaster.
Excellent English translation.

Andre Jute
The trouble with most people is not what they don't know, but what they
know for certain that isn't true. ---Mark Twain

[Chris Hornbeck]

On 20 Dec 2006 19:57:46 -0800, "Andre Jute" wrote:

One final, important, aspect of linearity that you didn't touch on.
Beyond a not very low level, it is not the amount of noise that is
important but its composition. An agreeable composition of noise falls
naturally with a Class A1 amp, which is of course the class of
operation most DHT fall into naturally, almost axiomatically, when
applied for hi-fi. Patrick was saying that he never heard a triode amp
with an ugly sound; I would go a smidgen further and say that it is
awfully difficult to build a really bad amp with triodes and especially
with DHT, simply because they have so much natural goodness going for
them (1).

And by saying "noise" above, let me interject that you mean
all artifacts, and that I can agree completely. I have
reservations about using the word more inclusively, but I
do understand the origins, so can't really quibble.

There is some extra "there" there, folks. Gotta follow along
in our hymnals, but it's worth the journey.

"Too soon oldt; too late schmardt."

Looks like the Amish version of the NY Jewish saying: The wisdom of age
is liking giving a bald man a comb.

Yeah, that's even better. Mine is Pennsylvania Dutch (Deutsch),
as remembered from a million years ago.

Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

[Chris Hornbeck]

On 20 Dec 2006 20:32:32 -0800, "Andre Jute" wrote:

snipped

What's with you guys? It's OUR newsgroup. Lighten up already.
Sheesh.

Still, I love men tilting at windmills; it is all the proof necessary
that there is hope for humanity. And while I'm on about Sancho Panza's
master, a super Spanish writer I discovered: Carlos Ruiz Zafon, acute
on the O; his The Shadow of the Wind is a must-read, a picaresque,
several love stories, several thrillers, a lamentation for the state of
literature, a coming-of-age novel, a history of Franco's Spain, an
evocation of Barcelona (my second-favorite Spanish city) in the 1950s,
and much, much besides, all on an unputdownable rollercoaster.
Excellent English translation.

Cool, I'll check it out. A high point of my sorry life was
performing in a _Man of LaMancha_. Besides a bit dragging an
incredibly beautiful woman offstage by her hair (stage work,
of course!) I had some very loud declamatory Guard lines in
one performance when the lead (the incredibly beautiful woman's
husband, actually) couldn't make call.

Ah, fame...


Much thanks, as always,

Chris Hornbeck
"Too soon oldt; too late schmardt."

[Eeyore]

Doug wrote:

Although tubes manufactured under a given part number do exhibit some
spread in their parameters (gm, rp, u etc.), they exhibit nowhere near
the parameter spead inherent in bipolar transistors or FETs. Just take
a look in any transistor data book to observe the spreads, for example,
in hfe for bipolar devices or in Vp and Idss for FETs. Great care must
be taken in fact in the design of solid state circuits to accomodate
these variations so that all final product in a production run will
meet its specifications.

With respect to bipolar transistors that spead in hfe hardly presents any great problems and certainly doesn't require 'great care' etc...... You simply design to accomadate the lowest hfe you're likely to meet and in most cases everythings fine.
Designing hfe senstive circuitry is to be deprecated.

In all other respects the parameters are highly consistent.

Graham

[Eeyore]

Doug wrote:

Many bipolar designs
from the 70s were very simple. These designs however relied heavily on
gobs of global negative feedback to linearize them, and in my humble
opinion the sound suffered as a consequence.

I rather doubt that nfb was the problem per se.

Graham

[Eeyore]

Chris Hornbeck wrote:

On 20 Dec 2006 17:51:32 -0800, "Peter Wieck" wrote:

Again, tubes have parameters that as compared to *similarly
functioning* transistors are wide enough for the QE2 to pass through at
speed. There are exceptions to every generalization of course.

No idea what you mean here, but taking the most obvious
reading, this is very incorrect.

Surely you're not saying that tolerances for semiconductors
are lower than tolerances for electron valves? That would
be wacky.

Do explain why.

Graham

[Arny Krueger]

"Doug" wrote in message
ups.com

Although tubes manufactured under a given part number do
exhibit some spread in their parameters (gm, rp, u etc.),
they exhibit nowhere near the parameter spead inherent in
bipolar transistors or FETs.

This would be in some alternative universe where tubes never degrade because
they simply wear out?

Just take a look in any
transistor data book to observe the spreads, for example,
in hfe for bipolar devices or in Vp and Idss for FETs.

Just take a look at what happens to tubes when you put old ones into a tube
checker.

Great care must be taken in fact in the design of solid
state circuits to accomodate these variations so that all
final product in a production run will meet its specifications.

Great care must be taken in fact in the design of tubed
circuits to accomodate the normal performance variations over the life of a
tube, so that all
product in actual use will meet its specifications as the tubes degrade.

This is problematical in certain common circuits such as the classic
twin-triode RIAA preamp. The tubes used often have barely enough gain to
maintain predictable operation at low frequencies when they are new.

Among the measures that are routinely taken in solid
state designs to accomodate these variations are the
application of considerable feedback, be it local, global
or both.

The essence of negative feedback is a basic amplifier stage that has enough
surplus gain so that this gain can be sacrificed so that the circuit
operates predictably and stably even with normal variation in the parameters
of active components. It is relatively easy and economical to build
solid-state circuits that have considerable excess gain that can be
sacrificed this way. Because of the size and cost of tubes and associated
power supplies, etc. it has been less common for tubed circuits to be
overbuilt to a comparable extent.

Bipolar devices in fact exhibit such a
non-linear transfer characterisitic that they are
unviable for audio amplification without these measures
being taken.

Historically, solid state products have had vastly lower amounts of
distortion than their tubed predecessors.

On the other hand, if you were to take as an example a
sample of US built 12AX7s and examine their
characteristic curves on a curve tracer you would be very
surprised at how closely their curves match. I do not
know how well their curves would match against Russian or
Chinese product, or against that of Telefunken, Amperex,
Mullard etc.

Ironically, preamps built with 12AX7s have never come close to providing the
low distortion operation that we take for granted with their solid state
predecessors. Admittedly, tubed equipment can have low enough distortion
that the higher measured levels of nonlienar distortion is not necessarily
an audible problem.

However, truly low distortion operation with tubes was more likely to be
reserved for premium-priced products. For example, compare the measured
performance of a Dyna Mark III with a McIntosh 75 after a year of use. As
many McIntosh clinics showed, McIntosh amplifiers as a rule could maintain
less than 0.25% THD 20-20 KHz year after year without renewal of tubes,
while Dyna and Eico amps often needed new tubes after a year or less of
regular operation. McIntosh amplifiers were well-known for their heavy use
of negative feedback, both local and loop.

It is certainly possible to build a solid state amplifier
with a similar component count to a tube counterpart.

Component count is a very narrow way to evaluate the performance of an
amplifier. For example, a Mc 75 has 9 tube functions handling the signal
(mostly triodes) while a Dyna Mark III had only 4, only one of which was a
triode.

Many bipolar designs from the 70s were very simple. These
designs however relied heavily on gobs of global negative
feedback to linearize them, and in my humble opinion the
sound suffered as a consequence.

An opinion that feedback hurts sound quality is contrary to the
generally-accepted principles of audio engineering, and therefore can never
be considered to be "humble". Both local and loop feedback was heavily-used
in tubed audio production equipment, for example. LPs were frequently cut
using McIntosh amplifiers that featured heavy use of local and loop
feedback. It needs to be explained why the laws of electrical engineering
would be different for recording versus playback equipment.

[Eeyore]

Arny Krueger wrote:

This is problematical in certain common circuits such as the classic
twin-triode RIAA preamp. The tubes used often have barely enough gain to
maintain predictable operation at low frequencies when they are new.

Note the recent thread about RIAA time constants. It becomes very difficult
indeed to create an accurate eq curve when you're short of open loop gain.


It is relatively easy and economical to build
solid-state circuits that have considerable excess gain that can be
sacrificed this way.

There's a simple 'rule of thumb' - in fact it's really quite accurate that a
single transistor common emitter stage can offer 20 x the supply voltage worth
of gain.

So with a once popular 18V supply the voltage gain without feedback will be ~
360x.

Using a high-voltage transistor such as MPSA42 with a 250V supply would offer a
remarkable 5,000x gain ! Using local feedback to reduce this to a more tube-like
25x would use 46dB ( 200x ) voltage feedback, thus reducing THD from say 5% to
0.025%.

Graham

[Ruud Broens]

"Eeyore" wrote in message
...
:
:
: Arny Krueger wrote:
:
: This is problematical in certain common circuits such as the classic
: twin-triode RIAA preamp. The tubes used often have barely enough gain to
: maintain predictable operation at low frequencies when they are new.
:
: Note the recent thread about RIAA time constants. It becomes very difficult
: indeed to create an accurate eq curve when you're short of open loop gain.
:
:
: It is relatively easy and economical to build
: solid-state circuits that have considerable excess gain that can be
: sacrificed this way.
:
: There's a simple 'rule of thumb' - in fact it's really quite accurate that a
: single transistor common emitter stage can offer 20 x the supply voltage worth
: of gain.
:
: So with a once popular 18V supply the voltage gain without feedback will be ~
: 360x.
:
: Using a high-voltage transistor such as MPSA42 with a 250V supply would offer a
: remarkable 5,000x gain ! Using local feedback to reduce this to a more
tube-like
: 25x would use 46dB ( 200x ) voltage feedback, thus reducing THD from say 5% to
: 0.025%.
:
: Graham

true, using fb will smooth all performance parameters.
there is a school of thought, any amplification stage should start out
with as little distortion as possible, before fb is applied,
where somewhat lower gain combined with lower fb
can then give impressive results also.
Av=25 with 0.025 % is not particularly hard to accomplish
with tubes , eg. in the hybrid circuit thread i gave the example
of using a PABC80 's triode doing 0.1 % at Av=400, with a B+ of 130V,
so that beats the MPSA42 at least 8 times, hehe.

pushing the envelope
Rudy

[Peter Wieck]

Chris Hornbeck wrote:

Surely you're not saying that tolerances for semiconductors
are lower than tolerances for electron valves? That would
be wacky.

Not at all. Semi tolerances are much tighter than those for tubes
(Parameters wide enough for the QE2 at speed does not suggest tight
tolerances). But also of an entirely different nature. Best analogy
that comes to mind is that a tube does function similarly to a valve, a
transistor similarly to a switch.

Similarly....

Not Exactly.

But the differences in performance are manifest. Were they not so,
there would be no need for (or fun in) tubes.

Peter Wieck
Wyncote, PA

[Peter Wieck]

Andy, you are near-enough brain-dead as not to matter.

Try to use what is left of your grey cells amongst the neurofibrillary
tangles to follow what is written vs. harkening only to your own
fantasies.

Peter Wieck
Wyncote, PA

[Patrick Turner]

Arny Krueger wrote:

"Patrick Turner" wrote in message


Well sometimes what we measure does not explain the sound
we hear.

Expand on that.


I have no intentiong of "expanding" on this.
Its all been said before.


Isolated measurements mean nothing. I measure an object and
it is 1/3 the length of my ruler. I say, "It is 33%".

My point is that no isolated measurement has any meaning without context.
Doesn't matter what it is. Your comment is therefore completely trivial and
meaningless.

Its pointless being obsessive about THD for example,
especially when its below 0.05% and all mainly 2H.

Now you contradicted what you just said. You've in effect said that 0.05% 2H
is less than that which is audibly significant. Of course, you added
context. You've proven my point, which is that what we measure can explain
what we hear if we add appropriate context.

But an arrangement of source, preamp, power amp and
speakers
will usually sound different to a different set of
componenents. Especially in a different room as well.

So what?

But the numbers are important though, and I like amps and
all the other bits to make
low amounts of THD and IMD, and its the IMD that so
easily does the damage.

Missing context: IMD and THD are both the results of nonlinearity. It is
difficult to contrive things so that one does not come without the other.

I don't recall hearing too many tube amps that were much
improved by lowering the THD from
0.1% to 0.01%, let alone 0.001%, a difficult feat for a
tube power amp.

If you study the literature and do some experiments, you will find that some
place around 0.1% nonlinear distortion, further improvements in linearity
tend to be moot, no matter what sort of listening test you try to do.

And an SS amp won't sound better just
because it can get down to 0.001% THD.

Agreed.

But nor does it suffer from being constructed so that
0.001% is typical at listening levels;

Not necessarily.

the absense of an OPT allows more FB so it can measure
better. Big deal.

Getting rid of the OPT has other benefits. Heck, even tubed amps can benefit
from getting rid of the OPT.

Getting rid of the OPT in tubed power amps usually means the
tubes must be run in low bias current class aB condition which least
suits their
natural linearity. Usually OTL amps have an appallingly bad load match,
and even with gobs of loop NFB they fail to be as linear or act low
enough Rout
compared to transformer coupled tubes.
The OPT has earned itself a bad name from anti-tubists mainly because
there have been so many very poor OPT around foisted on an unsuspecting
public.

But decent OPT don't have a detectably deleterious effect on music.

If an OPT isn'r wanted, then my advice is to use lots a power mosfets,
they are like pentoads on steroids.

In short, tube amps DO NOT benefit from removing the OPT.

You know how many EL34 in triode it takes to get a 5K : 8 ohm load match
without an OPT?

Its 625 x EL34.

But the amp will then make about 3,000 watts at 4% THD, and at 2 watts
THD is rather low....
Rout = Ra / 625 = about 2 ohms.

But with an OPT and ONE EL34 you can 2 watts and low enough THD and Rout
1ohm,
and with triode operation and with only 12 dB of loop NFB.

I've built 50 watts class A amps with mosfet outputs and bjt driver
circuits
with faily low open loop gain and only a total of 22 dB of loop NFB
and they could NOT be discerned to have had an OPT within unless the
listener is told
about it before hand.

The OPT allows a splendid variety of load matches, and there is no
possibility
of speaker damage if a mosfet ****s itself to a short circuit to save a
fuse from blowing
which all too often happens with solid state amps; my shed has a
continuous flow
of SS amps on the bench which have pooped themselves.

The OPT allows NPN or PNP devices to be used on each side of the PP
circuit.
So natural matching is possible.
The OPT can be an auto transformer, and is simple to wind compared to an
interleaved
multi turn tube OPT.
But at least whatever is allowed out of my shed gives OPTs a good name,
and shows what can be done
without bean counter influence.





Some people say they
are cold and clinical and lifeless, and one needs triodes
to present the real thing that was heard in the studio or
concert hall recording session.

Some people say the darndest things, especially if they think it will buy
them some influence over others.

Well all the hundrreds of tube haters who say the opposite to me
don't stop so many people banging on my door.



I don't argue, I just build the amps for them; I try to
get the numbers right while maintaining a semblance of
simplicity. After awhile one finds out what seems to
really work the best, but
many varieties of tube circuits work well.

It is usually a mistake to overdefine the solution. Asking that an amp does
not audibly change its input signal under any reasonble operating condition
should suffice.

From your last statement many things you might say don't necessarily
follow.

Patrick Turner.

[Eeyore]

Ruud Broens wrote:

"Eeyore" wrote in message
: Arny Krueger wrote:
:
: This is problematical in certain common circuits such as the classic
: twin-triode RIAA preamp. The tubes used often have barely enough gain to
: maintain predictable operation at low frequencies when they are new.
:
: Note the recent thread about RIAA time constants. It becomes very difficult
: indeed to create an accurate eq curve when you're short of open loop gain.
:
:
: It is relatively easy and economical to build
: solid-state circuits that have considerable excess gain that can be
: sacrificed this way.
:
: There's a simple 'rule of thumb' - in fact it's really quite accurate that a
: single transistor common emitter stage can offer 20 x the supply voltage worth
: of gain.
:
: So with a once popular 18V supply the voltage gain without feedback will be ~
: 360x.
:
: Using a high-voltage transistor such as MPSA42 with a 250V supply would offer a
: remarkable 5,000x gain ! Using local feedback to reduce this to a more
tube-like
: 25x would use 46dB ( 200x ) voltage feedback, thus reducing THD from say 5% to
: 0.025%.
:
: Graham

true, using fb will smooth all performance parameters.
there is a school of thought, any amplification stage should start out
with as little distortion as possible, before fb is applied,
where somewhat lower gain combined with lower fb
can then give impressive results also.
Av=25 with 0.025 % is not particularly hard to accomplish
with tubes , eg. in the hybrid circuit thread i gave the example
of using a PABC80 's triode doing 0.1 % at Av=400, with a B+ of 130V,
so that beats the MPSA42 at least 8 times, hehe.

Hey, if you're going to use a hybrid I'll use a compound pair. The THD with that
arrangement is vanishingly small. Easily in the 0.003 % area.

Graham

[Patrick Turner]

Eeyore wrote:

Doug wrote:

Although tubes manufactured under a given part number do exhibit some
spread in their parameters (gm, rp, u etc.), they exhibit nowhere near
the parameter spead inherent in bipolar transistors or FETs. Just take
a look in any transistor data book to observe the spreads, for example,
in hfe for bipolar devices or in Vp and Idss for FETs. Great care must
be taken in fact in the design of solid state circuits to accomodate
these variations so that all final product in a production run will
meet its specifications.

With respect to bipolar transistors that spead in hfe hardly presents any great problems and certainly doesn't require 'great care' etc...... You simply design to accomadate the lowest hfe you're likely to meet and in most cases everythings fine.
Designing hfe senstive circuitry is to be deprecated.

In all other respects the parameters are highly consistent.

Graham

In a recent precious period of 2 weeks of sparse spare time between
frantic
customer repairs before Xmas, I designed and built
a phono preamp for moving coil use using all solid state devices.

There are 3 stages of basically common source and common emitter stages
each with local current FB in the source/emitter circuits.

stage 1 which is 3 paralleled 2SK369 j-fets,
stage 2 is a kind of bootstrapped follower stage using j-fet and bjt
follower,
stage 3 is a similar kind of bootstrapped follower using two darlington
pairs.
All stages are single ended.
No loop FB is used and the RIAA eq is done with two passive networks.
after stg 1 there is 3180 and 318 uS time constant eq,
and after stg 2 there is 75us and 3 uS time constant eq.

Rail voltages are +35V and -35V.

The amp is capable of 10Vrms at any F between 4 Hz and 32 kHz without
severe saturation distortions,
ie, unlike many other phono amps this one does not begin to maul the
signal above 2Vrms at 5kHz.

THD is less than 0.01% at 2Vrms, 1 kHz.

Overall gain in most sensitive MC choice is about 60dB for 1 kHz.

Care was taken to choose the drain loads for the j-fet stages to be
about 6.4k ohms
because with Id = 5 mA, and Ed = +11V, there was a load where the 2H
dropped to zero.
below and above this load value the 2H was of differing phase. When this
load was used the 3H was
about 20dB lower than the 2H, so I can say i have found a "magical" load
value which results in naturally low THD.
the 2SK369 is like a pentode with µ = 3,200, Rd = 80,000 ohms and Gm =
0.04Amps per volt.

This is like a pentode on steroids, or like about four high Gm signal
pentodes such as 6EH7 all paralleled.

Of course the j-fets give a 20 dB better SNR than any normal tube, which
is why I have used them.

I will be publishing the full schematic of this circuit later for the
ppl
who like QUIET and LINEAR behaviour from SS preamps without reliance on
60db of loop NFB around an opamp.
This should give folks something to get their teeth into, and be a great
teaching lesson
about basic behaviours.

I am not aware of a "magical" load value for BJTs where the 2H virtually
dissappears.
The last stage of the phono amp has a gain of 27X which is the open loop
gain of around 1,500X
reduced by simple local current FB.
I though there was no need to find the magic load, knowing that a lot of
local current NFB was to be applied,
so whatever the THD od the stage was, it would be reduced to better than
triode proportions,
ie, the same 10Vrms from a triode output stage with gain = 27X would
give more rhater than less THD.

The Gm of the gain bjt used was effectively about 0.15A/V for Ic = 5mA,
and this may vary considerably between samples of bjt.
But the voltage gain is determined by the local current FB and the final
gain after FB is set
mainly by the ratio of the Re to Rc load, and just how I loaded the bjt
to get an open loop gain of
1,500 approx is a secret you will all have to wait for when i publish.
But the stage design is very like a typical bootstrapped follower stage
using two triodes, except
that i am using 4 bjts to make two darlington pairs to use in place of
two triodes.
Those not familiar with fine functioning bootstrapped follower triode
stage should go to
my schematic for the 'Rocket' preamp at
http://www.turneraudio.com.au/preamp...hono-2005.html
See V2 - V3.
But without the local emitter R current NFB the bjt is very non linear
in its voltage amplifying
ability even though the Ib to Ic ratio is a substantially linear
character for any bjt.


Graham is right about designing for lowest hfe, but hfe isn't
everything, and for best operation
darlington pairs should be routinely chosen because the connection gives
HIGH INPUT RESISTANCE,
ie, the total hfe is the product of the hfe of the two bjts, and hfe is
largely banished as something
one needs to worry about when designing simple discrete bjt stages.
Without any huge current draw in the base circuit the bjts involved can
better get on with business
without the output interacting with input in untoward ways.
For even better-in-some-ways darlington operation one can use a j-fet
direct connected to the bjt
to make the darlington pair.

Biasing becomes easier with darlingtons, and there is an absense of
rectification effects
which can grossly affect THD/IMD as the output voltage rises above a
trifle.
In effect, the darlington allows a simple common emitter stage to act
like a tube stage with its
beneficial zero current input, or very high input resistance.

Oh, and final SNR with respect to 0.5V output was excellent.

Not a stitch of global NFB was used as is normally done by the
bucketfull in nearly all other SS amps.

Bucket loads of NFB is also used in tube amps such as ARC, CJ, etc, and
rather excessively, and obsessively, IMHO.
Some of these are over complex, and with too much hybridisation so they
end up
being over engineered IMHO.


I continued with having fun with bits and peices when i re-built a wien
bridge oscillator
that uses all tubes, and the two necessary global loops of feedback, one
is positive, the other
is negative.
Now methinks that if I wanted less than the 0.04% THD that I ended up
with at 1 kHz, i should perhaps use
a gyrator circuit to mimic an inductance, and bypass it with a cap to
make an inductorless LC parallel
resonant circuit, and with a Q of say 12, and then the THD would be a
lot lower.

Now why would this occur, bearing in mind that the wien bridge Q is very
low?????

Let us see who is asleep or not!!!


Patrick Turner.

[Patrick Turner]

Eeyore wrote:

Ruud Broens wrote:

"Eeyore" wrote in message
: Arny Krueger wrote:
:
: This is problematical in certain common circuits such as the classic
: twin-triode RIAA preamp. The tubes used often have barely enough gain to
: maintain predictable operation at low frequencies when they are new.
:
: Note the recent thread about RIAA time constants. It becomes very difficult
: indeed to create an accurate eq curve when you're short of open loop gain.
:
:
: It is relatively easy and economical to build
: solid-state circuits that have considerable excess gain that can be
: sacrificed this way.
:
: There's a simple 'rule of thumb' - in fact it's really quite accurate that a
: single transistor common emitter stage can offer 20 x the supply voltage worth
: of gain.
:
: So with a once popular 18V supply the voltage gain without feedback will be ~
: 360x.
:
: Using a high-voltage transistor such as MPSA42 with a 250V supply would offer a
: remarkable 5,000x gain ! Using local feedback to reduce this to a more
tube-like
: 25x would use 46dB ( 200x ) voltage feedback, thus reducing THD from say 5% to
: 0.025%.
:
: Graham

true, using fb will smooth all performance parameters.
there is a school of thought, any amplification stage should start out
with as little distortion as possible, before fb is applied,
where somewhat lower gain combined with lower fb
can then give impressive results also.
Av=25 with 0.025 % is not particularly hard to accomplish
with tubes , eg. in the hybrid circuit thread i gave the example
of using a PABC80 's triode doing 0.1 % at Av=400, with a B+ of 130V,
so that beats the MPSA42 at least 8 times, hehe.

Hey, if you're going to use a hybrid I'll use a compound pair. The THD with that
arrangement is vanishingly small. Easily in the 0.003 % area.

Graham

Describe the circuit in more detail.

Please.


But an LTP with a pair of triodes is surprisingly linear
at signal voltage levels.....
And with matched triodes its even more surprising.


Patrick Turner.

[Ruud Broens]

"Eeyore" wrote in message
...
:
:
: Ruud Broens wrote:
:
: "Eeyore" wrote in message
: : Arny Krueger wrote:
: :
: : This is problematical in certain common circuits such as the classic
: : twin-triode RIAA preamp. The tubes used often have barely enough gain to
: : maintain predictable operation at low frequencies when they are new.
: :
: : Note the recent thread about RIAA time constants. It becomes very difficult
: : indeed to create an accurate eq curve when you're short of open loop gain.
: :
: :
: : It is relatively easy and economical to build
: : solid-state circuits that have considerable excess gain that can be
: : sacrificed this way.
: :
: : There's a simple 'rule of thumb' - in fact it's really quite accurate that
a
: : single transistor common emitter stage can offer 20 x the supply voltage
worth
: : of gain.
: :
: : So with a once popular 18V supply the voltage gain without feedback will be
~
: : 360x.
: :
: : Using a high-voltage transistor such as MPSA42 with a 250V supply would
offer a
: : remarkable 5,000x gain ! Using local feedback to reduce this to a more
: tube-like
: : 25x would use 46dB ( 200x ) voltage feedback, thus reducing THD from say 5%
to
: : 0.025%.
: :
: : Graham
:
: true, using fb will smooth all performance parameters.
: there is a school of thought, any amplification stage should start out
: with as little distortion as possible, before fb is applied,
: where somewhat lower gain combined with lower fb
: can then give impressive results also.
: Av=25 with 0.025 % is not particularly hard to accomplish
: with tubes , eg. in the hybrid circuit thread i gave the example
: of using a PABC80 's triode doing 0.1 % at Av=400, with a B+ of 130V,
: so that beats the MPSA42 at least 8 times, hehe.
:
with a 130V, Rc=33K, 2mA, rE being 12.5 Ohm, the MPSA42
will do 33000/12.5 for Av about 2700
with typical hfe of 100, input impedance is a rather unusable 1.2K..

: Hey, if you're going to use a hybrid I'll use a compound pair. The THD with
that
: arrangement is vanishingly small. Easily in the 0.003 % area.
:
: Graham

hmm, you should really go read that hybrid thread :-)
with 36V supply, flipper got this (6GM8, 'current loop config.') :
"Holy Hannah, spice says I'm getting an Av of 325 with 2'nd -92dB"

well, spiced, anyway ;-)

Rudy
into the socket

[Arny Krueger]

"Ruud Broens" wrote in message


in the hybrid circuit thread i gave the
example of using a PABC80 's triode doing 0.1 % at Av=400, with a
B+ of 130V, so that beats the MPSA42 at least 8 times,
hehe.

Apparently you like to invent tube specs out of thin air, Ruddy.

The PABC80 is speced by its producers to have a maximum Av of 60 and 0.3%
THD.

http://tubedata.tigahost.com/tubedat...4/e/EABC80.pdf

The impedances required to do this are very high, high enough to make good
AF response questionable.

[Eeyore]

Ruud Broens wrote:

with a 130V, Rc=33K, 2mA, rE being 12.5 Ohm, the MPSA42
will do 33000/12.5 for Av about 2700
with typical hfe of 100, input impedance is a rather unusable 1.2K..

But you don't usually need an Av of 2700. Add an emitter R of 1k and you've got a
rather more useful Av of 33x and Zin is 100k or so.

Of course a darlington arrangement will increase it quite happily by far more.


: Hey, if you're going to use a hybrid I'll use a compound pair. The THD with
: that: arrangement is vanishingly small. Easily in the 0.003 % area.
:
: Graham

hmm, you should really go read that hybrid thread :-)
with 36V supply, flipper got this (6GM8, 'current loop config.') :
"Holy Hannah, spice says I'm getting an Av of 325 with 2'nd -92dB"

Maybe he'd like to post that circuit arrangement ?

Graham

[Ruud Broens]

"Arny Krueger" wrote in message
. ..
: "Ruud Broens" wrote in message
:
:
: in the hybrid circuit thread ~ i gave the
: example of using a PABC80 's triode doing 0.1 % at Av=400, with a
: B+ of 130V, so that beats the MPSA42 at least 8 times,
: hehe.
:
: Apparently you like to invent tube specs out of thin air, Ruddy.
:
: The PABC80 is speced by its producers to have a maximum Av of 60 and 0.3%
: THD.
:
: http://tubedata.tigahost.com/tubedat...4/e/EABC80.pdf
:
: The impedances required to do this are very high, high enough to make good
: AF response questionable.


Don't worry, Army, some surplus GogglesT may await you,
under that tree vacuum impregnated, no less
;-)
~there
Rudy

[Ruud Broens]

"Eeyore" wrote in message
...
:
:
: Ruud Broens wrote:
:
: with a 130V, Rc=33K, 2mA, rE being 12.5 Ohm, the MPSA42
: will do 33000/12.5 for Av about 2700
: with typical hfe of 100, input impedance is a rather unusable 1.2K..
:
: But you don't usually need an Av of 2700. Add an emitter R of 1k and you've got
a
: rather more useful Av of 33x and Zin is 100k or so.
:
: Of course a darlington arrangement will increase it quite happily by far more.
:
:
: : Hey, if you're going to use a hybrid I'll use a compound pair. The THD with
: : that: arrangement is vanishingly small. Easily in the 0.003 % area.
: :
: : Graham
:
: hmm, you should really go read that hybrid thread :-)
: with 36V supply, flipper got this (6GM8, 'current loop config.') :
: "Holy Hannah, spice says I'm getting an Av of 325 with 2'nd -92dB"
:
: Maybe he'd like to post that circuit arrangement ?
:
: Graham

http://flipperhome.dyndns.org/TubeCurrentLoop.htm
was the first implementation.

cheers,
R.

[Arny Krueger]

"Ruud Broens" wrote in message

"Eeyore" wrote
in message ...


Ruud Broens wrote:

with a 130V, Rc=33K, 2mA, rE being 12.5 Ohm, the MPSA42
will do 33000/12.5 for Av about 2700
with typical hfe of 100, input impedance is a rather
unusable 1.2K..

But you don't usually need an Av of 2700. Add an emitter
R of 1k and you've got a rather more useful Av of 33x
and Zin is 100k or so.

Of course a darlington arrangement will increase it
quite happily by far more.


Hey, if you're going to use a hybrid I'll use a
compound pair. The THD with that: arrangement is
vanishingly small. Easily in the 0.003 % area.

Graham

hmm, you should really go read that hybrid thread :-)
with 36V supply, flipper got this (6GM8, 'current loop
config.') : "Holy Hannah, spice says I'm getting an Av
of 325 with 2'nd -92dB"

Maybe he'd like to post that circuit arrangement ?

Graham

http://flipperhome.dyndns.org/TubeCurrentLoop.htm
was the first implementation.

I strongly suspect you could get far more undistorted output from those 4
transistors if you left out the thermionic nonlinear distortion generator
and energy sink.

[Ruud Broens]

"Arny Krueger" wrote in message
. ..
: "Ruud Broens" wrote in message
:
: "Eeyore" wrote
: in message ...
:
:
: Ruud Broens wrote:
:
: with a 130V, Rc=33K, 2mA, rE being 12.5 Ohm, the MPSA42
: will do 33000/12.5 for Av about 2700
: with typical hfe of 100, input impedance is a rather
: unusable 1.2K..
:
: But you don't usually need an Av of 2700. Add an emitter
: R of 1k and you've got a rather more useful Av of 33x
: and Zin is 100k or so.
:
sure. the point would be, you desperately *need* nfb there
to get something useful - not so for the hybrid :-)
there you mainly need it to decrease input sensitivity
and bring distortion down ~even further~

: Of course a darlington arrangement will increase it
: quite happily by far more.
:
:
: Hey, if you're going to use a hybrid I'll use a
: compound pair. The THD with that: arrangement is
: vanishingly small. Easily in the 0.003 % area.
:
: Graham
:
: hmm, you should really go read that hybrid thread :-)
: with 36V supply, flipper got this (6GM8, 'current loop
: config.') : "Holy Hannah, spice says I'm getting an Av
: of 325 with 2'nd -92dB"
:
: Maybe he'd like to post that circuit arrangement ?
:
: Graham
:
: http://flipperhome.dyndns.org/TubeCurrentLoop.htm
: was the first implementation.
:
: I strongly suspect you could get far more undistorted output from those 4
: transistors if you left out the thermionic nonlinear distortion generator
: and energy sink.
:
i say, go for it. show us the circuit, Arry ;-)
R.
eealitycheck

[Arny Krueger]

"Patrick Turner" wrote in message

Arny Krueger wrote:

Getting rid of the OPT in tubed power amps usually means
the
tubes must be run in low bias current class aB condition
which least suits their
natural linearity. Usually OTL amps have an appallingly
bad load match, and even with gobs of loop NFB they fail
to be as linear or act low enough Rout
compared to transformer coupled tubes.
The OPT has earned itself a bad name from anti-tubists
mainly because there have been so many very poor OPT
around foisted on an unsuspecting public.

But decent OPT don't have a detectably deleterious effect
on music.

Seems like.

If an OPT isn'r wanted, then my advice is to use lots a
power mosfets, they are like pentoads on steroids.

Nothing wrong with BJTs.

In short, tube amps DO NOT benefit from removing the OPT.

Let's see how you justfiy that, Patrick.

You know how many EL34 in triode it takes to get a 5K : 8
ohm load match without an OPT?

Who in heaven's name would use El34s in a OPT-less power amp?

Aren't tubes like the 6C33 which modern OTL power amp builders seem to use a
lot, more to the point?

Its 625 x EL34.

Let's see how you justfiy that, Patrick.

But the amp will then make about 3,000 watts at 4% THD,
and at 2 watts THD is rather low....
Rout = Ra / 625 = about 2 ohms.

Sounds like a bad choice of output devices. Is this sort of weirdness
typical of your engineering technique, Patrick?

But with an OPT and ONE EL34 you can 2 watts and low
enough THD and Rout 1ohm,
and with triode operation and with only 12 dB of loop NFB.

Seems like you've totally missed the point, and just want to pull together
some justification for a prejudice against OTL tubed power amps, Patrick.

Are these guys lying?

http://members.aol.com/aria3/zout6c33.htm

Looks like maybe 16 6C33 output devices would make a failrly nice power OTL
amp.

16 625.

Obviously, if money and practicality were the goal, we'd go SS, but given
that we want to use modern tube technology to its fullest, why use an OPT
tube that is designed for high voltages and high impedances in a low
impedance circuit?