I've been looking at various options for my push pull
amplifier and have obtained some 4X150A's (6.0V fil) and D's
(26.5v fil) which are a precursor to the 4CX250's only with a
glass lower seal (may see filament glow here!!). These appear
to have identical ratings according to the RCA datasheets.

Now my question is this: How much power can I draw
from a concertina phase splitter and still have it remain linear?.

According to my datasheets I can use a B+ of 1250V and obtain
310 watts in class AB with a pair using zero watts of driving power
or I can alter the bias and get 425 watts in class AB2 with a
maximum peak driving power of 0.15 watts, 0.075 w nominal.

So if I can get 300 milliwatts peak out of a single
ECC84 valve (similar to a 6DJ8) concertina stage, I should
be able to make a 3 valve push pull amp with up to 425 watts
of class AB2 power.

Mark Harriss

"Mark Harriss" wrote in message
...
So if I can get 300 milliwatts peak out of a single
ECC84 valve (similar to a 6DJ8) concertina stage, I should
be able to make a 3 valve push pull amp with up to 425 watts
of class AB2 power.

There's little way around it, you have to use transformer coupling or
cathode followers. Something has to absorb the DC produced by the
rectifying effect of positive grid voltage. An RC-coupled stage will *not*
do, as the grid leak will have to be excessively small to produce an
acceptable bias drift, and this makes for a much larger driver.

A cathodyne will absolutely not produce useful output power into such a
varying load, because the plate has a high Zo, while the cathode is low.
What you'll see on the output is a little nipple forming on the peak of the
sine wave. I've seen it in Frankenhouse during testing.

I don't like the idea of a 3 tube amp because you'll need 20V to 40V input
for max. power output after wrapping it with NFB. Maybe add a 6AB4 or
something.

....Did Nick Sheldon ever say he was coming back some time? Someone e-mail
him about this project

Tim

--
"That's for the courts to decide." - Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

Tim Williams wrote:

There's little way around it, you have to use transformer coupling or
cathode followers. Something has to absorb the DC produced by the
rectifying effect of positive grid voltage. An RC-coupled stage will
*not* do, as the grid leak will have to be excessively small to produce an
acceptable bias drift, and this makes for a much larger driver.


Hi Tim, I don't know what Nick thinks of the project, it may be
a bit too small for him. The circuit I'm thinking of following is the "Main"
amp project in Morgan Jone's Valve Amplifiers book on page 233.
Morgan is using a cathode biased 6DJ8 concertina to drive a pair of RC
coupled 6BQ5. This amp has 10 watts output with 2 volt input sensitivity.

I felt I could alter the basic concertina to suit a 6SN7 or an
ECC84 to get the necessary voltage swing across the RC coupled
grids of the 4X150's although if I want to try a cathode feedback winding
on the OPT I'll have to come up with some more drive power.

Mark Harriss wrote:

I've been looking at various options for my push pull
amplifier and have obtained some 4X150A's (6.0V fil) and D's
(26.5v fil) which are a precursor to the 4CX250's only with a
glass lower seal (may see filament glow here!!). These appear
to have identical ratings according to the RCA datasheets.

Now my question is this: How much power can I draw
from a concertina phase splitter and still have it remain linear?.

If the CPI is resistance loaded, the loads seen by the anode and cathode
circuits have to be
equal through all parts of the cycle or the cathode and naode drive
voltage will vary,
because taken individually, Ro at anode is far higher than Ro at
cathode.





According to my datasheets I can use a B+ of 1250V and obtain
310 watts in class AB with a pair using zero watts of driving power
or I can alter the bias and get 425 watts in class AB2 with a
maximum peak driving power of 0.15 watts, 0.075 w nominal.

For wattless input, a CPI is fine, since no grid current is drawn.
But isn't the drive voltage required a bit high for CPI?
And for class AB2, low impedance drive is essential,
so a choke with CT could be used to bias the tubes, and the choke
ends driven directly by CF tubes operating off a differential gain
stage,
which is buffered from the effects of the grid current.



So if I can get 300 milliwatts peak out of a single
ECC84 valve (similar to a 6DJ8) concertina stage, I should
be able to make a 3 valve push pull amp with up to 425 watts
of class AB2 power.

That's like employing a flea to wield the whip on a
large plow horse, would not a proper drive tube be better,
like EL84 in triode or such like?

Patrick Turner.



Mark Harriss

Hi Pat,


Patrick Turner wrote:
Mark Harriss wrote:
Now my question is this: How much power can I draw
from a concertina phase splitter and still have it remain linear?.

If the CPI is resistance loaded, the loads seen by the anode and cathode
circuits have to be
equal through all parts of the cycle or the cathode and naode drive
voltage will vary,
because taken individually, Ro at anode is far higher than Ro at
cathode.

So the instant current is drawn from a concertina stage things start
to go pear shaped? (asymmetrical). If you could guarantee no
power is drawn by the driver valve grid things would be fine.


According to my datasheets I can use a B+ of 1250V and obtain
310 watts in class AB with a pair using zero watts of driving power
or I can alter the bias and get 425 watts in class AB2 with a
maximum peak driving power of 0.15 watts, 0.075 w nominal.

For wattless input, a CPI is fine, since no grid current is drawn.
But isn't the drive voltage required a bit high for CPI?

According to the datasheets (which triggered this line of enquiry)
a pair of 4X150's need -96 volts grid1 to grid1 with zero watts
required to get 300W.


And for class AB2, low impedance drive is essential,
so a choke with CT could be used to bias the tubes, and the choke
ends driven directly by CF tubes operating off a differential gain
stage,
which is buffered from the effects of the grid current.

I'm not sure I follow, does the choke supply the cathode followers
or do they drive the choke?.


So if I can get 300 milliwatts peak out of a single
ECC84 valve (similar to a 6DJ8) concertina stage, I should
be able to make a 3 valve push pull amp with up to 425 watts
of class AB2 power.

That's like employing a flea to wield the whip on a
large plow horse, would not a proper drive tube be better,
like EL84 in triode or such like?

Patrick Turner.

From what I can see the drive requirements aren't all that
savage, G1 has a maximum dissipation of 2 watts. At any rate
it looks like the cathode followers will keep things symmetrical
when the output stage is driven hard. I could probably use
6SN7's for concertina and cathode followers or alternatively
a 12AX7 for concertina and 12AU7 for followers.




Mark Harriss

Mark Harriss wrote:

Hi Pat,

Patrick Turner wrote:
Mark Harriss wrote:
Now my question is this: How much power can I draw
from a concertina phase splitter and still have it remain linear?.

If the CPI is resistance loaded, the loads seen by the anode and cathode
circuits have to be
equal through all parts of the cycle or the cathode and naode drive
voltage will vary,
because taken individually, Ro at anode is far higher than Ro at
cathode.

So the instant current is drawn from a concertina stage things start
to go pear shaped? (asymmetrical). If you could guarantee no
power is drawn by the driver valve grid things would be fine.

This should be the case.
The loads on the CPI are the A and K resistors, and the cap coupled grid bias
resistors,
and the if there is no grid current, the only other load is the
miller C and other stray C.





According to my datasheets I can use a B+ of 1250V and obtain
310 watts in class AB with a pair using zero watts of driving power
or I can alter the bias and get 425 watts in class AB2 with a
maximum peak driving power of 0.15 watts, 0.075 w nominal.

For wattless input, a CPI is fine, since no grid current is drawn.
But isn't the drive voltage required a bit high for CPI?

According to the datasheets (which triggered this line of enquiry)
a pair of 4X150's need -96 volts grid1 to grid1 with zero watts
required to get 300W.

So about 50 vrms to each grid.

The CPI then needs to have a very high B+ supply, but you got that OK,
maybe use +700v, and have 175 v across RLk and RLa,
and 350v across the tube, but I would use EL84 in triode,
with about 15 mA, which means RL a&k are about 12k ohms,
and a 5 ma rms current change in the tube will give 60 vrms change.
Since the EL84 in triode has a U of 20, the gain reduction
is from about 17 to 2, ( counting Vk and Va )
so thd would be reduced from around 3% to 0.35% at 60v,
and you have some reserve.
6V6 would look ok, even something like EL34....

The heaters need to be biased up at about 200v.




And for class AB2, low impedance drive is essential,
so a choke with CT could be used to bias the tubes, and the choke
ends driven directly by CF tubes operating off a differential gain
stage,
which is buffered from the effects of the grid current.

I'm not sure I follow, does the choke supply the cathode followers
or do they drive the choke?.

First you have an LTP, say a 6SN7 arranged with 68 k RLs to
say +600v supply, and fitted with a CCS tail to a -ve voltage.
This allows a single input to one side of the LTP.
at 50vrms output, thd should be less than 0.5%.
Then you have another 6SN7 with each half set up as CF,
with each grid RC coupled to the LTP anodes.
The CF cathodes are directly connected to about a 200H choke
with no gap, and with a CT taken to the bias voltage,
or with two separate windings, each of which is taken to
a separate bias supply, but arranged so the bias points form a
CT since both bias points are bypassed to ground with 470 uF caps,
and the bias impedance is low.

The grid current is supplied from the CF driving positively,
but the low DCR of the choke winding and absense of cap coupling
stops the AB2 operation charging up caps.

The CF bias needs to be set up so the right idle current
flows in the CF.
You need to anticipate the grid current flow, and perhaps a pair of EL84 in
triode
might be needed to cope best with the grid currents.
Series R of perhaps 220 ohms should be used as grid stoppers, and current
limiters,
since Ro from an EL84 triode CF is only 100 ohms,
and one don't want to over drive the opv grids.


So if I can get 300 milliwatts peak out of a single
ECC84 valve (similar to a 6DJ8) concertina stage, I should
be able to make a 3 valve push pull amp with up to 425 watts
of class AB2 power.

That's like employing a flea to wield the whip on a
large plow horse, would not a proper drive tube be better,
like EL84 in triode or such like?

Patrick Turner.

From what I can see the drive requirements aren't all that
savage, G1 has a maximum dissipation of 2 watts.

That isn't much, but you don't wanna pump more than 2 watts in....
It's doubtful that you ever would......

At any rate
it looks like the cathode followers will keep things symmetrical
when the output stage is driven hard. I could probably use
6SN7's for concertina and cathode followers or alternatively
a 12AX7 for concertina and 12AU7 for followers.

It depends on the whole train of tubes a bit.
If you need 50 vrms to drive the opvs, then Vin to the LTP
will have to be about 7vrms and that might need another gain tube,
say 12AU7, with unbupassed Rk for a gain of 7,
so the amp is sensitive to 1vrms.
On the other hand, if you want to use FB,
then V1 needs a lot more gain, and the miller C into
the LTP is about 8 x 3pF, so a paralleled 12AX7 with gain of 70
would do, and you could use say 12 dB of NFB and
need only 0.4 vrms input,
or perhaps 12AY7, or 12AT7, for less gain,
still have 12 dB of FB and still need only 1v input for 300 watts,
which means you sure don't want a preamp.

I have thought of using 2 x 13E1 in PP with CFB from OPT,
and its possible to have 800v anode supply,
very much lower Eg2 supply, and with about
5k RLa-a the po would be about 170 watts AB1,
with no need for AB2 at all, and with about
40 watts of class A.
Max Ia is about 800 mA, so RL could be as low as 4k,
for 200w.

But I know very few folks who need that much po.
But the benefit is that the first 40 watts are as clean as a whistle,
and since only 5 watts is used most days, thd is below
very low, depending on NFB.
But triode amps of 30 watts can still only have 1%,
And rarely ever sound bad, and at 5 watts are not much worse than
some huge amp using the same amount of NFB.
The larger amp does tend to sound effortless,
since there is little colouration added by distortions.

I'm just finishing the tests for a pair of SE amps using
4 x EH 6CA7 with 12.5% cfb from the OPT.
Its much harder to keep the thd low, even with NFB,
but so far the thd is 0.025% at 3 watts/5 ohms, which is as good as other
pp amps of 35 watts, with only 0.25% at 20 watts, but 1.8% at 36 watts.
I'm using 10 dB of opt cfb and 8 dB of global NFB.

Regardless of what you build, if the thd levels are low where you listen,
and you have tubes in class A, and not too much FB,
it should sound very healthy.

Amps vary their thd spectral content with regard to load, and
SE pentode amps like mine are no exception,
and if RL is a high value, the phase of the opv 2H becomes opposite
to what it is when RL is low.
In PP amps, which become more class AB as load is lowered,
the variations in RL should have some different effect on
thd and imd, depending on the speaker used, and different speakers
all have different impedance, so the thd/imd products will vary
from speaker to speaker, so we should at least try to keep
some of these artifacts low to minimise colouration.
So when we see thd figures for an R loaded amp, it has little
meaning unless we could see what the thd profiles were
for say 4, 5,6 ,8, and 12 ohms.
Having said that, many triode PP and SET amps with not a stitch
of NFB sound quite wonderful, regardless of possibly horrible measurements.
The experience of SE teaches one that there are no universally valid
pre-conceptions about how it will sound.
I found with my latest SE amp that if I biased the tubes with rated
dissipations at idle, the thd was higher at 5 watts than if
I had pd 20 watts instead of 25 watts.
So I am hoping for a sweeter sound with amp capable
of say 30 watts as opposed to one which makes 40 watts.
Many SE folks say the best sound comes from class A with the tubes
right up at full allowable conditions, but not I, and
the tubes' sound benefits from running below the ratings.
If the SE seems gutless, use more tubes, but still have each one
set up to work easy, not under duress.
Whether this idea translates into PP circuits is a moot point,
but I suspect even the 4X150A used in PP
would also perhaps sound better if configured to make an easy
200 watts rather than 425 watts.

Patrick Turner.





Mark Harriss

Hi Pat,
So from what you and Tim have said and from what I
can read about it, a concertina only, is possible but doesn't leave
much margin for any feedback or input sensitivity, any loading is
to be avoided.

My basic spec is now 1250v B+ into a 7.2K plate to plate
primary load for 310 watts output. The longtailed phase splitter is
your choice for the extra gain it would give over a concertina?.

Maybe I could have a cascode then concertina then cathode
follower and get the 48V peak (not RMS) for the 4X150 I need
from the concertina at a high B+ and biased filament. I could
use a pair of 6GW8's using the triode for the concertina and the
pentodes strapped as triode for the cathode followers.

With a cathode feedback winding on the OPT is the
value you get, say 12.5% part of the whole winding Z or is it in
addition to the winding?. I suppose either way if I don't use it I
can wire it back in on the anode side of things if it's part of the total
or leave it unconnected if it's in addition to the total P-P Z.

At this stage I need to wind a 7.2K P-P, 300W OPT with separate
screen winding and cathode feedback winding: should be fun!!.

At any rate you've given a lot of food for thought, but
I'm going to be digesting it all for a while yet.


regards
Mark Harriss

"Mark Harriss" wrote in message
...
Maybe I could have a cascode then concertina then cathode
follower and get the 48V peak (not RMS) for the 4X150 I need
from the concertina at a high B+ and biased filament. I could
use a pair of 6GW8's using the triode for the concertina and the
pentodes strapped as triode for the cathode followers.

Personally, if I were going to build one of these amps (of course, mine
would be 500W or more, since I have a pair of 4CX250Rs ), I'd do
something like... 6DJ8 CF drive.. (output is biased by the CF's grid like
Fred's Dogzilla), LTP splitter, donno what type, 12AU7 I guess, fed by
either a pentode or a 12AX7 depending on how much more gain I need.
Lesse.. 50V output drive for full output, say 10 gain in the 'AU, that's
5V.. bring that down to 1V, then say .05V open-loop input for full power
output. (Hehe, who needs a phono amp?) That makes an input gain of 100,
easily served by a pentode, but a gain of ~70 from a 6AV6 would work nicely.
If you can think of a triode/pentode whose triode mu = 100, that would work
perfectly. Wrap 23dB of NFB around it to bring input level back to a
non-relatavistic order of magnitude and improve the parameters, and you're
sitting good!
Reminds me of this guy:
http://webpages.charter.net/dawill/t...s/31LQ6_PP.gif

BTW, an early design of mine:
http://webpages.charter.net/dawill/t..._Power_Amp.gif

And probably some more he
http://webpages.charter.net/dawill/t..._Circuits.html

At any rate you've given a lot of food for thought, but
I'm going to be digesting it all for a while yet.

Better grab the Pepto-Bismol :^)

Tim

--
"That's for the courts to decide." - Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

Ahh very interesting, what made you decide on 4X150's
in the first place?, I picked mine up cheaply and they appear to
have identical ratings to the 4CX250 according to RCA, if I settled
for class AB2 I could pull 630W max for a pair.

For the phase splitter I think I'd prefer to have a concertina
as opposed to a LTP even though the there's no gain, because the
performance is better. I'd go for an EF86 pentode for the input but the
noise from a cascode 6DJ8 is much lower and has good gain.


Thanks for the indigestion!!!
Mark Harriss


P.S. the pattern for the cast heatsink is nearly finished in acrylic sheet
looks like a very big flat square motorcycle cylinder head.

"Mark Harriss" wrote in message
...
Ahh very interesting, what made you decide on 4X150's
in the first place?

Got 'em free with a bunch of other odds'n'ends that a generous guy shipped
me, "I'm not going to use my tube collection anymore so might as well give
it to someone who will." )

I picked mine up cheaply and they appear to
have identical ratings to the 4CX250 according to RCA, if I settled
for class AB2 I could pull 630W max for a pair.

Noooooo! The 150 and 250 are plate dissipation respectively. The 250 is
designed for either improved radiator structure, ceramic seals or both,
allowing it 100 more watts in the same size and basic design (AFAIK). The
'CX can operate hotter than a glass type because it's tougher.

For the phase splitter I think I'd prefer to have a concertina
as opposed to a LTP even though the there's no gain, because the
performance is better.

It is..?

I'd go for an EF86 pentode for the input but the
noise from a cascode 6DJ8 is much lower and has good gain.

I'd think a good high-Gm pentode would be equivalent to a cascode...after
all...that's all you're doing with the triodes.

P.S. the pattern for the cast heatsink is nearly finished in acrylic sheet
looks like a very big flat square motorcycle cylinder head.

Don't forget you need cooling on that screen seal..

Tim

--
"That's for the courts to decide." - Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

Mark Harriss wrote:

Hi Pat,
So from what you and Tim have said and from what I
can read about it, a concertina only, is possible but doesn't leave
much margin for any feedback or input sensitivity, any loading is
to be avoided.

My basic spec is now 1250v B+ into a 7.2K plate to plate
primary load for 310 watts output. The longtailed phase splitter is
your choice for the extra gain it would give over a concertina?.

I think yes.



Maybe I could have a cascode then concertina then cathode
follower and get the 48V peak (not RMS) for the 4X150 I need
from the concertina at a high B+ and biased filament. I could
use a pair of 6GW8's using the triode for the concertina and the
pentodes strapped as triode for the cathode followers.

I prefer discrete tubes for each element...



With a cathode feedback winding on the OPT is the
value you get, say 12.5% part of the whole winding Z or is it in
addition to the winding?.

12.5% of the whole winding, so 1,750 anode turns,
and 250 cathode turns.
No need to do CFB with a triode circuit.

I suppose either way if I don't use it I
can wire it back in on the anode side of things if it's part of the total
or leave it unconnected if it's in addition to the total P-P Z.

Always build with flexibility in mind for later.



At this stage I need to wind a 7.2K P-P, 300W OPT with separate
screen winding and cathode feedback winding: should be fun!!.

Be very careful with insulation.
Why a screen winding? why not take the screens to a fixed B+,
and have only CFB.

If the speaker sec sections are wound in two layers each,
then the CFB windings can be placed between these, since
they are earthy, and lower signal voltage, which makes insulation easy
from the anode windings.
5P x 6S sections are called for.



At any rate you've given a lot of food for thought, but
I'm going to be digesting it all for a while yet.

regards
Mark Harriss

Patrick Turner.

Tim Williams wrote:

"Mark Harriss" wrote in message
...
Ahh very interesting, what made you decide on 4X150's
in the first place?

Got 'em free with a bunch of other odds'n'ends that a generous guy shipped
me, "I'm not going to use my tube collection anymore so might as well give
it to someone who will." )

Hey Tim, you lucky bugger!!! I'm almost jealous

I picked mine up cheaply and they appear to
have identical ratings to the 4CX250 according to RCA, if I settled
for class AB2 I could pull 630W max for a pair.

Noooooo! The 150 and 250 are plate dissipation respectively. The 250 is
designed for either improved radiator structure, ceramic seals or both,
allowing it 100 more watts in the same size and basic design (AFAIK). The
'CX can operate hotter than a glass type because it's tougher.

You're absolutely right about the plate dissipation values, I just
made a major gaffe with my RCA datasheets: they are for the 7034 (250 W)
and the 4X150A/D (150W), the 7034 being a replacement 4X150A's.

My RCA datasheet claims you can get 630 watts class AB2 without any
trouble but they must be referring to the 7035's. I'm basing my specs on
the Eimac data which reckons i'll get 425 W from a pair with 85W plate
dissipation. I'll be trying for 310W with 90W on the plate.


For the phase splitter I think I'd prefer to have a concertina
as opposed to a LTP even though the there's no gain, because the
performance is better.

It is..?

It is unless the tail is much much less than the RL, although
you realise I'm just coughing this up from Morgan Jones "Valve
amplifiers book: according to him you can either have a large value
tail resistor, or improve things with a constant current source feeding
the anodes or you can just panel beat things with feedback.


I'd go for an EF86 pentode for the input but the
noise from a cascode 6DJ8 is much lower and has good gain.

I'd think a good high-Gm pentode would be equivalent to a cascode...after
all...that's all you're doing with the triodes.

Yep according to me old mate Morgan, an EF86 is 6 to
14 dB more noisy in pentode mode than it is in triode mode from
partition noise.


P.S. the pattern for the cast heatsink is nearly finished in acrylic
sheet looks like a very big flat square motorcycle cylinder head.

Don't forget you need cooling on that screen seal..

I'll be using commercial sockets with the screen heatsink
plate to try and take care of that, though I may augment them with
some extra copper sheet

Mark Harriss

On Wed, 21 Jan 2004 23:42:01 -0600, the highly esteemed Tim Williams
enlightened us with these pearls of wisdom:

"Mark Harriss" wrote in message
...
Ahh very interesting, what made you decide on 4X150's
in the first place?

Got 'em free with a bunch of other odds'n'ends that a generous guy shipped
me, "I'm not going to use my tube collection anymore so might as well give
it to someone who will." )

I picked mine up cheaply and they appear to have identical ratings to
the 4CX250 according to RCA, if I settled for class AB2 I could pull
630W max for a pair.

Noooooo! The 150 and 250 are plate dissipation respectively. The 250 is
designed for either improved radiator structure, ceramic seals or both,
allowing it 100 more watts in the same size and basic design (AFAIK). The
'CX can operate hotter than a glass type because it's tougher.

The original 4X150 was a metal-glass tube - it used glass as the seals.
Later it was modified to metal-ceramic construction and was given an
increased plate dissipation rating, and thus became the 4X150A. This
tube does indeed have a 250 watt plate dissipation rating. If Mark's
tubes have glass seals (the visual difference between the two is
obvious) then they are indeed limited to 150 watts Pd. However, if
they are truly 'A' versions, constructed with ceramic seals, then their
ratings are equivalent to the 4CX250B.

Here is the old Svetlana page for the 4X150A:
http://www.svetlana.com/docs/products/metal/4X150A.html

--
Greg

--The software said it requires Win2000 or better, so I installed Linux.

Patrick Turner wrote:
Be very careful with insulation.
Why a screen winding? why not take the screens to a fixed B+,
and have only CFB.

I felt if I was going wind my own tranny's I may as
well use ultralinear and CFB windings.


If the speaker sec sections are wound in two layers each,
then the CFB windings can be placed between these, since
they are earthy, and lower signal voltage, which makes insulation easy
from the anode windings.
5P x 6S sections are called for.

So the speaker windings are nearest to the core for the best
magnetic coupling?.


Well I'd better start some serious planning and cogitation
on the info now at hand, draw a few circuits and see what I get.

Mark Harriss

Mark Harriss wrote:

Patrick Turner wrote:
Be very careful with insulation.
Why a screen winding? why not take the screens to a fixed B+,
and have only CFB.

I felt if I was going wind my own tranny's I may as
well use ultralinear and CFB windings.

But there is *no* need for screen windings *and* cfb windings.

I used to have 6550 with UL taps and cfb, but removing the
screen taps and taking the screems to a B+ supply considerably lower than
anode B+ meant that the screen voltaghe was a lot lower, therefore
the grid 1 bias didn't need to be so great, and there was an increase in
gain
and an 8 db or more reduction in thd, so I
have abandoned the idea of any signal applied from the anode circuit.
With about 15% cfb windings, you will boost the required tetrode drive
voltage about 6 dB.
The signal difference voltage applied defacto between screen and cathode
acts
as screen FB as well as the FB applied in the grid circuit.
twe birds, only one shot.
The guys who invented the Quad II method of UL, or distributed load
were I think Hafler and Keroes.
But methinks its the best way to do local fb using the OPT.

Screen windings will complicate the OPT design with more
leakage inductances and stray capacitances, and
before you know it, you got an oscillator, especially
with such tubes you plan to use.





If the speaker sec sections are wound in two layers each,
then the CFB windings can be placed between these, since
they are earthy, and lower signal voltage, which makes insulation easy
from the anode windings.
5P x 6S sections are called for.

So the speaker windings are nearest to the core for the best
magnetic coupling?.

Nope.

Start with speaker windings and end with them
The P windings are between.
Think of about a 150mm stack of 51 mm tongue material.
Or even better, C-cores with about a 32 x 92 window, and a
strip width of 120, and build up of 2 x 32 mm.
Think big for the OPT.
The interleaving makes the location of the core irrelevant.
All the windings will "see" the core at LF,
but at above 20 kHz, its as if the cores are removed,
since its U has fallen to low levels.


Well I'd better start some serious planning and cogitation
on the info now at hand, draw a few circuits and see what I get.

Remember you will need a bigger drive voltage with CFB in the output stage.

Think small power tubes for a driver stage; I use LTP with EL84 to drive
even 4 x 6550. They will drive 12 x 6550 as well, no worries,
in an LTP, with RC coupling from the anode, no need to try for AB2.
Just a 6CG7 as V1 and 2 x EL84 as the driver in LTP is all I need.
only 3 active elements, and the LTP has less thd than a CPI.

Patrick Turner.





Mark Harriss

Hi Greg,

Greg Pierce wrote:
The original 4X150 was a metal-glass tube - it used glass as the seals.
Later it was modified to metal-ceramic construction and was given an
increased plate dissipation rating, and thus became the 4X150A. This
tube does indeed have a 250 watt plate dissipation rating. If Mark's
tubes have glass seals (the visual difference between the two is
obvious) then they are indeed limited to 150 watts Pd. However, if
they are truly 'A' versions, constructed with ceramic seals, then their
ratings are equivalent to the 4CX250B.

Here is the old Svetlana page for the 4X150A:
http://www.svetlana.com/docs/products/metal/4X150A.html


Sorry to be such a pain but................I have glass sealed
Eimac 4X150A and 4X150D's (26.5v fil) as per the photos on the
datasheets.

That neat little Svetlana unit has the identical flat top anode to a
4CX250 so apart from the number it's identical physically to a
4CX250B. My units are distinguishable by the glass seal, hence no
C for ceramic in the part number . Also the anodes are a flat topped
blunt cone.

But but but...I just looked at em again and they are labelled 7034
and 7035 which ARE the 250 watt units...WOOHOO!!!!

Regards
Mark

Hi Pat, Thanks for the information. It was Alan Dower Blumlein
who came up with the idea of "ultralinear" as well as another known as
"Stereo" these days. I don't think he called either by those names. He got
wiped out in WWII in a bomber crash while testing his radar prototype.

So if I used a LTP (Don't know why Morgan dislikes them so much) why
would I go for triode strapped EL84 if I can get the voltage swing with
much smaller triodes?. I mean if I don't need the extra power they provide
what would make them advantageous?.


regards
Mark

Tim Williams wrote:

Personally, if I were going to build one of these amps (of course, mine
would be 500W or more, since I have a pair of 4CX250Rs )

if you take that effort with transmitting tetrodes, it would be better
to use more linear tubes like 4CX350FJ, 4CX600J, 4CX1500B etc.
IM3 with 4X150/4CX250 types is not the best

Peter

Mark Harriss wrote:

Hi Pat, Thanks for the information. It was Alan Dower Blumlein
who came up with the idea of "ultralinear" as well as another known as
"Stereo" these days. I don't think he called either by those names. He got
wiped out in WWII in a bomber crash while testing his radar prototype.

I heard some Oz guys were using UL anyway well before Blumlein.



So if I used a LTP (Don't know why Morgan dislikes them so much) why
would I go for triode strapped EL84 if I can get the voltage swing with
much smaller triodes?. I mean if I don't need the extra power they provide
what would make them advantageous?.

An LTP with EL84 trioded assume you will go for class AB1, not
AB2, which is excessive, no?
If you have cfb in the opt, and you want say 90vrms to each opv grid,
then the LTP with a choke with CT for the DC supply,
abd a CCS tail is hard to beat for linearity and V swing.
Sure a single 6SN7 might do, especially
since the miller C of the cfb output stage is low because there is
no signal on the screens.

Two big tetrodes, and two gutsy triode drivers just seems right to me.
I find the trioded EL84 are good sounding....

I didn't know Morgan didn't like LTPs.
Ah well, one cannot please all the people all the time.

Try Allen Wright. He likes the output stage to be configured as a class
A triode LTP, with CCS tail to common cathodes of the triode output tubes.
He reckons it sounds the best.

So there are lotsa opinions out there.....

Patrick Turner.



regards
Mark

"Mark Harriss" wrote in message
...
You're absolutely right about the plate dissipation values, I just
made a major gaffe with my RCA datasheets: they are for the 7034 (250 W)
and the 4X150A/D (150W), the 7034 being a replacement 4X150A's.

Ok, I see.

It is unless the tail is much much less than the RL, although
you realise I'm just coughing this up from Morgan Jones "Valve
amplifiers book: according to him you can either have a large value
tail resistor, or improve things with a constant current source feeding
the anodes or you can just panel beat things with feedback.

Well I assume you're smert enough to use a CCS :^) Even without, you can
get good performance.
Personally, for this project, a transistor CCS would be best, since it's
mighty hard to find a 100 mu triode + pentode combo tube.

Yep according to me old mate Morgan, an EF86 is 6 to
14 dB more noisy in pentode mode than it is in triode mode from
partition noise.

But gain is at least 20dB higher, thus offsetting the noise by a good bit
(assuming his figures are dBV, and for the 20Hz-20kHz audio band).

Tim

--
"That's for the courts to decide." - Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

Hi Tim,


Tim Williams wrote:
Well I assume you're smert enough to use a CCS :^) Even without, you can
get good performance.
Personally, for this project, a transistor CCS would be best, since it's
mighty hard to find a 100 mu triode + pentode combo tube.

I have some 6GW8's which on paper appear to be very close
to half a 12AX7 and an EL84. A pair would let me try a long tailed
pair or concertina with an EF86 input stage, how does that sound to you Tim?



Yep according to me old mate Morgan, an EF86 is 6 to
14 dB more noisy in pentode mode than it is in triode mode from
partition noise.

But gain is at least 20dB higher, thus offsetting the noise by a good bit
(assuming his figures are dBV, and for the 20Hz-20kHz audio band).

Tim

You are 100% right there, I didn't consider that at all. Looks
like I'm just quoting from the usual "Common Knowledge" again
instead of considering it from all angles.


Regards
Mark

Mark Harriss wrote:

I have some 6GW8's which on paper appear to be very close
to half a 12AX7 and an EL84. A pair would let me try a long tailed
pair or concertina with an EF86 input stage, how does that sound to you
Tim?

I forgot to mention, the EL84-like pentode would be strapped
for triode to drive the output stage. So the lineup would be like this:

EF86 input, 12AX7-like triode, long tailed pair and EL84-like
triode strapped cathode follower driving the 7034/4X150A push pull
stage. Five valves in total

If I use discretes I'd have six valves......hang on!! six more
commonly available and currently manufactured valves!!!!.

Hmmm EF86,12AX7, 2x EL84, 2x 4X150's...how's that sound?


Regards
Mark

Patrick Turner wrote:



Mark Harriss wrote:

Hi Pat, Thanks for the information. It was Alan Dower Blumlein
who came up with the idea of "ultralinear" as well as another known as
"Stereo" these days. I don't think he called either by those names. He
got wiped out in WWII in a bomber crash while testing his radar
prototype.

I heard some Oz guys were using UL anyway well before Blumlein.

Thats right too! wasn't it on the Retrovox website?



Two big tetrodes, and two gutsy triode drivers just seems right to me.
I find the trioded EL84 are good sounding....

How does this sound for a lineup Pat? :
EF86, longtailed 12AX7,2x EL84 (triode) cathode follower, 2x 7034/4X150A's
for output stage with cathode feedback.




I didn't know Morgan didn't like LTPs.
Ah well, one cannot please all the people all the time.

One website I looked at last night compared phase splitter
preferences to religion, I think that's true.



Try Allen Wright. He likes the output stage to be configured as a class
A triode LTP, with CCS tail to common cathodes of the triode output tubes.
He reckons it sounds the best.

So there are lotsa opinions out there.....

Patrick Turner.



regards
Mark

"Mark Harriss" wrote in message
...
I have some 6GW8's which on paper appear to be very close
to half a 12AX7 and an EL84. A pair would let me try a long tailed
pair or concertina with an EF86 input stage, how does that sound to you
Tim?

I'll have to look up 6GW8.. it rings a bell as equiv. to ECL86, which is
like a lower power ECL82 = 6BM8.

So let's see, you have two triodes, two CF drivers and a pentode. You could
use the triodes and pentode for LTP, but that leaves the question of the
approx. 30-35dB gain you still need. If you used a transistor CCS, or even
just a resistor in the LTP, you could use the EF86 for front end in triode
or pentode, although that leaves either too little or too much gain up
front.

A pentode input stage feeding a cathodyne feeding the drivers could work,
given the splitter is fed with a high enough B+ as Pat T. observed. Input
for max. output might still be a bit on the high side, however.

If you want to skip class 2 output (i.e. grid current), you could use the
6GW8 pentodes in triode mode as a LTP, you'll need a beefier pentode, maybe
12BY7 or 6CL6 (or sure, EL84) for CCS, and as that'll net maybe 20 gain, an
EF86 front end should complete the deal with ample gain.

You could also take the "Williamson" route, gain splitter (usually
cathodyne) balanced driver/gain output.

Tim

--
"That's for the courts to decide." - Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

Mark Harriss questioned:

: I mean if I don't need the extra power they provide
: what would make them advantageous?.

Loosely bouncing off this to where the RSGB 3rd ed. is talking about SSB
linear amplifiers, and Choice of Class of Operation, p.322, an approach I like

"Before proceeding to a practical design, the question of whether the
amplifier is to be operated in class AB1 or AB2 must be resolved. An advantage
usually claimed for class AB1 is that, as grid current does not flow at any
time, it is possible to drive the stage with a voltage amplifier incapable of
delivering power, and that a regulated bias supply is un-necessary. There is
little or nothing in the first point; any driver, no matter how heavily
swamped by a resistive load, ought to be able to deliver the fraction of a
watt required to drive the amplifier to full input in class AB2. The second
point is valid, but potentially dangerous. In most cases it is virtually
impossible to guarantee that the grid will not be driven positive from time to
time, and when this does happen the distortion will be excessive. Even if the
intention is to operate an amplifier exclusively in class AB1 it is desirable
to design the driver, input, and bias circuitry as if operation in the
grid-current region were intended. The extra complication in construction will
be a small price to pay for effective insurance against high distortion if the
stage is slightly overdriven."

Radio Society of Great Britain "The Radio Amateur Handbook" 3rd Ed. '61-'62.
__

OK, they are talking about an RF Single SideBand linear amp here - but I
thought it was worth sharing, in the context of designing amps with RF valves,
and possibly worthy of a considered transposition to audio amp driver design.

Ross Matheson wrote:

Mark Harriss questioned:

: I mean if I don't need the extra power they provide
: what would make them advantageous?.

Loosely bouncing off this to where the RSGB 3rd ed. is talking about SSB
linear amplifiers, and Choice of Class of Operation, p.322, an approach I
like

"Before proceeding to a practical design, the question of whether the
amplifier is to be operated in class AB1 or AB2 must be resolved. An
advantage usually claimed for class AB1 is that, as grid current does not
flow at any time, it is possible to drive the stage with a voltage
amplifier incapable of delivering power, and that a regulated bias supply
is un-necessary. There is little or nothing in the first point; any
driver, no matter how heavily swamped by a resistive load, ought to be
able to deliver the fraction of a watt required to drive the amplifier to
full input in class AB2. The second point is valid, but potentially
dangerous. In most cases it is virtually impossible to guarantee that the
grid will not be driven positive from time to time, and when this does
happen the distortion will be excessive. Even if the intention is to
operate an amplifier exclusively in class AB1 it is desirable to design
the driver, input, and bias circuitry as if operation in the grid-current
region were intended. The extra complication in construction will be a
small price to pay for effective insurance against high distortion if the
stage is slightly overdriven."

Radio Society of Great Britain "The Radio Amateur Handbook" 3rd Ed.
'61-'62. __

OK, they are talking about an RF Single SideBand linear amp here - but I
thought it was worth sharing, in the context of designing amps with RF
valves, and possibly worthy of a considered transposition to audio amp
driver design.



Hi Ross, thanks very much for the input, it's good to know there
are sound reasons for that approach of having the cathode followers in
the design. Once I know the reasoning behind the choice, I'm happy to make
it.

Regards
Mark Harriss

Mark Harriss wrote:
Ahh very interesting, what made you decide on 4X150's
in the first place?, I picked mine up cheaply and they appear to
have identical ratings to the 4CX250 according to RCA, if I settled
for class AB2 I could pull 630W max for a pair.


I tried using one for SE a few years ago and did manage to get a little
bit of power out with running the plate at 350v. The driver was a SRPP
12AT7. One tube I always wanted to use for something is a 4-65a but the
specs aren't very good for audio.


Adam

Ross Matheson wrote:
Mark Harriss questioned:

: I mean if I don't need the extra power they provide
: what would make them advantageous?.

Loosely bouncing off this to where the RSGB 3rd ed. is talking about SSB
linear amplifiers, and Choice of Class of Operation, p.322, an approach I like

"Before proceeding to a practical design, the question of whether the
amplifier is to be operated in class AB1 or AB2 must be resolved. An advantage
usually claimed for class AB1 is that, as grid current does not flow at any
time, it is possible to drive the stage with a voltage amplifier incapable of
delivering power, and that a regulated bias supply is un-necessary. There is
little or nothing in the first point; any driver, no matter how heavily
swamped by a resistive load, ought to be able to deliver the fraction of a
watt required to drive the amplifier to full input in class AB2. The second
point is valid, but potentially dangerous. In most cases it is virtually
impossible to guarantee that the grid will not be driven positive from time to
time, and when this does happen the distortion will be excessive. Even if the
intention is to operate an amplifier exclusively in class AB1 it is desirable
to design the driver, input, and bias circuitry as if operation in the
grid-current region were intended. The extra complication in construction will
be a small price to pay for effective insurance against high distortion if the
stage is slightly overdriven."

Radio Society of Great Britain "The Radio Amateur Handbook" 3rd Ed. '61-'62.
__

OK, they are talking about an RF Single SideBand linear amp here - but I
thought it was worth sharing, in the context of designing amps with RF valves,
and possibly worthy of a considered transposition to audio amp driver design.
Hi
I made a 4cx250 amp using them as triodes.
I used a williamson type circuit with 6gu7 {12BH7} 6gu7 6sn7 as cathode
followers. has more than enough power. I used 5K A-A sowter transformer
as I had one. I like the sound of cathode coupled amps as the dynamics
are so much better than any thing else. I used 500 volts on the anodes
of the 4CX250's. Apart from the fans needed to cool these tubes they are
easier to use than a lot of other transmitting tubes as the heater
current is so low and you can use AC. Biult a PP 4E27 pentode amp with
cfb recently, major problem to supply the 5 volts DC due to the high current
Phil

Phil Lawrence wrote:

Hi
I made a 4cx250 amp using them as triodes.
I used a williamson type circuit with 6gu7 {12BH7} 6gu7 6sn7 as cathode
followers. has more than enough power. I used 5K A-A sowter transformer
as I had one. I like the sound of cathode coupled amps as the dynamics
are so much better than any thing else. I used 500 volts on the anodes
of the 4CX250's. Apart from the fans needed to cool these tubes they are
easier to use than a lot of other transmitting tubes as the heater
current is so low and you can use AC. Biult a PP 4E27 pentode amp with
cfb recently, major problem to supply the 5 volts DC due to the high
current Phil


You know I was eyeing off the 6SN7 as I can have up to 250 volts
A-K on it which whould make getting the 96 volt swing I need for drive a
bit easier to get on the phase splitter.

Mark Harriss

Adam Stouffer wrote:

Mark Harriss wrote:
Ahh very interesting, what made you decide on 4X150's
in the first place?, I picked mine up cheaply and they appear to
have identical ratings to the 4CX250 according to RCA, if I settled
for class AB2 I could pull 630W max for a pair.


I tried using one for SE a few years ago and did manage to get a little
bit of power out with running the plate at 350v. The driver was a SRPP
12AT7. One tube I always wanted to use for something is a 4-65a but the
specs aren't very good for audio.


Adam


Hi Adam, I'd be interested in hearing more about your amp
project. Thanks for the 12AT7 idea, I'm looking at a 6SN7 for a phase
splitter because I can have up to 250V A-K . Now looking at the 12AT7
I see I can try 250V on it too.

Regards
Mark

Mark Harriss wrote:

Hi Adam, I'd be interested in hearing more about your amp
project. Thanks for the 12AT7 idea, I'm looking at a 6SN7 for a phase
splitter because I can have up to 250V A-K . Now looking at the 12AT7
I see I can try 250V on it too.

Regards
Mark

Heres the schematic:
http://151.201.223.39:808/~tesla/4x150a%20SE%20amp.gif

It really wasn't much of a project, just me trying to get a tube made
for VHF to play some music


Adam

"Phil Lawrence" wrote in message
u...
I made a 4cx250 amp using them as triodes.
I used a williamson type circuit with 6gu7 {12BH7} 6gu7 6sn7 as cathode
followers.

I believe 'GU is closer to 6CG7/'FQ.

Tim

--
"That's for the courts to decide." - Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

Mark Harriss wrote:
Phil Lawrence wrote:


Hi
I made a 4cx250 amp using them as triodes.
I used a williamson type circuit with 6gu7 {12BH7} 6gu7 6sn7 as cathode
followers. has more than enough power. I used 5K A-A sowter transformer
as I had one. I like the sound of cathode coupled amps as the dynamics
are so much better than any thing else. I used 500 volts on the anodes
of the 4CX250's. Apart from the fans needed to cool these tubes they are
easier to use than a lot of other transmitting tubes as the heater
current is so low and you can use AC. Biult a PP 4E27 pentode amp with
cfb recently, major problem to supply the 5 volts DC due to the high
current Phil



You know I was eyeing off the 6SN7 as I can have up to 250 volts
A-K on it which whould make getting the 96 volt swing I need for drive a
bit easier to get on the phase splitter.

Mark Harriss
Hi
Even better than the 6sn7 the 6BL7 can handle 600 volts as a cathode
follower is capable of driving 811a's with 600 volts on the anode which
gives 100 watts. But I doubt you would need them to drive 4CX250.
Phil

Tim Williams wrote:

"Phil Lawrence" wrote in message
u...

I made a 4cx250 amp using them as triodes.
I used a williamson type circuit with 6gu7 {12BH7} 6gu7 6sn7 as cathode
followers.


I believe 'GU is closer to 6CG7/'FQ.

Tim

--
"That's for the courts to decide." - Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

HI
The 6gu7 is the 6 volt version of the 12BH7 and they are still cheap.
from the RCA manual


6cg7. 6gu7 12bh7
plate:250
Grid -8 -10.5 -10.5
amplication 20 17 17
p resist 7700 5500 5500
trans 2600 3100 3100
plate current 9ma 11.5 11.5
I find there is a very big difference in the drive voltage between the
6cg7 and 6GU7/12BH7 when used as the push pull drivers in the williamson
circuit. the Russian 6N6 { has the same Pin connections} can give even
higher drive voltages. Also I find the slightly lower amplifiction
factor is better for the horns i use 105db
Phil

Tim Williams wrote:

"Mark Harriss" wrote in message
...
You're absolutely right about the plate dissipation values, I just
made a major gaffe with my RCA datasheets: they are for the 7034 (250 W)
and the 4X150A/D (150W), the 7034 being a replacement 4X150A's.

Ok, I see.

It is unless the tail is much much less than the RL, although
you realise I'm just coughing this up from Morgan Jones "Valve
amplifiers book: according to him you can either have a large value
tail resistor, or improve things with a constant current source feeding
the anodes or you can just panel beat things with feedback.

Well I assume you're smert enough to use a CCS :^) Even without, you can
get good performance.
Personally, for this project, a transistor CCS would be best, since it's
mighty hard to find a 100 mu triode + pentode combo tube.

Yep according to me old mate Morgan, an EF86 is 6 to
14 dB more noisy in pentode mode than it is in triode mode from
partition noise.

But gain is at least 20dB higher, thus offsetting the noise by a good bit
(assuming his figures are dBV, and for the 20Hz-20kHz audio band).

But the triode amp is quieter.
Pentodes are OK for later stages in the amp chain
where snr is not so important.

Patrick Turner.



Tim

--
"That's for the courts to decide." - Homer Simpson
Website @ http://webpages.charter.net/dawill/tmoranwms

Mark Harriss wrote:

Patrick Turner wrote:



Mark Harriss wrote:

Hi Pat, Thanks for the information. It was Alan Dower Blumlein
who came up with the idea of "ultralinear" as well as another known as
"Stereo" these days. I don't think he called either by those names. He
got wiped out in WWII in a bomber crash while testing his radar
prototype.

I heard some Oz guys were using UL anyway well before Blumlein.

Thats right too! wasn't it on the Retrovox website?

Two big tetrodes, and two gutsy triode drivers just seems right to me.
I find the trioded EL84 are good sounding....

How does this sound for a lineup Pat? :
EF86, longtailed 12AX7,2x EL84 (triode) cathode follower, 2x 7034/4X150A's
for output stage with cathode feedback.

Nah,
Use 6SN7 or 6CG7 for all input stages and LTP.
Then perhaps EL84 for CF drivers for AB2.
The LTP out put drive voltage is somewhat high,
and I wouldn't use 12AX7.
EF86 and 12AX7 have far too much gain...

Patrick Turner.



I didn't know Morgan didn't like LTPs.
Ah well, one cannot please all the people all the time.

One website I looked at last night compared phase splitter
preferences to religion, I think that's true.


Try Allen Wright. He likes the output stage to be configured as a class
A triode LTP, with CCS tail to common cathodes of the triode output tubes.
He reckons it sounds the best.

So there are lotsa opinions out there.....

Patrick Turner.



regards
Mark

Patrick Turner wrote:



Mark Harriss wrote:

Patrick Turner wrote:



Mark Harriss wrote:

How does this sound for a lineup Pat? :
EF86, longtailed 12AX7,2x EL84 (triode) cathode follower, 2x
7034/4X150A's for output stage with cathode feedback.

Nah,
Use 6SN7 or 6CG7 for all input stages and LTP.
Then perhaps EL84 for CF drivers for AB2.
The LTP out put drive voltage is somewhat high,
and I wouldn't use 12AX7.
EF86 and 12AX7 have far too much gain...



Just prototyped an ECC84 cascode input stage and after that
an EF86 input stage connected to a 12AT7 concertina. The concertina
works fine but the top triode on the ECC84 seems to markedly distort the
waveform when viewed on a CRO. Thats with 190V at the anode, a 39K
anode resistor, 270V supply , -3V on the grid and 64V on the cathode. I get
really nice squarewave up to 3KHz but after that thing gradually go to a
sawtooth (square in) at 55KHz.

The EF86 was noticeable distorted right from the start and
nothing I tried seemed to help. I'll have to play with it tommorrow.


Mark Harriss

Mark Harriss wrote:

Patrick Turner wrote:



Mark Harriss wrote:

Patrick Turner wrote:



Mark Harriss wrote:

How does this sound for a lineup Pat? :
EF86, longtailed 12AX7,2x EL84 (triode) cathode follower, 2x
7034/4X150A's for output stage with cathode feedback.

Nah,
Use 6SN7 or 6CG7 for all input stages and LTP.
Then perhaps EL84 for CF drivers for AB2.
The LTP out put drive voltage is somewhat high,
and I wouldn't use 12AX7.
EF86 and 12AX7 have far too much gain...


Just prototyped an ECC84 cascode input stage and after that
an EF86 input stage connected to a 12AT7 concertina. The concertina
works fine but the top triode on the ECC84 seems to markedly distort the
waveform when viewed on a CRO. Thats with 190V at the anode, a 39K
anode resistor, 270V supply , -3V on the grid and 64V on the cathode. I get
really nice squarewave up to 3KHz but after that thing gradually go to a
sawtooth (square in) at 55KHz.

Cascode triode stages are not good for power amps.
The bottom tube of the pair sees a load = RL of the top tube / gain of the top
tube.
The 2H distortion of each tube is additive.
If RL was 39k, and top triode gain was 16, then the RL sen by the bottom triode
is
2.4k, and thd is very high at the voltages being asked.
I only use cascode stages using a j-fet and triode for a low noise phono stage.

with such tiny input voltages, the thd is minimal.
In a power amp it'd be hopeless,
but such a hybrid cascode is fairly linear beacuse of the
fets high drain resistance, so the top tube effectively has a high
impedance as its cathode load, and is thus subject to
current feedback, far more so than if low impedance triodes were used.



The EF86 was noticeable distorted right from the start and
nothing I tried seemed to help. I'll have to play with it tommorrow.

EF86 has a lot of thd compared to 6SN7 etc.
I have to confess that I have not built an amp with
a pentode signal amp section anywhere.
I have used pentodes strapped as triodes,
or in UL, or with cfb.
And strapping a signal pentode as a triode
often gives us a flawless triode.
I use 6EJ7 frame grid pentode as a triode in the phono stage,
and its U is 60, and a gain of 50 is available with
a 22 k RL, and the Ra is low because Gm is so high.
The gain is about 200 in pentode with 22k,
but way too high, and not linear enough for several volts output.
Triodes are better for all signal work.

RDH4 compares signal triodes with pentodes re thd
and the pentode has less thd than the triode
up to about 10v rms output.
This comparison is very misleading to anyone looking for a general rule.
Their example was with a 6SJ7 either pentode connected,
or triode strapped.
Below 10v output, the pentode had less thd.
Alas there was no comparison between a 1/2 6SN7
and the 6SJ7 pentode.
And the 2H thd of a pentode can be made extremely
low by careful choice of RL.
Below the value of RL which results in very low 2H, the pentode
has 2H like a triode, with pointed negative going anode voltage swings,
and flattened positive going voltage swings.
But above the "ideal" RL, the phase of the thd reverses,
and with high gain and high RL values, the 2H is large.
Even at the ideal RL, the pentode struggles to
have less 3H that a triode, which has very low 3H at smallish signal voltages.
Power tetrodes and pentodes also display the same character, in comparison with

real triodes, or triodes made from tet/pentodes.

I have spent a lot of time over driver amps, and the
line up I use is the simplest for gain, linearity and bw,
so moderate nfb can be used to get 0.2% thd into 4 ohms
near full po with class Ab1.
a paralleled 12AT7 driving a pair of 2 x 6CG7/6SN7 also paralleled in an LTP
with CCS
is hard to beat.
I found LTPs using cascode halves limited voltage output and had lots of
3H, and paraphase is too bw limited, and twice the thd
of a simple LTP, etc.
I have been tempted to making class AB2 amps,
but then you need a CF driver stage, and simplicity dissappears,
and I like to plan all my power needs around Ab1 operation.
The extra bit of power from a tetrode Ab2 isn't worth the bother.
With 807/6L6 set up as triodes, 30 watts Ab2 instead of the usual
16 watts Ab1 in triode is available.
I'd much rather use 4 x 807 and settle for
32 watts, but at about 1/4 the thd of the Ab2 amp,
and with a much wider class A sweetzone.
In tetrode mode with moderate B+ of say 400v,
the extra Ab2 po is negligible because the anode voltage swing
intersects the 0V line below or at the knee of the 0V grid bias Ra line.
3 or 4% thd is possible in class AB1, say 35 watts thius way.
I have a Brimar tube data book with an 807 recipe for
80 watts tetrode Ab2, with Ea = 600v, Eg2 = 300v,
and the thd is 14% at 80 watts,
and very little class A.
It is public address amp territory, not hi-fi.
Such equipment was used as pa amps at railway stations,
and the train announcements were totally incomprehendable.
(Part of the reason train travel is unattractive.)
The 300 watt amps I made have an enormous %
of class A operation, and thd is negligible,
because there is 300 watts of plate dissipation
in the 12 x EH6550 used.

So what does your load line analysis
reveal about thd with a pair of 4X150D without any nfb?
What is the thd at 1,3,5,and 10 watts of output?

The use of CFB for the OPT tends to make the tubes have a
more benign cut off behaviour like triodes and UL,
and hence less switching nasties of the tetrode/pentode.
Switching in RF circuits makes little difference to RF tank circuits
which are tuned to one F, thus excluding RF harmonics, but AF
amps are not tuned, so the harmonics of crossover thd remain
in their uglyness, and in hi-fi amps its been said that
its best to not have to rely totally on nfb to remove thd.
Its also been said that once thd has been lowered sufficiently
by class A design, then the FB is only needed to reduce
Ro of the amp to get a reasonable damping factor of
say 10 or less, which means Ro = about 0.5 ohms is OK
with most speakers.

Food for thought,

Patrick Turner.







Mark Harriss

Mark Harriss wrote:

: Ross Matheson wrote:
:
: Mark Harriss questioned:
:
: : I mean if I don't need the extra power they provide
: : what would make them advantageous?.
:
: Loosely bouncing off this to where the RSGB 3rd ed. is talking about SSB
: linear amplifiers, and Choice of Class of Operation, p.322, an approach I
: like
:
: "Before proceeding to a practical design, the question of whether the
: amplifier is to be operated in class AB1 or AB2 must be resolved. An
: advantage usually claimed for class AB1 is that, as grid current does not
: flow at any time, it is possible to drive the stage with a voltage
: amplifier incapable of delivering power, and that a regulated bias supply
: is un-necessary. There is little or nothing in the first point; any
: driver, no matter how heavily swamped by a resistive load, ought to be
: able to deliver the fraction of a watt required to drive the amplifier to
: full input in class AB2. The second point is valid, but potentially
: dangerous. In most cases it is virtually impossible to guarantee that the
: grid will not be driven positive from time to time, and when this does
: happen the distortion will be excessive. Even if the intention is to
: operate an amplifier exclusively in class AB1 it is desirable to design
: the driver, input, and bias circuitry as if operation in the grid-current
: region were intended. The extra complication in construction will be a
: small price to pay for effective insurance against high distortion if the
: stage is slightly overdriven."
:
: Radio Society of Great Britain "The Radio Amateur Handbook" 3rd Ed.
: '61-'62. __
:
: OK, they are talking about an RF Single SideBand linear amp here - but I
: thought it was worth sharing, in the context of designing amps with RF
: valves, and possibly worthy of a considered transposition to audio amp
: driver design.
:
:
:
: Hi Ross, thanks very much for the input, it's good to know there
: are sound reasons for that approach of having the cathode followers in
: the design. Once I know the reasoning behind the choice, I'm happy to make
: it.
:
: Regards
: Mark Harriss

Be aware that there are potential red herrings lurking in them thar waters...
They are talking about RF linear amps, with, as PT points out, a tuned LC tank
circuit as the anode load. The picture is not as simple as in a class C RF amp
operating (and modulated) at one frequency, where the harmonics will be
filtered out - the problem with SSB is that the valve must be operated in a
linear region else it will act as a mixer and IM products will fall within the
acceptance range of the load, right on the carrier and lower sideband
frequencies which so much trouble has been taken to suppress earlier.

RF power amplifiers are generally operated in AB1, or AB2 with suitable
valves, for the sake of efficiency - class A may be used in a driver stage.

An audio amplifier is unlikely to be overdriven to the extent mentioned
earlier that might occur in an RF amplifier, so transposing those concerns to
an AF amp might not be so necessary. Sorry if my note was irrelevant.
As PT points out, aiming for much class A operation would be better.

One point that stands out in the RSGB discussion is the care needed with
screen supplies and the need to ensure that tetrodes are not driven so hard
that the min. anode voltage swing falls below the screen supply voltage ...

I've just found this paragraph, too, which I hope is not too off-putting:- (

"The belief seems to be fairly widespread that aligned radial-grid tetrodes
such as the 4-125A and 4-250A have characteristics which make them especially
suitable for SSB linear amplifier service. Apart from the fact that they may
be operated at higher frequencies than the cheaper 813, they have no
outstanding advantages. In fact, they are rather more susceptible both to non
linearity and to screen damage if overdriven" (p.330)

They mention the 6146 as unusually susceptible to screen damage through
overunning too, earlier, and claim that "unlike the 807 (it) has been designed
expressly for AB1 operation, and the minimal extra output obtained by driving
it into the grid-current region is not worth the effort involved".

So I suppose taking advantage of the massive plate dissipation of your valves
and running as much class A as possible, as PT says, is the best way to go.

Best of luck!

RdM

Hi Pat,
I'll have to try a LTP next or maybe a hedge circuit
next and test the performance, although a suitable HT supply
is next on my list at the moment.

The cascode was straight out of 'Valve Amplifiers"
and the pentode circuit was from Svetlana:
http://www.svetlana.com/graphics/TB/no.23fig1.jpg

Both the pentode and the cascode showed HF dropoff
at 20 KHz but I'll try them at a higher B+ and different anode R's.
I don't think my CRO is causing any loading as I could get a nice
squarewave at the anode of the first triode. The distortion is seen
on the output of respective gain stage capacitor, where triangles
become sines and squares become sawtooths at 20 KHz. After
the concertina things seems to improve a bit but I still see HF
dropoff on squares and bow legged triangle waves.

I've yet to try more than plotting a few loadlines for the 4X150.
I'll leave that til last while I play around with the front end a bit
more. If you are interested I can post waveforms, circuits and
whatever limited data I have recorded so far.


Mark Harriss