Hi RATS,

I'm seeing that combining a grounded cathode gain stage in front of a
split load phase splitter in one preamp tube is a common and solid
performing package.
I was investigating the option of current sourcing either or both with
extra preamp tubes when it seemed to me that possibly a good option
could be to use one or both triodes of a separate tube as a single
common current source to both stages in the other tube.

Function or folly?

Wessel

Wessel Dirksen wrote:

Hi RATS,

I'm seeing that combining a grounded cathode gain stage in front of a
split load phase splitter in one preamp tube is a common and solid
performing package.
I was investigating the option of current sourcing either or both with
extra preamp tubes when it seemed to me that possibly a good option
could be to use one or both triodes of a separate tube as a single
common current source to both stages in the other tube.

Function or folly?

There is a benefit to loading the input gain triode of most
power amps with a current source.
The CS need not be extraordunarily high, but once over
say 20 x Ra, the low signal linearity approaches as good as it gets.
So the easiest trick is to make the input gain stage a mu follower,
or load it with a pentode as a CCS, or use a PNP transistor as a CCS.

The split load concertina phase inverter, CPI, cannot have CCS loads
because
one needs equal output voltages of opposite amplitude so
the loads must be resistive so that the same current flow through
anode load, tube, and cathode load will generate the wanted equal voltages

if RLa = RLk.

CPI are very linear stages.
If the triode is a 12AT7 there might be 22k for RLa and RLk,
and for -/+10 v out from a and k there is a total of 20v acrosss a to k,
and the total load is 44 k.
The rquired Vg-k needed to swing a voltage across 44 k is
Vout / gain.
Gain for all tubes = u x RL / ( RL + Ra )
= 50 x 44k / ( 44k + 20k )
= 34 for 12AT7 approximately.

The Vgk will be +0.589v, and since Vk = +10v, Vg = +10.589v.
Half the total output voltage is +10v, and it appears at k
as a feedback voltage. Gain reduction factor due to the series current NFB

= closed loop gain / open loop gain
= ( 20 / 10.589 ) / 34 = 1/18 = 0.055.

Therefore if the thd was say 2% 2H at 20v with 44k all in the plate
circuit,
then the thd with 22k in both plate and cathode circuits
will be 2% x 0.055 = 0.11%, 2H.

This amount of thd in a CPI will occur when the output
stage is producing a lot of power and the thd in the output stage is
probably 1%, about 10 times the level of the CPI.
In a Williamson, the CPI would only ever have to produce
about +/- 2vrms output, and its thd is therefore maybe 0.01%,
and utterly negligible.


So the resistance loaded CPI makes a negligible contribution
to the thd of the amp, and its a pretty blameless stage, rather like a
cathode follower.

A blameless input config for where you have just a gain
triode and CPI only as a driver amp for where the drive voltage is low,
as in the case of a pair of EL84,
then a 12AT7 with CCS, direct coupled to another 12AT7 CPI,
with 22k per a and k , will give the lowest thd.
The input tube has to make sightly more voltage than applied to each
output tube grid.

The CCS loading of the input gain tube or use of mu follower
will reduce thd perhaps by 15 db over a plain resistance load.
That is because when the load is a high value, there is very little signal
current change,
and triodes are amazingly linear if we prevent them from
incurring a large current swing as well as a large voltage swing.
In load line terms, when the load line tends to becoming horizontal,
it tends to become a CCS, which is a truly horizontal load line,
and if you plot the horizontal loadline, the difference between
Ea min and Ea max for an equal grid voltage swing will be minimized.
The input tube with a CCS and direct coupling may have a gain of 50
with 12AT7.
12AX7 with both halves in parallel and with CCS of about 1.5 mA is also
very usable,
and slightly more linear, and warmer sounding than 12AT7, which should
sound
faster.
12AX7 in U foll or CCS loaded will have gain maybe up to 95.

Does this answer your question??

Patrick Turner.







Wessel

In article , Patrick Turner
wrote:

The split load concertina phase inverter, CPI, cannot have CCS loads
because
one needs equal output voltages of opposite amplitude so
the loads must be resistive so that the same current flow through
anode load, tube, and cathode load will generate the wanted equal voltages

if RLa = RLk.

I don't see how that excludes using current sources in the cathode and
plate circuits of a "CPI", the following grid resistors become the load
and as long as they are matched I would think everything should be OK.


CPI are very linear stages.

Until the stage they are driving starts drawing grid current, then the
"CPI" quickly becomes very nonlinear.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

John Byrns wrote:

In article , Patrick Turner
wrote:

The split load concertina phase inverter, CPI, cannot have CCS loads
because
one needs equal output voltages of opposite amplitude so
the loads must be resistive so that the same current flow through
anode load, tube, and cathode load will generate the wanted equal voltages

if RLa = RLk.

I don't see how that excludes using current sources in the cathode and
plate circuits of a "CPI", the following grid resistors become the load
and as long as they are matched I would think everything should be OK.

Two issues.
The CCS would have to be closely matched current sources.
What happens when anode current exceeeds cathode current?

But you are right, you could at least theoretically have CCS to a and k,
and rely on the cap coupled Rg bias R following to set the RLs,
which could then be say 100k each, so open loop CPI tube gain is high,
and so thd will be extremely low.
personally, i don't think its necessary to go to all that trouble with a CPI.




CPI are very linear stages.

Until the stage they are driving starts drawing grid current, then the
"CPI" quickly becomes very nonlinear.

Yes, but in a Williamson, the CPI drives the balanced amp with a low voltage,
so this balanced amp will overload after the output stage clips.
CPI overload is only likely when it drives an output stage without the balanced
amp.
The positive going OPV g1 voltages with grid current will affect the
anode circuit much more than the cathode circuit, since the Rout of anode
and cathode when considerd individually are very different.
But while a and k both see the same resistive load the balance stays put.
One output tube sometimes wears out sooner than the other in a guitar amp with
CPI.
But most guitar amps have the LTP driver, which oveloads more symetrically.

Oveload in hi-fi amps such as the Williamson isn't a worry;
nobody should be driving the amps that hard; the idea is to keep well away from
overload.

Patrick Turner.


Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

From a practical point of view, using a CCS to load the first (gain) stage
is a good idea. The split-load phase splitter is as good and simple as it
can be if You use it like in a Williamson to provide a small signal voltage
to a following differential amplifier, therefore I wouldn't go thru any
extra trouble.
Directly driving an output tube from a concertina is not a very good idea,
unless these latter are, say, EL84s or similar (or the concertina is made
out of a very "robust" triode, say a 6DR7, 6BX7...).
I'd say that it is impossible to have one single CCS feeding the 2 triodes
in parallel. It will keep the SUM of the 2 currents at a constant level,
which does not mean that the whole will work.
I'd consider using a pentode (say a 6AU6 or an EF86) as a CCS for the 1st
half of an asymmetric double tube (say 6DR7 or a trioded 6BM8/6GV8...), DC
coupled to the 2nd half (the 7W one) of the same tube used as the
concertina. The other way, the pentode of a ECF80 can current-source the
triode of the same tube, but then there are few single triodes to be used as
a concertina, 6J5 maybe, or trioded pentodes.

Ciao

Fabio


"Wessel Dirksen" ha scritto nel messaggio
m...
Hi RATS,

I'm seeing that combining a grounded cathode gain stage in front of a
split load phase splitter in one preamp tube is a common and solid
performing package.
I was investigating the option of current sourcing either or both with
extra preamp tubes when it seemed to me that possibly a good option
could be to use one or both triodes of a separate tube as a single
common current source to both stages in the other tube.

Function or folly?

Wessel

Patrick Turner wrote in message ...
John Byrns wrote:

In article , Patrick Turner
wrote:

The split load concertina phase inverter, CPI, cannot have CCS loads
because
one needs equal output voltages of opposite amplitude so
the loads must be resistive so that the same current flow through
anode load, tube, and cathode load will generate the wanted equal voltages

if RLa = RLk.

I don't see how that excludes using current sources in the cathode and
plate circuits of a "CPI", the following grid resistors become the load
and as long as they are matched I would think everything should be OK.

Two issues.
The CCS would have to be closely matched current sources.
What happens when anode current exceeeds cathode current?

But you are right, you could at least theoretically have CCS to a and k,
and rely on the cap coupled Rg bias R following to set the RLs,
which could then be say 100k each, so open loop CPI tube gain is high,
and so thd will be extremely low.
personally, i don't think its necessary to go to all that trouble with a CPI.

This is what I was referring to as far as the CPI goes. I have seen a
circuit where this was done but there it was specified that the
inheirant balance in the CPI is not guaranteed and thus would require
playing around with the load on the Rk leg, ie making RLk unequal to
restore proper balance.




CPI are very linear stages.

Until the stage they are driving starts drawing grid current, then the
"CPI" quickly becomes very nonlinear.

Yes, but in a Williamson, the CPI drives the balanced amp with a low voltage,
so this balanced amp will overload after the output stage clips.
CPI overload is only likely when it drives an output stage without the balanced
amp.
The positive going OPV g1 voltages with grid current will affect the
anode circuit much more than the cathode circuit, since the Rout of anode
and cathode when considerd individually are very different.
But while a and k both see the same resistive load the balance stays put.
One output tube sometimes wears out sooner than the other in a guitar amp with
CPI.
But most guitar amps have the LTP driver, which oveloads more symetrically.

Oveload in hi-fi amps such as the Williamson isn't a worry;
nobody should be driving the amps that hard; the idea is to keep well away from
overload.

Patrick Turner.

In response to other input. Patrick I appreciate the way you expound
your responses. I save all this stuff and get more out of it that way
for reference.




Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

"Fabio Berutti" wrote in message ...
From a practical point of view, using a CCS to load the first (gain) stage
is a good idea. The split-load phase splitter is as good and simple as it
can be if You use it like in a Williamson to provide a small signal voltage
to a following differential amplifier, therefore I wouldn't go thru any
extra trouble.
Directly driving an output tube from a concertina is not a very good idea,
unless these latter are, say, EL84s or similar (or the concertina is made
out of a very "robust" triode, say a 6DR7, 6BX7...).

This I am also curious about. I have heard this advice more often
because of the limited voltage swing of a CPI. Yet what I am doing now
is playing around with a Vanderveen design where there is a single 14x
gain ECC81 grounded cathode input stage with ONF which is Cap coupled
to a ECC82 with 14x gain grounded cathode stage cascaded into a CPI
which directly drives a 4xEL-34 output in triode mode. It sounds
pretty good. However, it sounds even better without the ECC81. But
with too little gain and a wish for ONF, I thought about using the
ECC81 as the final line stage on its own, getting the first half to
produce +/- 28x gain on its own with the ONF and bingo. The CCS was an
idea to use the other tube socket that is already in the neighborhood.

Forgetting about the CCS for a minute, I believe I've got a
theoretical winner for the single ECC81 option with 350V. As Patrick
points out 22K resistors come out good for Rk and Ra on the CPI part.
Using 33K for the cathode resistor and 330 ohm for the cathode
resistor for the grounded cathode stage seems to fit. I've not thought
about how the ONF will affect it, if at all. Still maybe I should
check . . . Will I blow anything up with this config?

I can tell I'm getting myself into a bit of ignorant and potentially
destructive mischief but I love it.

I'd say that it is impossible to have one single CCS feeding the 2 triodes
in parallel. It will keep the SUM of the 2 currents at a constant level,
which does not mean that the whole will work.
I'd consider using a pentode (say a 6AU6 or an EF86) as a CCS for the 1st
half of an asymmetric double tube (say 6DR7 or a trioded 6BM8/6GV8...), DC
coupled to the 2nd half (the 7W one) of the same tube used as the
concertina. The other way, the pentode of a ECF80 can current-source the
triode of the same tube, but then there are few single triodes to be used as
a concertina, 6J5 maybe, or trioded pentodes.

Ciao

Fabio


"Wessel Dirksen" ha scritto nel messaggio
m...
Hi RATS,

I'm seeing that combining a grounded cathode gain stage in front of a
split load phase splitter in one preamp tube is a common and solid
performing package.
I was investigating the option of current sourcing either or both with
extra preamp tubes when it seemed to me that possibly a good option
could be to use one or both triodes of a separate tube as a single
common current source to both stages in the other tube.

Function or folly?

Wessel