[Patrick Turner]

On Aug 20, 5:19*pm, "Paul D. Spiegel" wrote:
flipper wrote:
Does anyone have experience with the Crowhurst "Twin Coupled"
amplifier design, or variants? I've got the original articles so what
I'm looking for is real life experience, tips, comments,
'improvements', etc.

I was happy to see your post. The Twin Coupled is an intriguing design
that is rarely used.

I have built a Twin Coupled amp and I'm very pleased with the results. *
It is now my main stereo amp. *However, my version is somewhat different,
as I will explain.

I originally built a PP cathode follower amp. *I thought it was a shame
that all that high voltage was going to waste and then I remembered that
crazy Crowhurst design with another OT on the plates in addition to the
cathodes. *So I tried it and after fiddling with it for awhile I got some
good measured and sonic results.

Crowhurst's design runs the output tubes as AB pentodes. *I like the
clever way that the screens are attached to the opposite tube plates so
that they will "swing" with the cathodes to ensure pentode operation. *He
also uses global feedback as is common with PP pentode amps to get the
output impedance down. *This design looks a little like the Macintosh
amps, except with a second transformer instead of a bifilar winding for
the cathodes.

For updates to the original Crowhurst concept and some construction hints
you should check out the articles by John Stewart in August 2004 and
September 2005 of _Glass_Audio_.

I did a few things different with my amp. *Starting from the output, I
regulated the screens at a fixed voltage. *With both the plates and
cathodes swinging, I don't know what to call this mode of operation, but
the tubes seem to behave as triodes this way. *As with some other PP
triode amps I've built, I added a constant current souce to the junction
of the output tube cathodes. *With the Twin Coupled configuration this is
the center tap of the cathode OT. *This is a trade off. *The CCS forces
the amp into class A operations and reduces the output power, but makes
the response very linear and clean sounding.

As with CF output stages you need a huge driver voltage swing. *I find the
positive feedback bootstrap approach inelegant, so I use a diff-amp 6BL7
and a Lundahl 1667/5ma center tapped inductor intended for this use. *(I
suppose I could also have used a small OT and terminated the secondary
with a resistor to get the right plate load.) *With the Lundahl and a 450v
plate supply I measured 1kHz sign wave at up to 440 Vrms before clipping, *
That's 1250 v peak-to-peak!

I used output transformers from a Dynaco ST70, supplemented with two
clones from Triode Electronics. *I started with EL34's at 450v on the
plates. *I later upgrade to KT90's at 550v for even more power. *I know. *
It's crazy.

Some measurements with a 1kHz sine wave and 4 ohm resistive load:
Cathode Follower *EL34's : 4.6W at 0.7 ohms Zout
Twin Coupled EL34's * * *: 14.8W at 0.4 ohms Zout
Twin Coupled KT90's * * *: 25W at 0.4 ohms Zout

Low Zout (for a tube amp) with no global feedback. *A damping factor of
~10 keeps the voltage consistent for most speaker systems. *I like the way
this amp sounds with my single driver 8" Hempcones.

I hope this gives you some inspiration. *Have fun with your project!

- Paul

I posted last night on the subject but I think my post went to who I
replied to rather than to the group.
I'm finding it not quite so easy to post via Google groups rather than
via the driect feed from Usenet my ISP used to give me, but won't damn
well give to me anymore.

Anyway, I hope I've aimed my post right this time. I don't know how to
make a copy of what I sent out last night.
But basically, the idea of "unity gain output stages" ah la McIntosh
in principle have been done in a variety if ways by other makers such
as Paravicini's EAR509 where he used equal turns for cathode and anode
windings and then had caps across the windings where equal signal
voltage exists to shunt the leakage inductance rather than to try to
tight couple such windings by the very difficult method using bifilar
turns of wire with extra thick insulation against dc arcs.

Crowshurst and others are crusing around the McI idea.

I happen to think that the Circlotron is a better idea than the McI.
You only need one winding which is at 0V potential and the turns are
half of the total required for both anode and cathode windings of the
McI, or all of the turns in a conventional PP ranny with just one
anode winding.

I've never bothered with the McI or Circlotron or Crowshurst idea
because the drive voltage needed is too high. So I've settled for
cathode winding FB between 12.5% and 25% of the total OPT primary
turns, ie, rather like Quad-II, but with higher CFB %.

But say someone has a Dynaco ST70 which makes 35W/ch class AB into the
5k anode to anode load.

Should they be interested in fidelity from class A working, they could
connect their 8 ohm speakers to the 4 ohm outlet, and the RL a-a
becomes 10k, and the power is nearly all class A.

But they coulod go further to make a Dybaco clean up its act.

What is needed is a choke of about 80H with a CT and no air gap. This
can be a 25mm stack of GOSS with 25mm tongue, then just fill the
window up with 0.35mm dia wire. But there must be a CT, which is
grounded.
Each cathode of each output tube is taken to the ends of the choke.
The screens are disconnected from the UL taps on the original 5k OPT,
and fed from B+ at the original OPT CT, but via a 4k7 and 100uF filter
cap, so Eg2 is a little lower than Ea, and Eg2 is well filtered of the
hum at the OPT CT.

Now each output tube cathode is connected to the UL tap on the OPT
originally meant for the screen of the *other* tube.
This is done using a 470uF x 450V cap, with plus+ to the OPT screen
tap. So it looks like the cathodes are cross coupled, and this is
wanted so signal currents generated by each output tube flow around
one 1/2 of the OPT primary *plus* 43% of the other half. 43% is near
to 3/7 of the 1/2 P winding.

So when working you would see the following signal voltages across the
OPT and V1 and V2,
+140V at V1 anode, +60V at V2 cathode, -60V at V1 cathode and -140V at
V2 anode.

Now you have 200V across each tube, so its equivalent to 400V anode to
anode in a conventional amp.

But the OPT has 280V between each anode, and this is 25 times the
secondary speaker winding voltage if meant to get a 5k:8 load match.
But we have 400V anode to anode if we include the voltages happening
at the cathodes so the new connection doubles the impedance matching,
so the load the tubes see is 10k anode to anode rather than the 5k
Dynaco intended.

Now you have 60V at each cathode, the Vg-k must be about 15V to
produce the 200V across each tube, so you'd need a drive signal of 75V
at each grid and this is easier to do than 150V if you have a McI type
of circuit.

There is a huge drop in Rout of the amp due to the large amount of
effective CFB, equal to 30%. This is plenty, and no need to go the
whole hog and 50% as in the case of the McI, Circlotron, etc.

Any old amp with UL taps less than 43% would be very suitable for this
mod; Manley Labs Snappers would be much improved because their 25% UL
taps get to do some real benefits to the tube working, and the load
value becomes higher and the class A% becomes greater etc, etc.

Of course you can make a bigger choke and cross couple cathodes to the
anode connection of the other tube.
The choke is designed so its Bmax is not in excess of 1.2Tesla at
15Hz. It will then have plenty of inductance which won't shunt the
load badly at LF. The choke needed for the connection using screen
taps also needs to obey the rule for saturation. For the cathode to
screen tap connection one could also use a pair of constant current
sinks taken down to a -Vdc that is more than the peak sgnal voltage
expected at the cathodes.
This would stabilise the Idc of each tube and provide individual
biasing, and you would not need a choke, ( or the second OPT like
Crowshurst.)

Obviously, there are several ways to connect up PP output tubes....

Patrick Turner.

[John Byrns]

In article ,
flipper wrote:

On Mon, 21 Sep 2009 23:37:43 -0700 (PDT), Patrick Turner
wrote:

On Aug 20, 5:19*pm, "Paul D. Spiegel" wrote:
flipper wrote:
Does anyone have experience with the Crowhurst "Twin Coupled"
amplifier design, or variants? I've got the original articles so what
I'm looking for is real life experience, tips, comments,
'improvements', etc.

I was happy to see your post. The Twin Coupled is an intriguing design
that is rarely used.

I have built a Twin Coupled amp and I'm very pleased with the results. *
It is now my main stereo amp. *However, my version is somewhat different,
as I will explain.

I originally built a PP cathode follower amp. *I thought it was a shame
that all that high voltage was going to waste and then I remembered that
crazy Crowhurst design with another OT on the plates in addition to the
cathodes. *So I tried it and after fiddling with it for awhile I got some
good measured and sonic results.

Crowhurst's design runs the output tubes as AB pentodes. *I like the
clever way that the screens are attached to the opposite tube plates so
that they will "swing" with the cathodes to ensure pentode operation. *He
also uses global feedback as is common with PP pentode amps to get the
output impedance down. *This design looks a little like the Macintosh
amps, except with a second transformer instead of a bifilar winding for
the cathodes.

For updates to the original Crowhurst concept and some construction hints
you should check out the articles by John Stewart in August 2004 and
September 2005 of _Glass_Audio_.

I did a few things different with my amp. *Starting from the output, I
regulated the screens at a fixed voltage. *With both the plates and
cathodes swinging, I don't know what to call this mode of operation, but
the tubes seem to behave as triodes this way. *As with some other PP
triode amps I've built, I added a constant current souce to the junction
of the output tube cathodes. *With the Twin Coupled configuration this is
the center tap of the cathode OT. *This is a trade off. *The CCS forces
the amp into class A operations and reduces the output power, but makes
the response very linear and clean sounding.

As with CF output stages you need a huge driver voltage swing. *I find the
positive feedback bootstrap approach inelegant, so I use a diff-amp 6BL7
and a Lundahl 1667/5ma center tapped inductor intended for this use. *(I
suppose I could also have used a small OT and terminated the secondary
with a resistor to get the right plate load.) *With the Lundahl and a 450v
plate supply I measured 1kHz sign wave at up to 440 Vrms before clipping, *
That's 1250 v peak-to-peak!

I used output transformers from a Dynaco ST70, supplemented with two
clones from Triode Electronics. *I started with EL34's at 450v on the
plates. *I later upgrade to KT90's at 550v for even more power. *I know. *
It's crazy.

Some measurements with a 1kHz sine wave and 4 ohm resistive load:
Cathode Follower *EL34's : 4.6W at 0.7 ohms Zout
Twin Coupled EL34's * * *: 14.8W at 0.4 ohms Zout
Twin Coupled KT90's * * *: 25W at 0.4 ohms Zout

Low Zout (for a tube amp) with no global feedback. *A damping factor of
~10 keeps the voltage consistent for most speaker systems. *I like the way
this amp sounds with my single driver 8" Hempcones.

I hope this gives you some inspiration. *Have fun with your project!

- Paul

I posted last night on the subject but I think my post went to who I
replied to rather than to the group.
I'm finding it not quite so easy to post via Google groups rather than
via the driect feed from Usenet my ISP used to give me, but won't damn
well give to me anymore.

Anyway, I hope I've aimed my post right this time. I don't know how to
make a copy of what I sent out last night.
But basically, the idea of "unity gain output stages" ah la McIntosh

The McIntosh is not unity "gain," it's unity "coupled," and the point
is for the bifilar winding.

in principle have been done in a variety if ways by other makers such
as Paravicini's EAR509 where he used equal turns for cathode and anode
windings and then had caps across the windings where equal signal
voltage exists to shunt the leakage inductance rather than to try to
tight couple such windings by the very difficult method using bifilar
turns of wire with extra thick insulation against dc arcs.

Crowshurst and others are crusing around the McI idea.

I happen to think that the Circlotron is a better idea than the McI.
You only need one winding which is at 0V potential and the turns are
half of the total required for both anode and cathode windings of the
McI, or all of the turns in a conventional PP ranny with just one
anode winding.

I've never bothered with the McI or Circlotron or Crowshurst idea
because the drive voltage needed is too high. So I've settled for
cathode winding FB between 12.5% and 25% of the total OPT primary
turns, ie, rather like Quad-II, but with higher CFB %.

Despite appearances the McIntosh isn't true CFB (past maybe some
residual). The output tubes float between the windings and the boot
straps take care of that (on the drivers).

This argument, assuming it is valid, would seem to apply only to the
later McIntosh amps, the original McIntosh design didn't use
bootstrapped drivers IIRC.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Patrick Turner]

On Sep 22, 5:26*pm, flipper wrote:
On Mon, 21 Sep 2009 23:37:43 -0700 (PDT), Patrick Turner


snip,

The McIntosh is not unity "gain," it's unity "coupled," and the point
is for the bifilar winding.

To my mind, the McIntosh and other circuits like it all have 50% of
the Va-k signal fed back in series with the grid input voltage.

Calling it a unity gain circuit is a conventional way of describing it
but it is bull****, because you have a total of just over twice the
input voltage appearing across the tube, and gain isn't unity, or less
than unity like a cathode follower, but more like nearly 2.0 if the
load valie is high enough. In fact, the load for each tube in an McI
circuit is rather low, and so open loop gain is low and the 50% of
output sugnalapplied via the cathode is a rather low amount of series
voltage negative feedback.

snip,

I've never bothered with the McI or Circlotron or Crowshurst idea
because the drive voltage needed is too high. So I've settled for
cathode winding FB between 12.5% and 25% of the total OPT primary
turns, ie, rather like Quad-II, but with higher CFB %.

Despite appearances the McIntosh isn't true CFB (past maybe some
residual). The output tubes float between the windings and the boot
straps take care of that (on the drivers).

Huh? The McI looks to me like 50% CFB. Like a concertina phase
inverter, except that the anode and cathode loads are magnetically
coupled.

The output tubes don't float in the McI OP stage.

The bootstrapping of the driver LTP anode loads has nothing to do with
the OP tube CFB.

The bootstrapping gives rise to some positive FB from output stage to
driver but because the OP stage gain is so low and because the Ra of
the triode LTP drivers is so low, the amount of PFB is small. There is
LTP driver boost to gain and reduction of its own THD by means of the
effective raising of the effective anode load via the bootstrapping,
to the very slight increase to THD via the PFB is offset by the
benefits of the bootstrapping.

As a purist, I prefer to make a driver which does not rely on
bootstrapping. There is less re-cycling of error signal corrections.

See my website for details from pages on my 300W amps about how to
make a fabulous PP driver circuit using CT choke plus resistance loads
for EL84 strapped as triodes.

Patrick Turner.




But say someone has a Dynaco ST70 which makes 35W/ch class AB into the
5k anode to anode load.

Should they be interested in fidelity from class A working, they could
connect their 8 ohm speakers to the 4 ohm outlet, and the RL a-a
becomes 10k, and the power is nearly all class A.

But they coulod go further to make a Dynaco clean up its act.

What is needed is a choke of about 80H with a CT and no air gap. This
can be a 25mm stack of GOSS with 25mm tongue, then just fill the
window up with 0.35mm dia wire. But there must be a CT, which is
grounded.
Each cathode of each output tube is taken to the ends of the choke.
The screens are disconnected from the UL taps on the original 5k OPT,
and fed from B+ at the original OPT CT, but via a 4k7 and 100uF filter
cap, so Eg2 is a little lower than Ea, and Eg2 is well filtered of the
hum at the OPT CT.

Now each output tube cathode is connected to the UL tap on the OPT
originally meant for the screen of the *other* tube.
This is done using a 470uF x 450V cap, with plus+ to the OPT screen
tap. *So it looks like the cathodes are cross coupled, and this is
wanted so signal currents generated by each output tube flow around
one 1/2 of the OPT primary *plus* 43% of the other half. 43% is near
to 3/7 of the 1/2 P winding.

So when working you would see the following signal voltages across the
OPT and V1 and V2,
+140V at V1 anode, +60V at V2 cathode, -60V at V1 cathode and -140V at
V2 anode.

Now you have 200V across each tube, so its equivalent to 400V anode to
anode in a conventional amp.

But the OPT has 280V between each anode, and this is 25 times the
secondary speaker winding voltage if meant to get a 5k:8 load match.
But we have 400V anode to anode if we include the voltages happening
at the cathodes so the new connection doubles the impedance matching,
so the load the tubes see is 10k anode to anode rather than the 5k
Dynaco intended.

Now you have 60V at each cathode, the Vg-k must be about 15V to
produce the 200V across each tube, so you'd need a drive signal of 75V
at each grid and this is easier to do than 150V if you have a McI type
of circuit.

There is a huge drop in Rout of the amp due to the large amount of
effective CFB, equal to 30%. This is plenty, and no need to go the
whole hog and 50% as in the case of the McI, Circlotron, etc.

Any old amp with UL taps less than 43% would be very suitable for this
mod; Manley Labs Snappers would be much improved because their 25% UL
taps get to do some real benefits to the tube working, and the load
value becomes higher and the class A% becomes greater etc, etc.

Of course you can make a bigger choke and cross couple cathodes to the
anode connection of the other tube.
The choke is designed so its Bmax is not in excess of 1.2Tesla at
15Hz. It will then have plenty of inductance which won't shunt the
load badly at LF. The choke needed for the connection using screen
taps also needs to obey the rule for saturation. For the cathode to
screen tap connection one could also use a pair of constant current
sinks taken down to a -Vdc that is more than the peak sgnal voltage
expected at the cathodes.
This would stabilise the Idc of each tube and provide individual
biasing, and you would not need a choke, ( or the second OPT like
Crowshurst.)

Obviously, there are several ways to connect up PP output tubes....

Patrick Turner.- Hide quoted text -

- Show quoted text -- Hide quoted text -

- Show quoted text -- Hide quoted text -

- Show quoted text -

[John Byrns]

In article ,
flipper wrote:

On Tue, 22 Sep 2009 07:37:07 -0500, John Byrns
wrote:

In article ,
flipper wrote:

On Mon, 21 Sep 2009 23:37:43 -0700 (PDT), Patrick Turner
wrote:

On Aug 20, 5:19*pm, "Paul D. Spiegel" wrote:
flipper wrote:
Does anyone have experience with the Crowhurst "Twin Coupled"
amplifier design, or variants? I've got the original articles so what
I'm looking for is real life experience, tips, comments,
'improvements', etc.

I was happy to see your post. The Twin Coupled is an intriguing design
that is rarely used.

I have built a Twin Coupled amp and I'm very pleased with the results.
*
It is now my main stereo amp. *However, my version is somewhat
different,
as I will explain.

I originally built a PP cathode follower amp. *I thought it was a shame
that all that high voltage was going to waste and then I remembered
that
crazy Crowhurst design with another OT on the plates in addition to the
cathodes. *So I tried it and after fiddling with it for awhile I got
some
good measured and sonic results.

Crowhurst's design runs the output tubes as AB pentodes. *I like the
clever way that the screens are attached to the opposite tube plates so
that they will "swing" with the cathodes to ensure pentode operation.
*He
also uses global feedback as is common with PP pentode amps to get the
output impedance down. *This design looks a little like the Macintosh
amps, except with a second transformer instead of a bifilar winding for
the cathodes.

For updates to the original Crowhurst concept and some construction
hints
you should check out the articles by John Stewart in August 2004 and
September 2005 of _Glass_Audio_.

I did a few things different with my amp. *Starting from the output, I
regulated the screens at a fixed voltage. *With both the plates and
cathodes swinging, I don't know what to call this mode of operation,
but
the tubes seem to behave as triodes this way. *As with some other PP
triode amps I've built, I added a constant current souce to the
junction
of the output tube cathodes. *With the Twin Coupled configuration this
is
the center tap of the cathode OT. *This is a trade off. *The CCS forces
the amp into class A operations and reduces the output power, but makes
the response very linear and clean sounding.

As with CF output stages you need a huge driver voltage swing. *I find
the
positive feedback bootstrap approach inelegant, so I use a diff-amp
6BL7
and a Lundahl 1667/5ma center tapped inductor intended for this use.
*(I
suppose I could also have used a small OT and terminated the secondary
with a resistor to get the right plate load.) *With the Lundahl and a
450v
plate supply I measured 1kHz sign wave at up to 440 Vrms before
clipping, *
That's 1250 v peak-to-peak!

I used output transformers from a Dynaco ST70, supplemented with two
clones from Triode Electronics. *I started with EL34's at 450v on the
plates. *I later upgrade to KT90's at 550v for even more power. *I
know. *
It's crazy.

Some measurements with a 1kHz sine wave and 4 ohm resistive load:
Cathode Follower *EL34's : 4.6W at 0.7 ohms Zout
Twin Coupled EL34's * * *: 14.8W at 0.4 ohms Zout
Twin Coupled KT90's * * *: 25W at 0.4 ohms Zout

Low Zout (for a tube amp) with no global feedback. *A damping factor of
~10 keeps the voltage consistent for most speaker systems. *I like the
way
this amp sounds with my single driver 8" Hempcones.

I hope this gives you some inspiration. *Have fun with your project!

- Paul

I posted last night on the subject but I think my post went to who I
replied to rather than to the group.
I'm finding it not quite so easy to post via Google groups rather than
via the driect feed from Usenet my ISP used to give me, but won't damn
well give to me anymore.

Anyway, I hope I've aimed my post right this time. I don't know how to
make a copy of what I sent out last night.
But basically, the idea of "unity gain output stages" ah la McIntosh

The McIntosh is not unity "gain," it's unity "coupled," and the point
is for the bifilar winding.

in principle have been done in a variety if ways by other makers such
as Paravicini's EAR509 where he used equal turns for cathode and anode
windings and then had caps across the windings where equal signal
voltage exists to shunt the leakage inductance rather than to try to
tight couple such windings by the very difficult method using bifilar
turns of wire with extra thick insulation against dc arcs.

Crowshurst and others are crusing around the McI idea.

I happen to think that the Circlotron is a better idea than the McI.
You only need one winding which is at 0V potential and the turns are
half of the total required for both anode and cathode windings of the
McI, or all of the turns in a conventional PP ranny with just one
anode winding.

I've never bothered with the McI or Circlotron or Crowshurst idea
because the drive voltage needed is too high. So I've settled for
cathode winding FB between 12.5% and 25% of the total OPT primary
turns, ie, rather like Quad-II, but with higher CFB %.

Despite appearances the McIntosh isn't true CFB (past maybe some
residual). The output tubes float between the windings and the boot
straps take care of that (on the drivers).

This argument, assuming it is valid, would seem to apply only to the
later McIntosh amps, the original McIntosh design didn't use
bootstrapped drivers IIRC.

I'd need a model number to look at because all the schematics I've got
do.

The model number is the McIntosh 50W-1, the schematic is in the RDH4.
There is also a model 50W-2 which I assume uses a similar circuit, and
lower power models 20W-1 &20W-2 using 6V6 output tubes.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[John Byrns]

In article ,
John Byrns wrote:

In article ,
flipper wrote:

On Tue, 22 Sep 2009 07:37:07 -0500, John Byrns
wrote:

In article ,
flipper wrote:

On Mon, 21 Sep 2009 23:37:43 -0700 (PDT), Patrick Turner
wrote:

On Aug 20, 5:19*pm, "Paul D. Spiegel" wrote:
flipper wrote:
Does anyone have experience with the Crowhurst "Twin Coupled"
amplifier design, or variants? I've got the original articles so
what
I'm looking for is real life experience, tips, comments,
'improvements', etc.

I was happy to see your post. The Twin Coupled is an intriguing
design
that is rarely used.

I have built a Twin Coupled amp and I'm very pleased with the
results.
*
It is now my main stereo amp. *However, my version is somewhat
different,
as I will explain.

I originally built a PP cathode follower amp. *I thought it was a
shame
that all that high voltage was going to waste and then I remembered
that
crazy Crowhurst design with another OT on the plates in addition to
the
cathodes. *So I tried it and after fiddling with it for awhile I got
some
good measured and sonic results.

Crowhurst's design runs the output tubes as AB pentodes. *I like the
clever way that the screens are attached to the opposite tube plates
so
that they will "swing" with the cathodes to ensure pentode operation.
*He
also uses global feedback as is common with PP pentode amps to get
the
output impedance down. *This design looks a little like the Macintosh
amps, except with a second transformer instead of a bifilar winding
for
the cathodes.

For updates to the original Crowhurst concept and some construction
hints
you should check out the articles by John Stewart in August 2004 and
September 2005 of _Glass_Audio_.

I did a few things different with my amp. *Starting from the output,
I
regulated the screens at a fixed voltage. *With both the plates and
cathodes swinging, I don't know what to call this mode of operation,
but
the tubes seem to behave as triodes this way. *As with some other PP
triode amps I've built, I added a constant current souce to the
junction
of the output tube cathodes. *With the Twin Coupled configuration
this
is
the center tap of the cathode OT. *This is a trade off. *The CCS
forces
the amp into class A operations and reduces the output power, but
makes
the response very linear and clean sounding.

As with CF output stages you need a huge driver voltage swing. *I
find
the
positive feedback bootstrap approach inelegant, so I use a diff-amp
6BL7
and a Lundahl 1667/5ma center tapped inductor intended for this use.
*(I
suppose I could also have used a small OT and terminated the
secondary
with a resistor to get the right plate load.) *With the Lundahl and a
450v
plate supply I measured 1kHz sign wave at up to 440 Vrms before
clipping, *
That's 1250 v peak-to-peak!

I used output transformers from a Dynaco ST70, supplemented with two
clones from Triode Electronics. *I started with EL34's at 450v on the
plates. *I later upgrade to KT90's at 550v for even more power. *I
know. *
It's crazy.

Some measurements with a 1kHz sine wave and 4 ohm resistive load:
Cathode Follower *EL34's : 4.6W at 0.7 ohms Zout
Twin Coupled EL34's * * *: 14.8W at 0.4 ohms Zout
Twin Coupled KT90's * * *: 25W at 0.4 ohms Zout

Low Zout (for a tube amp) with no global feedback. *A damping factor
of
~10 keeps the voltage consistent for most speaker systems. *I like
the
way
this amp sounds with my single driver 8" Hempcones.

I hope this gives you some inspiration. *Have fun with your project!

- Paul

I posted last night on the subject but I think my post went to who I
replied to rather than to the group.
I'm finding it not quite so easy to post via Google groups rather than
via the driect feed from Usenet my ISP used to give me, but won't damn
well give to me anymore.

Anyway, I hope I've aimed my post right this time. I don't know how to
make a copy of what I sent out last night.
But basically, the idea of "unity gain output stages" ah la McIntosh

The McIntosh is not unity "gain," it's unity "coupled," and the point
is for the bifilar winding.

in principle have been done in a variety if ways by other makers such
as Paravicini's EAR509 where he used equal turns for cathode and anode
windings and then had caps across the windings where equal signal
voltage exists to shunt the leakage inductance rather than to try to
tight couple such windings by the very difficult method using bifilar
turns of wire with extra thick insulation against dc arcs.

Crowshurst and others are crusing around the McI idea.

I happen to think that the Circlotron is a better idea than the McI.
You only need one winding which is at 0V potential and the turns are
half of the total required for both anode and cathode windings of the
McI, or all of the turns in a conventional PP ranny with just one
anode winding.

I've never bothered with the McI or Circlotron or Crowshurst idea
because the drive voltage needed is too high. So I've settled for
cathode winding FB between 12.5% and 25% of the total OPT primary
turns, ie, rather like Quad-II, but with higher CFB %.

Despite appearances the McIntosh isn't true CFB (past maybe some
residual). The output tubes float between the windings and the boot
straps take care of that (on the drivers).

This argument, assuming it is valid, would seem to apply only to the
later McIntosh amps, the original McIntosh design didn't use
bootstrapped drivers IIRC.

I'd need a model number to look at because all the schematics I've got
do.

The model number is the McIntosh 50W-1, the schematic is in the RDH4.
There is also a model 50W-2 which I assume uses a similar circuit, and
lower power models 20W-1 &20W-2 using 6V6 output tubes.

The McIntosh 50W-1 schematic is also available on the web he
http://www.triodeel.com/images/mac50w1p.jpg
Looks like the verbiage is in Spanish though.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/