[Fai C]

My friend has sent me a link last night that is about a French winder
offers SE output transformer with no air gap and low flux. However,
this is absolutely a series/direct feed type SE transformer.

http://www.audiodesignguide.com/abb/abb.html

This is the link for the documentations about this particular SE
output transformer called SC-OPT:

http://www.polisois-audio.com/informazioni_eng.html

Looks like this is not a conventional type SE output transformer, but
how can it get any inductance???

[Patrick Turner]

On Mar 17, 5:07*pm, Fai C wrote:
My friend has sent me a link last night that is about a French winder
offers SE output transformer with no air gap and low flux. However,
this is absolutely a series/direct feed type SE transformer.

http://www.audiodesignguide.com/abb/abb.html

This is the link for the documentations about this particular SE
output transformer called SC-OPT:

http://www.polisois-audio.com/informazioni_eng.html

Looks like this is not a conventional type SE output transformer, but
how can it get any inductance???

The above links contain NO amp schematics and NO details of how the
OPTs are wound so what these ppl are trying to do is to baffle
everyone with bull**** and make out they have invented some new way of
building a class A SE amp differently / better / more innovatively
than ever before, and therefore make themselves famous and ride off
into the sunset with bags full of profits.

If there was a way of building an SE OPT without an air gap, it would
have probably been discovered before 1930 when most amps anyobe could
afford were SET types, and damned expensive, so any attempt at
increased tube efficiency and simpler easier lighter better OPTs was
very welcomed. But a gapless OPT for a conventionally used OPT between
B+ and anode is not to be seen mentioned in any books I have ever
read.

Polisois mentions in the second link how he has TWO channels of audio
signals operating through SINGLE OPT core. All constructional details
and test results are entirely missing, so its got to be a trick, ie,
bull****, until proven otherwise.

It is possible to have no Idc flow in an OPT primary winding despite
the coil being in series with tubes and B+ supplies. Its been done for
years in a Circlotron but there are 2 tubes 2 PSUs floating, and the
whole single primary has dc flow in BOTH directions which sum to no DC
flow, which of course does not make sense to anyone. The slightest
imbalance of dc in either tube shows up as a net Dc current. As far as
I know, there is no way to have the anode current flow from one tube
through 2 equal turn primary windings so the magnetizing effect of
each winding opposes each other leaving a core free of any
magnetization. If the Ia anode current is reduced in a signal wave,
you might think you would see a rising voltage across one coil and
falling voltage across the other, and the signal magnetization would
oppose itself, preventing such voltage changes occuring. One might
like to think one could reverse wind one of the coils, and magically
find that you get a voltage at the sec which could power something,
but in books such as RDH4, Tremain's Audio Encyclopedia, there isn't
any mention of magic at all anywhere, just cold hard engineering,
which all bean counters and charlatans try desperately to avoid.

I could be wrong. Lets see who else wants to chime in with NEW
delectable magnetophic delights for audiophiles.

One way to avoid air gapping in SE OPTs is to apply a DC current
through a winding on an OPT such that the amp-turns are the same as
the anode winding. If such Idc is supplied from a constant current
source, then there is no signal loading of such a DC source, and you
can get the predicted SE power from a tube just as anyone might
calculate it, and without distortions due to partial heavy dc
magnetization of the core. Such schemes have been discussed here quite
often over the last 12 years, but I know of nobody who has ever
successfully built such an amp, or tried to sell examples to the
public. One could indeed set up a normal PP class A amp with two OP
tubes, and have one tube arranged with a fixed grid1 voltage, ie, no
signal input, then use a constant current source between its cathode
and 0V, thus causing the effective anode resistance to be virtually an
infinite ohm value. The one remaining tube of the pair would happily
work without being affected by the presence of the lazy companion
tube.

Perhaps Polisois has some not-so-difficult-to-figure-out current
cancelling circuit under the chassis.

Patrick Turner.

[flipper]

On Fri, 16 Mar 2012 23:07:58 -0700 (PDT), Fai C
wrote:

My friend has sent me a link last night that is about a French winder
offers SE output transformer with no air gap and low flux. However,
this is absolutely a series/direct feed type SE transformer.

http://www.audiodesignguide.com/abb/abb.html

This is the link for the documentations about this particular SE
output transformer called SC-OPT:

http://www.polisois-audio.com/informazioni_eng.html

Looks like this is not a conventional type SE output transformer, but
how can it get any inductance???

I imagine it gets inductance "the old fashioned way" with wire turns
around an iron core.

But the basic "idea" seems simple enough. Two primaries wound in
series opposition so the same DC current through both cancels each
other's magnetic field for a net DC flux of 0. The 'tertiary', as they
seem to call it, is then AC bypassed so it's 'out of circuit' for
audio, leaving the varying magnetic field of only one to couple into
the secondary.

How well it works I don't know since, as the saying goes, the devil is
in the details and simple ideas are often plagued by pesky
practicalities

[John L Stewart]

One way to avoid air gapping in SE OPTs is to apply a DC current
through a winding on an OPT such that the amp-turns are the same as
the anode winding. If such Idc is supplied from a constant current
source, then there is no signal loading of such a DC source, and you
can get the predicted SE power from a tube just as anyone might
calculate it, and without distortions due to partial heavy dc
magnetization of the core. Such schemes have been discussed here quite
often over the last 12 years, but I know of nobody who has ever
successfully built such an amp, or tried to sell examples to the
public. One could indeed set up a normal PP class A amp with two OP
tubes, and have one tube arranged with a fixed grid1 voltage, ie, no
signal input, then use a constant current source between its cathode
and 0V, thus causing the effective anode resistance to be virtually an
infinite ohm value. The one remaining tube of the pair would happily
work without being affected by the presence of the lazy companion
tube.

Perhaps Polisois has some not-so-difficult-to-figure-out current
cancelling circuit under the chassis.

Patrick Turner.[/quote]

For the adventuress & the curious here is a cct that in some ways is similar to something proposed by Patrick above. It is driven SE but the OP is PP, so the DC in the OPT is cancelled. The drive required is twice that of a normal SE cct. The choke in the OP cathodes forces equal AC currents in the sides of the PP OPT.

The choke in the simulation is a Hammond of 30H, 595R. The switch Z allows comparison with the choke switched out.

I did quite a few tests about a year ago on one of these things using PP 6V6's running into a Hammond 125E OPT. Driver is a 6AU6. The performance was more than passable.

I was curious why the ancients in the 30's did not use this simple hookup in some of their receivers. Came to the conclusion that many of those receivers connected all the cathodes straigt to the chassis. Used back biasing so I guess production was simplified.

Today we could use a FET in place of the choke.

Cheers, John

[John L Stewart]

Originally Posted by Patrick Turner

On Mar 17, 5:07*pm, Fai C wrote:
My friend has sent me a link last night that is about a French winder
offers SE output transformer with no air gap and low flux. However,
this is absolutely a series/direct feed type SE transformer.

http://www.audiodesignguide.com/abb/abb.html

This is the link for the documentations about this particular SE
output transformer called SC-OPT:

http://www.polisois-audio.com/informazioni_eng.html

Looks like this is not a conventional type SE output transformer, but
how can it get any inductance???

The above links contain NO amp schematics and NO details of how the
OPTs are wound so what these ppl are trying to do is to baffle
everyone with bull**** and make out they have invented some new way of
building a class A SE amp differently / better / more innovatively
than ever before, and therefore make themselves famous and ride off
into the sunset with bags full of profits.

If there was a way of building an SE OPT without an air gap, it would
have probably been discovered before 1930 when most amps anyobe could
afford were SET types, and damned expensive, so any attempt at
increased tube efficiency and simpler easier lighter better OPTs was
very welcomed. But a gapless OPT for a conventionally used OPT between
B+ and anode is not to be seen mentioned in any books I have ever
read.

Polisois mentions in the second link how he has TWO channels of audio
signals operating through SINGLE OPT core. All constructional details
and test results are entirely missing, so its got to be a trick, ie,
bull****, until proven otherwise.

It is possible to have no Idc flow in an OPT primary winding despite
the coil being in series with tubes and B+ supplies. Its been done for
years in a Circlotron but there are 2 tubes 2 PSUs floating, and the
whole single primary has dc flow in BOTH directions which sum to no DC
flow, which of course does not make sense to anyone. The slightest
imbalance of dc in either tube shows up as a net Dc current. As far as
I know, there is no way to have the anode current flow from one tube
through 2 equal turn primary windings so the magnetizing effect of
each winding opposes each other leaving a core free of any
magnetization. If the Ia anode current is reduced in a signal wave,
you might think you would see a rising voltage across one coil and
falling voltage across the other, and the signal magnetization would
oppose itself, preventing such voltage changes occuring. One might
like to think one could reverse wind one of the coils, and magically
find that you get a voltage at the sec which could power something,
but in books such as RDH4, Tremain's Audio Encyclopedia, there isn't
any mention of magic at all anywhere, just cold hard engineering,
which all bean counters and charlatans try desperately to avoid.

I could be wrong. Lets see who else wants to chime in with NEW
delectable magnetophic delights for audiophiles.

One way to avoid air gapping in SE OPTs is to apply a DC current
through a winding on an OPT such that the amp-turns are the same as
the anode winding. If such Idc is supplied from a constant current
source, then there is no signal loading of such a DC source, and you
can get the predicted SE power from a tube just as anyone might
calculate it, and without distortions due to partial heavy dc
magnetization of the core. Such schemes have been discussed here quite
often over the last 12 years, but I know of nobody who has ever
successfully built such an amp, or tried to sell examples to the
public. One could indeed set up a normal PP class A amp with two OP
tubes, and have one tube arranged with a fixed grid1 voltage, ie, no
signal input, then use a constant current source between its cathode
and 0V, thus causing the effective anode resistance to be virtually an
infinite ohm value. The one remaining tube of the pair would happily
work without being affected by the presence of the lazy companion
tube.

Perhaps Polisois has some not-so-difficult-to-figure-out current
cancelling circuit under the chassis.

Patrick Turner.

A bit of digging pulls this up-

http://www.polisois-audio.com/docume..._Is sue_3.pdf

For the schema go to page 8 & all will be revealed. Looks to me like an ordinary PP OPT would do.

Cheers, John

[Roger Jones]

On Mar 17, 9:06*am, flipper wrote:
On Fri, 16 Mar 2012 23:07:58 -0700 (PDT), Fai C
wrote:
(snip)


But the basic "idea" seems simple enough. Two primaries wound in
series opposition so the same DC current through both cancels each
other's magnetic field for a net DC flux of 0. The 'tertiary', as they
seem to call it, is then AC bypassed so it's 'out of circuit' for
audio, leaving the varying magnetic field of only one to couple into
the secondary.

(snip)

At face value this would appear to short out the spare secondary to
signal since the B+ side is zero impedance and so is the presumed AC
bypass cap, thus the whole of the OPT is shorted. You'd need an "near
infinite" impedance current source to do this, and match the DC mA's.
What might I have missed?
Cheers.
Roger

[flipper]

On Sat, 17 Mar 2012 18:02:38 -0700 (PDT), Roger Jones
wrote:

On Mar 17, 9:06*am, flipper wrote:
On Fri, 16 Mar 2012 23:07:58 -0700 (PDT), Fai C
wrote:
(snip)


But the basic "idea" seems simple enough. Two primaries wound in
series opposition so the same DC current through both cancels each
other's magnetic field for a net DC flux of 0. The 'tertiary', as they
seem to call it, is then AC bypassed so it's 'out of circuit' for
audio, leaving the varying magnetic field of only one to couple into
the secondary.

(snip)

At face value this would appear to short out the spare secondary to
signal since the B+ side is zero impedance and so is the presumed AC
bypass cap, thus the whole of the OPT is shorted. You'd need an "near
infinite" impedance current source to do this, and match the DC mA's.
What might I have missed?

Only the opposing primary (tertiary) is AC bypassed, not both.

At AC it looks like a plain old SE OPT.

Cheers.
Roger

[John L Stewart]

Originally Posted by Roger Jones

On Mar 17, 9:06*am, flipper wrote:
On Fri, 16 Mar 2012 23:07:58 -0700 (PDT), Fai C
wrote:
(snip)


But the basic "idea" seems simple enough. Two primaries wound in
series opposition so the same DC current through both cancels each
other's magnetic field for a net DC flux of 0. The 'tertiary', as they
seem to call it, is then AC bypassed so it's 'out of circuit' for
audio, leaving the varying magnetic field of only one to couple into
the secondary.

(snip)

At face value this would appear to short out the spare secondary to
signal since the B+ side is zero impedance and so is the presumed AC
bypass cap, thus the whole of the OPT is shorted. You'd need an "near
infinite" impedance current source to do this, and match the DC mA's.
What might I have missed?
Cheers.
Roger

That series opposed secondary doesn't look good driving 100 microfarads. Have to wonder if anyone actually tried the transformer in a cct.

Perhaps looking further in the references posted we might find a correction, one way or another!

Cheers, JS

[John L Stewart]

Originally Posted by John L Stewart

That series opposed secondary doesn't look good driving 100 microfarads. Have to wonder if anyone actually tried the transformer in a cct.

Perhaps looking further in the references posted we might find a correction, one way or another!

Cheers, JS

For the curious & others, here is more info on split loading, part of something I authored in Glass Audio about 10 years ago. It was in regard to a line driver, this one has an output Zo of 600 ohms. The schemas did not scan well. They are from pencil dwgs. But you get the idea.

Cheers to all, John

Split Loading

Some confusion exists when we try to understand how a split load
transformer should be connected. I know of at least two examples in
current publications were the authors managed to get it wrong.

A comparison with the common pushpull stage will help since it uses a
similar output transformer. Winding polarity is indicated by dots.
Refer to figures six & seven.

First of all we can make the common assumption that current runs downhill.
That will help lots even as it does for those working in the semiconductor
industry. That is were the little arrow on the transistors are going.

In the figures referred to the arrows indicate the commonly assumed
direction of current flow, that being from positive to negative.
In the case of the pushpull circuit, these currents are as equal as possible
& running in opposite directions so that they will cancel most of the dc flux
in the transformer core.

Not so in the split load connections. Flux resulting
from current in the two parts of the transformer winding is additive
& may result in core saturation. We will need a core with a built in air gap,
just as if we were dealing with a single ended amplifier.

Again referring to the pushpull circuit, for a useful output to occur we will
need as nearly as possible equal but opposite voltages at each end of the
transformer primary. Since the grids are driven equal & opposite that
condition will result.

In the split load circuit the same conditions are needed to produce an output.
When the tube is turned on by a positive going signal on its grid, the plate will
go negative while the cathode goes positive in order to follow the grid.
A useful output occurs.

[John Byrns]

In article ,
Roger Jones wrote:

On Mar 17, 9:06*am, flipper wrote:
On Fri, 16 Mar 2012 23:07:58 -0700 (PDT), Fai C
wrote:
(snip)


But the basic "idea" seems simple enough. Two primaries wound in
series opposition so the same DC current through both cancels each
other's magnetic field for a net DC flux of 0. The 'tertiary', as they
seem to call it, is then AC bypassed so it's 'out of circuit' for
audio, leaving the varying magnetic field of only one to couple into
the secondary.

(snip)

At face value this would appear to short out the spare secondary to
signal since the B+ side is zero impedance and so is the presumed AC
bypass cap, thus the whole of the OPT is shorted. You'd need an "near
infinite" impedance current source to do this, and match the DC mA's.
What might I have missed?

I don't think you have missed anything, it looks to me like C2 the 100 uF cap
pretty well shorts out the audio, excepting the effect of the transformer half
primary winding resistance which prevents the capacitor from being a zero
impedance short, letting a little audio sneak through.

--
Regards,

John Byrns

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

[flipper]

On Tue, 20 Mar 2012 12:45:59 -0500, John Byrns
wrote:

In article ,
Roger Jones wrote:

On Mar 17, 9:06*am, flipper wrote:
On Fri, 16 Mar 2012 23:07:58 -0700 (PDT), Fai C
wrote:
(snip)


But the basic "idea" seems simple enough. Two primaries wound in
series opposition so the same DC current through both cancels each
other's magnetic field for a net DC flux of 0. The 'tertiary', as they
seem to call it, is then AC bypassed so it's 'out of circuit' for
audio, leaving the varying magnetic field of only one to couple into
the secondary.

(snip)

At face value this would appear to short out the spare secondary to
signal since the B+ side is zero impedance and so is the presumed AC
bypass cap, thus the whole of the OPT is shorted. You'd need an "near
infinite" impedance current source to do this, and match the DC mA's.
What might I have missed?

I don't think you have missed anything, it looks to me like C2 the 100 uF cap
pretty well shorts out the audio, excepting the effect of the transformer half
primary winding resistance which prevents the capacitor from being a zero
impedance short, letting a little audio sneak through.

Why do you think shorting one winding shorts the other?

[Nick Gorham]

On 21/03/12 05:19, flipper wrote:

Why do you think shorting one winding shorts the other?

Because they are effectivly in parallel

--
Nick

[John L Stewart]

Originally Posted by flipper

On Tue, 20 Mar 2012 12:45:59 -0500, John Byrns
wrote:

In article ,
Roger Jones wrote:

On Mar 17, 9:06*am, flipper wrote:
On Fri, 16 Mar 2012 23:07:58 -0700 (PDT), Fai C
wrote:
(snip)


But the basic "idea" seems simple enough. Two primaries wound in
series opposition so the same DC current through both cancels each
other's magnetic field for a net DC flux of 0. The 'tertiary', as they
seem to call it, is then AC bypassed so it's 'out of circuit' for
audio, leaving the varying magnetic field of only one to couple into
the secondary.

(snip)

At face value this would appear to short out the spare secondary to
signal since the B+ side is zero impedance and so is the presumed AC
bypass cap, thus the whole of the OPT is shorted. You'd need an "near
infinite" impedance current source to do this, and match the DC mA's.
What might I have missed?

I don't think you have missed anything, it looks to me like C2 the 100 uF cap
pretty well shorts out the audio, excepting the effect of the transformer half
primary winding resistance which prevents the capacitor from being a zero
impedance short, letting a little audio sneak through.

Why do you think shorting one winding shorts the other?

Try that on a pwr transformer, then plug it into the line. Should be a convincing demonstration!

Cheers, John

[flipper]

On Wed, 21 Mar 2012 12:32:31 +0000, Nick Gorham
wrote:

On 21/03/12 05:19, flipper wrote:

Why do you think shorting one winding shorts the other?

Because they are effectivly in parallel

I don't see how you come up with that either.

The two primaries are in series opposition so Idc cancels. One is AC
bypassed so it's out of circuit at audio.

I don't see the big mystery.

[Nick Gorham]

On 21/03/12 15:30, flipper wrote:
On Wed, 21 Mar 2012 12:32:31 +0000, Nick
wrote:

On 21/03/12 05:19, flipper wrote:

Why do you think shorting one winding shorts the other?

Because they are effectivly in parallel

I don't see how you come up with that either.

The two primaries are in series opposition so Idc cancels. One is AC
bypassed so it's out of circuit at audio.

I don't see the big mystery.

Well, you say bipassed, I say shorted. Or maybe I should look at the
circuit. But I don't see how it can be simply bipassed without the end
of the winding being effectivly connected together at AC.

I suppose if you had two sets of windings, the first in the anode as
normal, the second in the cathode out of phase with the first, and a CCS
also in series with the winding in the cathode circuit. Then both the
CCS and the winding bipassed with a cap it would do it.

--
Nick

[John L Stewart]

Originally Posted by flipper

On Wed, 21 Mar 2012 12:32:31 +0000, Nick Gorham
wrote:

On 21/03/12 05:19, flipper wrote:

Why do you think shorting one winding shorts the other?

Because they are effectivly in parallel

I don't see how you come up with that either.

The two primaries are in series opposition so Idc cancels. One is AC
bypassed so it's out of circuit at audio.

I don't see the big mystery.

Series opposition of the windings gets you cancellation of the DC component, provided the ampere-turns are equal. Since the IT of each section is equal, any driving AC current will show up as a voltage in the other winding. The cap will short that at audio frequencies in both windings.

Perhaps a correction was made later. But sometimes I wonder if proposed circuits are ever tried in the Real World!

Cheers to all, John

[YD]

Late at night, by candle light, flipper penned this
immortal opus:

On Tue, 20 Mar 2012 12:45:59 -0500, John Byrns
wrote:

In article ,
Roger Jones wrote:

On Mar 17, 9:06*am, flipper wrote:
On Fri, 16 Mar 2012 23:07:58 -0700 (PDT), Fai C
wrote:
(snip)


But the basic "idea" seems simple enough. Two primaries wound in
series opposition so the same DC current through both cancels each
other's magnetic field for a net DC flux of 0. The 'tertiary', as they
seem to call it, is then AC bypassed so it's 'out of circuit' for
audio, leaving the varying magnetic field of only one to couple into
the secondary.

(snip)

At face value this would appear to short out the spare secondary to
signal since the B+ side is zero impedance and so is the presumed AC
bypass cap, thus the whole of the OPT is shorted. You'd need an "near
infinite" impedance current source to do this, and match the DC mA's.
What might I have missed?

I don't think you have missed anything, it looks to me like C2 the 100 uF cap
pretty well shorts out the audio, excepting the effect of the transformer half
primary winding resistance which prevents the capacitor from being a zero
impedance short, letting a little audio sneak through.

Why do you think shorting one winding shorts the other?

Because they're magnetically coupled, just like in any old
transformer.

- YD.
--
Remove HAT if replying by mail.

[flipper]

On Thu, 22 Mar 2012 00:50:44 +0000, John L Stewart
wrote:


flipper;954369 Wrote:
On Wed, 21 Mar 2012 12:32:31 +0000, Nick Gorham
wrote:
-
On 21/03/12 05:19, flipper wrote:
-
Why do you think shorting one winding shorts the other?-

Because they are effectivly in parallel-

I don't see how you come up with that either.

The two primaries are in series opposition so Idc cancels. One is AC
bypassed so it's out of circuit at audio.

I don't see the big mystery.

Series opposition of the windings gets you cancellation of the DC
component, provided the ampere-turns are equal. Since the IT of each
section is equal, any driving AC current will show up as a voltage in
the other winding. The cap will short that at audio frequencies in both
windings.

I guess that means everyone has decided to simply ignore he claims his
design has "tight" DC coupling but "loose" AC coupling, apparently due
to whatever that 'flux escape' thing is.


Perhaps a correction was made later. But sometimes I wonder if proposed
circuits are ever tried in the Real World!

Assuming one is curious 'enough' it would be easy to find out because
he sells the transformers.


Cheers to all, John


+-------------------------------------------------------------------+
+-------------------------------------------------------------------+

[flipper]

On Thu, 22 Mar 2012 20:41:57 -0300, YD wrote:

Late at night, by candle light, flipper penned this
immortal opus:

On Tue, 20 Mar 2012 12:45:59 -0500, John Byrns
wrote:

In article ,
Roger Jones wrote:

On Mar 17, 9:06*am, flipper wrote:
On Fri, 16 Mar 2012 23:07:58 -0700 (PDT), Fai C
wrote:
(snip)


But the basic "idea" seems simple enough. Two primaries wound in
series opposition so the same DC current through both cancels each
other's magnetic field for a net DC flux of 0. The 'tertiary', as they
seem to call it, is then AC bypassed so it's 'out of circuit' for
audio, leaving the varying magnetic field of only one to couple into
the secondary.

(snip)

At face value this would appear to short out the spare secondary to
signal since the B+ side is zero impedance and so is the presumed AC
bypass cap, thus the whole of the OPT is shorted. You'd need an "near
infinite" impedance current source to do this, and match the DC mA's.
What might I have missed?

I don't think you have missed anything, it looks to me like C2 the 100 uF cap
pretty well shorts out the audio, excepting the effect of the transformer half
primary winding resistance which prevents the capacitor from being a zero
impedance short, letting a little audio sneak through.

Why do you think shorting one winding shorts the other?

Because they're magnetically coupled, just like in any old
transformer.

That would be good logic if it really were "just like in any old
transformer" but he claims the design is not "just like in any old
transformer."

[flipper]

On Wed, 21 Mar 2012 16:23:49 +0000, Nick Gorham
wrote:

On 21/03/12 15:30, flipper wrote:
On Wed, 21 Mar 2012 12:32:31 +0000, Nick
wrote:

On 21/03/12 05:19, flipper wrote:

Why do you think shorting one winding shorts the other?

Because they are effectivly in parallel

I don't see how you come up with that either.

The two primaries are in series opposition so Idc cancels. One is AC
bypassed so it's out of circuit at audio.

I don't see the big mystery.

Well, you say bipassed, I say shorted. Or maybe I should look at the
circuit. But I don't see how it can be simply bipassed without the end
of the winding being effectivly connected together at AC.

The "end of the winding" is. It's the magnetic path he claims isn't;
the 'loose coupling' and his 'flux escape'.


I suppose if you had two sets of windings, the first in the anode as
normal, the second in the cathode out of phase with the first, and a CCS
also in series with the winding in the cathode circuit. Then both the
CCS and the winding bipassed with a cap it would do it.