Output classes A and AB
 

In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Eeyore wrote:

Patrick Turner wrote:

One cannot have distortion cancelling by one tube cancelling that in
another when one is cut off.

THANK YOU !

Basics do matter.

Indeed they do, but neither Patrick, myself, or anyone else is correct
on every issue. In this case Patrick has vigorously asserted that this
view, which he holds in common with you, is true, but he has failed to
even attempt an argument that might demonstrate its truth. Patrick is
an extremely skilled and talented fellow in the practical aspects of
tube amp design and construction, but he has a very limited
understanding of what is going on behind the scenes in the theory of
tube amp operation.

I think my website might indicate that your are not quite right about my
levels of understanding.

Can you point out where on your website it explains why "One cannot have
distortion cancelling by one tube cancelling that in another when one is
cut off"?

I admit I do not understand this notion that even order distortion can't
be canceled when one tube is cut off, and I would really appreciate an
explanation from someone that understands why cancellation should fail
to occur?

Why I don't understand this is because even order distortion is canceled
in class B amplifiers, where one or the other tube is cutoff over almost
the entire cycle, yet cancellation still occurs! I am having trouble
reconciling this with your claim that cancelation can't occur. Also the
related observation that there is no even order distortion in a sine
wave which has had both the tops and bottoms of the sine wave
symmetrically clipped off.

Please explain.


Regards,

John Byrns

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

Output classes A and AB
 

"John Byrns" wrote in message
...
In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Eeyore wrote:

Patrick Turner wrote:

One cannot have distortion cancelling by one tube cancelling that
in
another when one is cut off.

THANK YOU !

Basics do matter.

Indeed they do, but neither Patrick, myself, or anyone else is correct
on every issue. In this case Patrick has vigorously asserted that this
view, which he holds in common with you, is true, but he has failed to
even attempt an argument that might demonstrate its truth. Patrick is
an extremely skilled and talented fellow in the practical aspects of
tube amp design and construction, but he has a very limited
understanding of what is going on behind the scenes in the theory of
tube amp operation.

I think my website might indicate that your are not quite right about my
levels of understanding.

Can you point out where on your website it explains why "One cannot have
distortion cancelling by one tube cancelling that in another when one is
cut off"?

I admit I do not understand this notion that even order distortion can't
be canceled when one tube is cut off, and I would really appreciate an
explanation from someone that understands why cancellation should fail
to occur?

Why I don't understand this is because even order distortion is canceled
in class B amplifiers, where one or the other tube is cutoff over almost
the entire cycle, yet cancellation still occurs! I am having trouble
reconciling this with your claim that cancelation can't occur. Also the
related observation that there is no even order distortion in a sine
wave which has had both the tops and bottoms of the sine wave
symmetrically clipped off.

Please explain.


Regards,

John Byrns

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

John:

I think that the problem that Patrick may be having with this concept is
more one of semantics than anything else. It's hard to intuitively view
"cancellation" of even harmonic terms if both tubes are not conducting at
the same time. Let's view it with a very small amount of math thrown in.
Sometimes intuition does not take one far enough.

Take a pure Class B amplifier. Each tube or transistor conducts for 180
degrees (forget about any crossover effects). The current in each half of
the OPT primary, neglecting any magnetizing current is a half sine. Do a
Fourier expansion on this half sine and it's full of harmonics...all even
ones. But as long as the two tubes are well matched, the two half primary
currents are of identical shape and add in the OPT secondary to produce a
full sine, symmetrical about its zero voltage axis.
The even harmonics are gone. Is this cancellation? Well I guess you might
call it that.

Basically any waveform symmetrical about its zero voltage axis will contain
no even harmonics...even a symmetrical square wave. Similarly, if we
symmetrically clip the tops off both the positive and negative half cycles
of our sine we do not see any even harmonic content either for the same
reason. The ideal push-pull output stage using matched devices is not going
to produce even order distortion regardless of its operating class, whether
it be Class A, Class AB or Class B.

If in any doubt, see pages 300-301 of RDH4, or any university level math
text covering Fourier analysis.

Best Regards : Doug Bannard, P. Eng.

Output classes A and AB
 

On Oct 28, 2:29 am, "Doug Bannard" wrote:
"John Byrns" wrote in message

...





In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Eeyore wrote:

Patrick Turner wrote:

One cannot have distortion cancelling by one tube cancelling that
in
another when one is cut off.

THANK YOU !

Basics do matter.

Indeed they do, but neither Patrick, myself, or anyone else is correct
on every issue. In this case Patrick has vigorously asserted that this
view, which he holds in common with you, is true, but he has failed to
even attempt an argument that might demonstrate its truth. Patrick is
an extremely skilled and talented fellow in the practical aspects of
tube amp design and construction, but he has a very limited
understanding of what is going on behind the scenes in the theory of
tube amp operation.

I think my website might indicate that your are not quite right about my
levels of understanding.

Can you point out where on your website it explains why "One cannot have
distortion cancelling by one tube cancelling that in another when one is
cut off"?

I admit I do not understand this notion that even order distortion can't
be canceled when one tube is cut off, and I would really appreciate an
explanation from someone that understands why cancellation should fail
to occur?

Why I don't understand this is because even order distortion is canceled
in class B amplifiers, where one or the other tube is cutoff over almost
the entire cycle, yet cancellation still occurs! I am having trouble
reconciling this with your claim that cancelation can't occur. Also the
related observation that there is no even order distortion in a sine
wave which has had both the tops and bottoms of the sine wave
symmetrically clipped off.

Please explain.

Regards,

John Byrns

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

John:

I think that the problem that Patrick may be having with this concept is
more one of semantics than anything else. It's hard to intuitively view
"cancellation" of even harmonic terms if both tubes are not conducting at
the same time. Let's view it with a very small amount of math thrown in.
Sometimes intuition does not take one far enough.

Take a pure Class B amplifier. Each tube or transistor conducts for 180
degrees (forget about any crossover effects). The current in each half of
the OPT primary, neglecting any magnetizing current is a half sine. Do a
Fourier expansion on this half sine and it's full of harmonics...all even
ones. But as long as the two tubes are well matched, the two half primary
currents are of identical shape and add in the OPT secondary to produce a
full sine, symmetrical about its zero voltage axis.
The even harmonics are gone. Is this cancellation? Well I guess you might
call it that.

Basically any waveform symmetrical about its zero voltage axis will contain
no even harmonics...even a symmetrical square wave. Similarly, if we
symmetrically clip the tops off both the positive and negative half cycles
of our sine we do not see any even harmonic content either for the same
reason. The ideal push-pull output stage using matched devices is not going
to produce even order distortion regardless of its operating class, whether
it be Class A, Class AB or Class B.

If in any doubt, see pages 300-301 of RDH4, or any university level math
text covering Fourier analysis.

Best Regards : Doug Bannard, P. Eng.- Hide quoted text -

- Show quoted text -

That looks like not created v. cancelled from here...:)
cheers,
Douglas

Output classes A and AB
 

On Oct 27, 4:29 pm, "Doug Bannard" wrote:
"John Byrns" wrote in message
{snip]
I admit I do not understand this notion that even order distortion can't
be canceled when one tube is cut off, and I would really appreciate an
explanation from someone that understands why cancellation should fail
to occur?

Why I don't understand this is because even order distortion is canceled
in class B amplifiers, where one or the other tube is cutoff over almost
the entire cycle, yet cancellation still occurs! I am having trouble
reconciling this with your claim that cancelation can't occur. Also the
related observation that there is no even order distortion in a sine
wave which has had both the tops and bottoms of the sine wave
symmetrically clipped off.

Please explain.

Regards,

John Byrns

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

John:

I think that the problem that Patrick may be having with this concept is
more one of semantics than anything else. It's hard to intuitively view
"cancellation" of even harmonic terms if both tubes are not conducting at
the same time. Let's view it with a very small amount of math thrown in.
Sometimes intuition does not take one far enough.

Take a pure Class B amplifier. Each tube or transistor conducts for 180
degrees (forget about any crossover effects). The current in each half of
the OPT primary, neglecting any magnetizing current is a half sine. Do a
Fourier expansion on this half sine and it's full of harmonics...all even
ones. But as long as the two tubes are well matched, the two half primary
currents are of identical shape and add in the OPT secondary to produce a
full sine, symmetrical about its zero voltage axis.
The even harmonics are gone. Is this cancellation? Well I guess you might
call it that.

Well, I'd call it cancellation. I didn't want to give my take on this
and get the subthread invaded by the usual fartcatchers who follow me
around before John received a sound answer, and yours is a good and
complete answer; thanks Doug from me too. I don't see how anyone but a
pedant can object to the "removal result" of the combination of a
positive and a negative half-cycle being called a cancellation; it is
an "addition" only in a mathematical sense (and it goes against the
grain!). But Patrick, clinging so tightly to the formal definition,
had me wondering whether there was something else that made my
commonsense view wrong.

Basically any waveform symmetrical about its zero voltage axis will contain
no even harmonics...even a symmetrical square wave. Similarly, if we
symmetrically clip the tops off both the positive and negative half cycles
of our sine we do not see any even harmonic content either for the same
reason. The ideal push-pull output stage using matched devices is not going
to produce even order distortion regardless of its operating class, whether
it be Class A, Class AB or Class B.

If in any doubt, see pages 300-301 of RDH4, or any university level math
text covering Fourier analysis.

Nah. We don't need to resort to such desperate measures while we have
you and Patrick to explain things to us.

Best Regards : Doug Bannard, P. Eng.

Thanks again for restoring my faith in the simple things...

Andre Jute
There are more things in heaven and earth, Horatio -- Will the Shake
Your first take on a problem is usually the right one -- John Z
DeLorean, when we were young

Output classes A and AB
 

John Byrns wrote:

In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Eeyore wrote:

Patrick Turner wrote:

One cannot have distortion cancelling by one tube cancelling that in
another when one is cut off.

THANK YOU !

Basics do matter.

Indeed they do, but neither Patrick, myself, or anyone else is correct
on every issue. In this case Patrick has vigorously asserted that this
view, which he holds in common with you, is true, but he has failed to
even attempt an argument that might demonstrate its truth. Patrick is
an extremely skilled and talented fellow in the practical aspects of
tube amp design and construction, but he has a very limited
understanding of what is going on behind the scenes in the theory of
tube amp operation.

I think my website might indicate that your are not quite right about my
levels of understanding.

Can you point out where on your website it explains why "One cannot have
distortion cancelling by one tube cancelling that in another when one is
cut off"?

No, because I endorse what many books tell us such as RDH4.

I refuse to spoon feed ppl about the basics.

But simple use of the brain about what happens inside each tube of a PP
pair
in class A, AB or B or C etc reveals to most minds what happens.

I admit I do not understand this notion that even order distortion can't
be canceled when one tube is cut off, and I would really appreciate an
explanation from someone that understands why cancellation should fail
to occur?

Once one tube is well and truly cut right off, the other is conducting
delivery
of all the power through only 1/2 the OPT primary, and its as if the
other cut off tube is unplugged
for that part of the wave.

Each output tube takes a turn at current delivery by means of turn on
harder to produce
voltage change which then SUM but without cancelation of current.
The cancelation thinge is when the tubes are in class A and the 2H is
cancelled, similarly
to in any LTP. The Class A Wiulliamson has a common cathode resistance
which is unbypassed
to assist the cancellation, but ideally, a CCS should be used, or even a
choke for the cathode
resistance. One may argue whether having each cathode with separate RC
bypass networks
is better than a shared cathode impedance/resistance.
But the individual cathode RC or fixed bias does allow class AB
operation, wheras the
common cathode R/Z does not.

Work out the 2H current in each tube while in class A of each tube and
see how such currents are applied across the primary.
The reason 2H is low in class A is that the 2H currents are the same
phase at each end of the
OPT primary which cannot produce voltage in the load if the current is
applied in common mode to
both ends of the load.


Why I don't understand this is because even order distortion is canceled
in class B amplifiers, where one or the other tube is cutoff over almost
the entire cycle, yet cancellation still occurs!

Class B and C amps have a summing action of their turn on currents which
pull the voltage
in oposite directio. There is huge distortion in the Class B device
current,
but the total action produces a linear voltage outcome.
a pair of complementary pnp and npn SS devices do exactly the same
thing.

But its not unusual that the amount of V swing achieved in each
direction by each class B device is not equal so
some 2H will allways appear in the output.
But usually, the 3H dominates H distortion products.


I am having trouble
reconciling this with your claim that cancelation can't occur.

You have to see the distinction between each tube in class A "sharing
the load"
and the summing actions once cut off has occurred.
One could say correctly that the severe current distortions of each
device in class B
are cancelled by means of the summing. In effect they are.

In class A, what one tube does with the load current affect the other
tube.
If you have one tube with high gm and the other with a low gm, then
the amount of class A power produced in each tube varies.
This in effect is because the load seen by each tube working as an SE
tube
varies, and where you have RLa-a = 10k, then the class A load of each
tube = 5k in theory.
Tubes ain't any more perfect than I am, and you will find that perhaps
one tube "sees" 4.5k,
and the other sees 5.5k. Careful measurements with 10 ohm current
sensing R in each anode or cathode circuit
will tell you about the current flow in each tube.


Also the
related observation that there is no even order distortion in a sine
wave which has had both the tops and bottoms of the sine wave
symmetrically clipped off.

A sine wave symetrically clipped usually has a pile of odd numbered H,
and not much 2H. Some 2H is usually there though.

Try measuring one of your AB amps at 1W, 3.2W, 10W, 32W and as clipping
progesses from mild to severe.

Patrick Turner.

Please explain.

Regards,

John Byrns

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

Output classes A and AB
 

In article . com,
Multi-grid wrote:

On Oct 28, 2:29 am, "Doug Bannard" wrote:
"John Byrns" wrote in message

...

In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Eeyore wrote:

Patrick Turner wrote:

One cannot have distortion cancelling by one tube cancelling that
in
another when one is cut off.

THANK YOU !

Basics do matter.

Indeed they do, but neither Patrick, myself, or anyone else is correct
on every issue. In this case Patrick has vigorously asserted that this
view, which he holds in common with you, is true, but he has failed to
even attempt an argument that might demonstrate its truth. Patrick is
an extremely skilled and talented fellow in the practical aspects of
tube amp design and construction, but he has a very limited
understanding of what is going on behind the scenes in the theory of
tube amp operation.

I think my website might indicate that your are not quite right about my
levels of understanding.

Can you point out where on your website it explains why "One cannot have
distortion cancelling by one tube cancelling that in another when one is
cut off"?

I admit I do not understand this notion that even order distortion can't
be canceled when one tube is cut off, and I would really appreciate an
explanation from someone that understands why cancellation should fail
to occur?

Why I don't understand this is because even order distortion is canceled
in class B amplifiers, where one or the other tube is cutoff over almost
the entire cycle, yet cancellation still occurs! I am having trouble
reconciling this with your claim that cancelation can't occur. Also the
related observation that there is no even order distortion in a sine
wave which has had both the tops and bottoms of the sine wave
symmetrically clipped off.

Please explain.


John:

I think that the problem that Patrick may be having with this concept is
more one of semantics than anything else. It's hard to intuitively view
"cancellation" of even harmonic terms if both tubes are not conducting at
the same time. Let's view it with a very small amount of math thrown in.
Sometimes intuition does not take one far enough.

Take a pure Class B amplifier. Each tube or transistor conducts for 180
degrees (forget about any crossover effects). The current in each half of
the OPT primary, neglecting any magnetizing current is a half sine. Do a
Fourier expansion on this half sine and it's full of harmonics...all even
ones. But as long as the two tubes are well matched, the two half primary
currents are of identical shape and add in the OPT secondary to produce a
full sine, symmetrical about its zero voltage axis.
The even harmonics are gone. Is this cancellation? Well I guess you might
call it that.

Basically any waveform symmetrical about its zero voltage axis will contain
no even harmonics...even a symmetrical square wave. Similarly, if we
symmetrically clip the tops off both the positive and negative half cycles
of our sine we do not see any even harmonic content either for the same
reason. The ideal push-pull output stage using matched devices is not going
to produce even order distortion regardless of its operating class, whether
it be Class A, Class AB or Class B.

If in any doubt, see pages 300-301 of RDH4, or any university level math
text covering Fourier analysis.

Best Regards : Doug Bannard, P. Eng.


That looks like not created v. cancelled from here...:)

It wouldn't look like "not created" if you used a current probe to
observe the anode current waveforms of each of the two PP output tubes.


Regards,

John Byrns

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

Output classes A and AB
 

In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Eeyore wrote:

Patrick Turner wrote:

One cannot have distortion cancelling by one tube cancelling that in
another when one is cut off.

THANK YOU !

Basics do matter.

Indeed they do, but neither Patrick, myself, or anyone else is correct
on every issue. In this case Patrick has vigorously asserted that this
view, which he holds in common with you, is true, but he has failed to
even attempt an argument that might demonstrate its truth. Patrick is
an extremely skilled and talented fellow in the practical aspects of
tube amp design and construction, but he has a very limited
understanding of what is going on behind the scenes in the theory of
tube amp operation.

I think my website might indicate that your are not quite right about my
levels of understanding.

Can you point out where on your website it explains why "One cannot have
distortion cancelling by one tube cancelling that in another when one is
cut off"?

No, because I endorse what many books tell us such as RDH4.

Yes, and if we read, understand, and apply what the RDH4 tells us, we
understand that even harmonic cancellation does occur in class AB
amplifiers, even when the tubes are cutoff for part of the cycle. That
is what this discussion is all about, Multi-grid, at least I think
Multi-grid was the first, made the claim that even harmonic cancellation
doesn't occur when the tubes in a class AB amplifier are cut off, then
both you and Eeyore jumped in to back up his claim.

Have you read pages 300 & 301 of the RDH4 as Doug Bannard, not to be
confused with the Multi-grid Doug, suggested? There is not a lot there,
but it is a start and hopefully might lead you to follow up by pursuing
some more complete references on Fourier analysis and the theory behind
it.

I refuse to spoon feed ppl about the basics.

But simple use of the brain about what happens inside each tube of a PP
pair
in class A, AB or B or C etc reveals to most minds what happens.

Simple use of the brain reveals that the cathode current of each tube of
a class A, AB or B amplifier contains even order harmonic currents,
among others. What simple use of the brain apparently doesn't reveal to
some is the reason these even harmonic currents are canceled in the
output of a PP amplifier. If the two PP sides of the amplifier are
identical, then the cancellation will be complete, if the to sides are
not identical, for any one of a multitude of reasons, including but not
limited to, mismatched tubes then the cancellation will only be partial.

I admit I do not understand this notion that even order distortion can't
be canceled when one tube is cut off, and I would really appreciate an
explanation from someone that understands why cancellation should fail
to occur?

Once one tube is well and truly cut right off, the other is conducting
delivery
of all the power through only 1/2 the OPT primary, and its as if the
other cut off tube is unplugged
for that part of the wave.

Yes, that is obvious, but not especially relevant.

Each output tube takes a turn at current delivery by means of turn on
harder to produce
voltage change which then SUM but without cancelation of current.

It is not necessary for each tube to be continuously delivering current
for the cancellation of the even order harmonic currents from the two
tubes to cancel, it is only necessary that the two currents exhibit the
required symmetry with respect to one another. The harmonic currents
flow even when the tube is cutoff, it's just that all the currents,
including the 2H current, happen to add to zero for the period the tube
is cutoff.

The cancelation thinge is when the tubes are in class A and the 2H is
cancelled, similarly
to in any LTP.

"The cancellation thingy" is also when the tubes are in class AB or even
class B, where the 2H is also canceled.

The Class A Wiulliamson has a common cathode resistance
which is unbypassed
to assist the cancellation, but ideally, a CCS should be used, or even a
choke for the cathode
resistance. One may argue whether having each cathode with separate RC
bypass networks
is better than a shared cathode impedance/resistance.
But the individual cathode RC or fixed bias does allow class AB
operation, wheras the
common cathode R/Z does not.

This is a diversion into a side issue but let me ask, why does a common
cathode RC not allow class AB? Granted fixed bias provides far better
class AB operation than does cathode bias, but given that we are using
cathode bias I fail to see how a common cathode RC would prevent class
AB operation anymore than separate cathode RCs do? What separate
cathode RCs do is reduce to some extent the effect of mismatched tubes
in the two sides of the PP circuit, but given well matched tubes a
common RC does not prevent class AB operation.

Work out the 2H current in each tube while in class A of each tube and
see how such currents are applied across the primary.
The reason 2H is low in class A is that the 2H currents are the same
phase at each end of the
OPT primary which cannot produce voltage in the load if the current is
applied in common mode to
both ends of the load.

The same is also true of the even order harmonic currents in class AB
and class B amplifiers!

I am beginning to get a glimmer of what your problem is here, it appears
that you are not correctly identifying the even order harmonic currents
that are applied to the output transformer primary in common mode. You
are looking at the total current waveform from each tube, which is
causing you to become confused. When the tube is cut off, the even
order harmonic currents continue to flow through the tube, its just that
the DC, fundamental, and all the harmonics together sum to zero during
the time the tube is cutoff, however this does not imply that the
current of any particular harmonic is aero during the time of cutoff,
hence cancellation goes on as if the tube had not cutoff.

Why I don't understand this is because even order distortion is canceled
in class B amplifiers, where one or the other tube is cutoff over almost
the entire cycle, yet cancellation still occurs!

Class B and C amps have a summing action of their turn on currents which
pull the voltage
in oposite directio.

It sounds like you are describing the push pull action of any class of
amplifier?

There is huge distortion in the Class B device
current,
but the total action produces a linear voltage outcome.

And that is exactly the point, which you persist in trying to deny.
Although odd order non-linearity does remain in class A, AB, and B
amplifiers.

a pair of complementary pnp and npn SS devices do exactly the same
thing.

But its not unusual that the amount of V swing achieved in each
direction by each class B device is not equal so
some 2H will allways appear in the output.
But usually, the 3H dominates H distortion products.

You are simply stating that in the real world, devices, drive signals
and other parameters of the two sides of a PP amplifier may not be
perfectly matched, and as a result the even harmonic cancellation will
be incomplete, but partial cancellation still occurs reducing the even
order harmonic distortion, even when one or the other tube is cutoff.

I am having trouble
reconciling this with your claim that cancelation can't occur.

You have to see the distinction between each tube in class A "sharing
the load"
and the summing actions once cut off has occurred.
One could say correctly that the severe current distortions of each
device in class B
are cancelled by means of the summing. In effect they are.

In class A, what one tube does with the load current affect the other
tube.
If you have one tube with high gm and the other with a low gm, then
the amount of class A power produced in each tube varies.
This in effect is because the load seen by each tube working as an SE
tube
varies, and where you have RLa-a = 10k, then the class A load of each
tube = 5k in theory.
Tubes ain't any more perfect than I am, and you will find that perhaps
one tube "sees" 4.5k,
and the other sees 5.5k. Careful measurements with 10 ohm current
sensing R in each anode or cathode circuit
will tell you about the current flow in each tube.

Not only will the load seen by each of the two tubes in a class A
amplifier vary depending on the relative gm of the two tubes, but even
with two perfectly matched tubes the relative load each sees will vary
over the complete audio cycle due to the dynamic changes in the tubes
characteristics over the cycle.

Also the
related observation that there is no even order distortion in a sine
wave which has had both the tops and bottoms of the sine wave
symmetrically clipped off.

A sine wave symetrically clipped usually has a pile of odd numbered H,
and not much 2H. Some 2H is usually there though.

There is no 2H if the clipping is symmetrical.

Try measuring one of your AB amps at 1W, 3.2W, 10W, 32W and as clipping
progesses from mild to severe.

I'm not sure what you think this experiment proves? All it proves is
that a voltage amplifier or driver stage is producing some 2H at high
levels of drive, or that the output stage begins clipping asymmetrically
at high output levels due to some circuit feature or imperfection, I
don't see what it has to do with the issue of even order harmonic
cancellation? If one wanted to go to the trouble I have little doubt
that given some effort one could design a class AB amplifier that
wouldn't exhibit the problems you are alluding to. You are simply
dragging a Red Herring across the path with this issue which is
unrelated to the issue of even order harmonic cancellation in class AB
amplifiers when the tubes cutoff.

And these sorts of problems aren't restricted to class AB amplifiers,
there are class A PP amplifiers that show a rise in even order,
including 2H, distortion even before clipping sets in.


Regards,

John Byrns

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

Output classes A and AB
 

In article ,
flipper wrote:

On Sun, 28 Oct 2007 13:12:57 GMT, Patrick Turner
wrote:

In class A, what one tube does with the load current affect the other
tube.

Not with pentodes (or any SS device) it doesn't. (Ideal) Pentodes are
essentially voltage controlled current sources and will source that
current regardless of what the 'other' is doing.

Yes, but what the other tube is doing will affect load seen by the first
tube, which will affect the voltage on its anode, if not its anode
current.

I think all this is just going to further confuse poor old Patrick, we
should probably let him alone so he can get back to his shed to wind
some output transformers and catch up with the backlog of Big US amps
awaiting modifications. While he working on that, his sub conscious can
work on the cutoff problem and perhaps an Epiphany will eventually come.


Regards,

John Byrns

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

Output classes A and AB
 

In article ,
flipper wrote:

On Sun, 28 Oct 2007 21:40:26 GMT, John Byrns
wrote:

In article ,
flipper wrote:

On Sun, 28 Oct 2007 13:12:57 GMT, Patrick Turner
wrote:

In class A, what one tube does with the load current affect the other
tube.

Not with pentodes (or any SS device) it doesn't. (Ideal) Pentodes are
essentially voltage controlled current sources and will source that
current regardless of what the 'other' is doing.

Yes, but what the other tube is doing will affect load seen by the first
tube, which will affect the voltage on its anode, if not its anode
current.

Of course, but Patrick was speaking of what happens 'in the tube' and
(ideal) pentode conduction is independent of anode voltage

The load change will affect voltage but as long as everything is
'identical' (which is never the 'real world' case, of course) both
sides will see and do the same thing so, again, even harmonics are
canceled. Odd harmonics are another matter.

The point here, though, was that 'tube interaction', and 'mutual'
conduction, isn't the 'canceling' mechanism. It's the OPT summing two
equal but opposite waveforms and it's inherent to what even harmonics
'are'. With both halves 'equal but opposite' they just plain don't
exist no matter 'how' you got there and how many you started with,
because they 'cancel'. Even if it's just equal but opposite 1ns blips,
no even order harmonics.


I think all this is just going to further confuse poor old Patrick, we
should probably let him alone so he can get back to his shed to wind
some output transformers and catch up with the backlog of Big US amps
awaiting modifications. While he working on that, his sub conscious can
work on the cutoff problem and perhaps an Epiphany will eventually come.

Well, let's not get too smug about it because there could/should be
'some' interaction with triodes so his description has intuitive merit
in that case. But it's a special case due to plate feedback in triodes
(which could/should affect all harmonics to some extent) and not PP,
per see.

I may not have been very clear, but my second sentence was referring to
the issue of whether or not cutoff makes even order cancelation
impossible. I consider the interaction question I referred to in the
first sentence to be only a side issue to the original question of
cutoff defeating even order cancelation.


Regards,

John Byrns

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

Output classes A and AB
 

John Byrns wrote:

In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Eeyore wrote:

Patrick Turner wrote:

One cannot have distortion cancelling by one tube cancelling that in
another when one is cut off.

THANK YOU !

Basics do matter.

Indeed they do, but neither Patrick, myself, or anyone else is correct
on every issue. In this case Patrick has vigorously asserted that this
view, which he holds in common with you, is true, but he has failed to
even attempt an argument that might demonstrate its truth. Patrick is
an extremely skilled and talented fellow in the practical aspects of
tube amp design and construction, but he has a very limited
understanding of what is going on behind the scenes in the theory of
tube amp operation.

I think my website might indicate that your are not quite right about my
levels of understanding.

Can you point out where on your website it explains why "One cannot have
distortion cancelling by one tube cancelling that in another when one is
cut off"?

No, because I endorse what many books tell us such as RDH4.

Yes, and if we read, understand, and apply what the RDH4 tells us, we
understand that even harmonic cancellation does occur in class AB
amplifiers, even when the tubes are cutoff for part of the cycle. That
is what this discussion is all about, Multi-grid, at least I think
Multi-grid was the first, made the claim that even harmonic cancellation
doesn't occur when the tubes in a class AB amplifier are cut off, then
both you and Eeyore jumped in to back up his claim.

I have lost track of whatever it was that Multigrid might have said.

But once one tube cuts off, the part of the cycle handled by the other
tube alone
generates a substantially linear output voltage when summed with the
other tube's efforts. IMHO, the summing action of tubes or other devices
in class B does not include the cancelation of harmonics
because of the common mode application of the near identical even
harmonics
of the same phase of both tubes across the whole primary.
The class B action means the current wave of the device is a series
of 1/2 sine waves with a flat part and many harmonics are present
if you ever wanted to filter them out and quantify them.
But these harmonics don't appear in the load. Fact.
There isn't mutual cancellation like in class A.
But there is crossover distortion as one device turns off, and the other
turns on.

Class B is like the two guys sawing the log with the long bush saw,
and each guy only pulls the saw from centre position his way and pushes
it back
to centre then lets go. The mens' current production is severely
distorted, and jerky,
stop'start in fact, but they manage to saw the log with the saw
travelling
one way and the other like an output voltage due to the summed
applications of the
efforts, ie, currents.

Once one device cuts off, no current flows, so there are no distortion
currents produced
by that device.

So the only relevant issue is what each tube does when it turns on, and
has the load
power under its control.


Have you read pages 300 & 301 of the RDH4 as Doug Bannard, not to be
confused with the Multi-grid Doug, suggested?

Page 299 through to pg 303 is about Fourier Series and Harmonics.

There is so much high falooting math I don't understand very much, and
i don't need to. The pages do not contain a large range of
down to earth simple current wave forms of PP amps with the
list and typical amplitudes of the harmonics present.


There is not a lot there,
but it is a start and hopefully might lead you to follow up by pursuing
some more complete references on Fourier analysis and the theory behind
it.

I do not need to become fully familiar with Fourier analysis.
I am not formally university educated, and don'y undertsand the math.
I make amops and do whatever relevant hands on measuring and analysis
that I need to do
without putting on the garb of the white coated boffin.
I do know that were I to measure the harmonics of a typical AB PP amps
current waves,
I would find a shirtfull and bootfull of harmonics when the amp is well
into class AB.
But a very small fraction of them find their way into the secondary of
an OPT,
or into a direct connected load of complementary pnp and npn devices.


I refuse to spoon feed ppl about the basics.

But simple use of the brain about what happens inside each tube of a PP
pair
in class A, AB or B or C etc reveals to most minds what happens.

Simple use of the brain reveals that the cathode current of each tube of
a class A, AB or B amplifier contains even order harmonic currents,
among others. What simple use of the brain apparently doesn't reveal to
some is the reason these even harmonic currents are canceled in the
output of a PP amplifier.

In a class B amp, the output current produced by each output tube is
substantially
linear with input voltage for each 1/2 wave of the sine wave.
Its tuned off for the other 1/2.

Its mate does the same substantially linear job on the other 1/2 of the
wave not handled by
the first.

Its a simple case of each device substantially linearly amplifying each
1/2 wave.
The Fourier Analysis might describe things with math, and any given
repeating waveform can be expressed in terms of various amplitudes of
sine wave harmonics
with amplitudes and phase relationships required to give the original
un-analysed non sine wave.
The quantified harmonics are simply not relevant in the class B case.

It could be argued cancelation does take place, but I see things simply
as a
arched 1/2 wave being a linear effort with ZERO effort made upon the
other 1/2 of the wave.

Now were you to analyse and quantify and draw the graphs for the
fundememental
and harmonics of EACH class B device, you could congratulate yourself on
your wizardry,
but just getting a PC spectral analysis program to do it all for you
could be
less prone to errors after days of work doing the RDH4 math "by hand".

After doing the graphs of the current waves displaying them so they
represent the currents
flowing in/out of each end of the OPT primary, perhaps you'd find that
lo and behold,
when the harmonic waves are all viewed, it is seen they applied in
common mode to each
end of the load and thus cannot appear as load voltage current since the
amplitude and phase of the harmonics is the same at each end of the
load, ie,
each end of the OPT.

But I see no need to do this labourious and unecessary math and graphing
task which may satisfy
the lofty smartarse acedemics.
RDH4 does not have too many full descriptions to simply illustrate what
I have been saying.
So what use are the smartarse acedemics? have they earned their dinner?
I don't see they have
unless they make easy for lesser mortals to understand the concepts.
So the Fourier analysis is merely confusing.
I see the tubes simply turning on and off to do various parts of the
wave, and while 'on',
they have a linear control of output current, but while off, they have
no control.

In some SS amps, there are two crossover regions as the output voltage
lifts its own rail voltages
as in class H, all very complicated in Fourier terms but all you need to
know is the
timing sequence of what happens during wave cycles.
The Soundcraftsmen amp was class H.


If the two PP sides of the amplifier are
identical, then the cancellation will be complete, if the to sides are
not identical, for any one of a multitude of reasons, including but not
limited to, mismatched tubes then the cancellation will only be partial.

The perfectly complementary action of two devices in class B never ever
occurs.
Crossover distortion of some kind spoils the attempt at perfection.

Where you have EL34 on one side and 6L6 on the other of a PP AB amp,
expect a large amplitude DIFFERENCE in the applied commonality of
harmonic currents applied
to each end of the OPT, so you'd get considerably more resultant THD
than if you had a matched pair
of the same tube type number.

But in common sense terms, each of the pair of tubes are just switching
on and off,
and while off, each tube is oblivious to its Fourier productions. Its
bloody turned off,
to there ain't nothin going on. If a tube could talk, it'd tell you
I am NOT making any harmonics while I is turned off, OK.
You might answer the cheeky brat of a tube by saying but all your
harmonics
add to a flat line of zero current while you are turned off,
and don't argue, because you can't argue with that guy Fourier.
So all these many harmonics are there, but appear not to be there for
1/2 a wave....


I admit I do not understand this notion that even order distortion can't
be canceled when one tube is cut off, and I would really appreciate an
explanation from someone that understands why cancellation should fail
to occur?

Once one tube is well and truly cut right off, the other is conducting
delivery
of all the power through only 1/2 the OPT primary, and its as if the
other cut off tube is unplugged
for that part of the wave.

Yes, that is obvious, but not especially relevant.

We cannot remove tubes and plug them back in fast enough to demonstrate
the common sense notion...

Each output tube takes a turn at current delivery by means of turn on
harder to produce
voltage change which then SUM but without cancelation of current.

It is not necessary for each tube to be continuously delivering current
for the cancellation of the even order harmonic currents from the two
tubes to cancel, it is only necessary that the two currents exhibit the
required symmetry with respect to one another. The harmonic currents
flow even when the tube is cutoff, it's just that all the currents,
including the 2H current, happen to add to zero for the period the tube
is cutoff.

See the above comments.

The cancelation thinge is when the tubes are in class A and the 2H is
cancelled, similarly
to in any LTP.

"The cancellation thingy" is also when the tubes are in class AB or even
class B, where the 2H is also canceled.

One could prove almost anything with Fourier math....

Its easier to see the mutually cancelling 2H of each tube in a class A
amp.

But while tubes are cut right off, summed distortion currents in that
tube = zero.

I have to say this does not mean the distortion currents are NOT
PRESENT.
They must be, but all summ to zero amps for the flat line of the tube
when cut off.
If you filter out harmonics of the current wave of a single tube of the
clas B amp,
you'll see the plethera of harmonics, and these are all sine wave
signals
which flow continuously, and without cutting off, but which when added
will give you zero current for
1/2 the cycle wave at fundememtal F.



The Class A Wiulliamson has a common cathode resistance
which is unbypassed
to assist the cancellation, but ideally, a CCS should be used, or even a
choke for the cathode
resistance. One may argue whether having each cathode with separate RC
bypass networks
is better than a shared cathode impedance/resistance.
But the individual cathode RC or fixed bias does allow class AB
operation, wheras the
common cathode R/Z does not.

This is a diversion into a side issue but let me ask, why does a common
cathode RC not allow class AB?

Because after one tube cuts off, the common Rk is the Rk for one tube
only,
and the Ra of the tube rises acording to (µ + 1) x Rk, and severe 3H
distortion results.



Granted fixed bias provides far better
class AB operation than does cathode bias, but given that we are using
cathode bias I fail to see how a common cathode RC would prevent class
AB operation anymore than separate cathode RCs do?

Well, the C bypassing the common Rk does improve AB working
considerably.
Quad-II has 180 ohms plus 40uF between cathode FB winding and 0V.
So for music at well below clipping the biasing stays fixed.
But with a sine wave the biasing voltage Ek rises dramatically in AB
and is as bad as separate RC to each cathode.

What separate
cathode RCs do is reduce to some extent the effect of mismatched tubes
in the two sides of the PP circuit, but given well matched tubes a
common RC does not prevent class AB operation.

True.

Its the low ZC that counts. If the C is ommitted, class AB is horrid.

The separate RC give far better self regulation of bias than a common RC
does.



Work out the 2H current in each tube while in class A of each tube and
see how such currents are applied across the primary.
The reason 2H is low in class A is that the 2H currents are the same
phase at each end of the
OPT primary which cannot produce voltage in the load if the current is
applied in common mode to
both ends of the load.

The same is also true of the even order harmonic currents in class AB
and class B amplifiers!

Well yes, but they are less easy to see, as i explained above.

In effect we agree then, but practically speaking a tube cut off has
zero
harmonic distortion currents for the time its cut off.


I am beginning to get a glimmer of what your problem is here, it appears
that you are not correctly identifying the even order harmonic currents
that are applied to the output transformer primary in common mode. You
are looking at the total current waveform from each tube, which is
causing you to become confused. When the tube is cut off, the even
order harmonic currents continue to flow through the tube, its just that
the DC, fundamental, and all the harmonics together sum to zero during
the time the tube is cutoff, however this does not imply that the
current of any particular harmonic is aero during the time of cutoff,
hence cancellation goes on as if the tube had not cutoff.

I explained this above.

The humble man in the street can be forgiven if he doubts any currents
of any kind flow
when the DC and harmonic currents are ZERO when the tube is cut off.



Why I don't understand this is because even order distortion is canceled
in class B amplifiers, where one or the other tube is cutoff over almost
the entire cycle, yet cancellation still occurs!

Class B and C amps have a summing action of their turn on currents which
pull the voltage
in oposite directio.

It sounds like you are describing the push pull action of any class of
amplifier?

There is huge distortion in the Class B device
current,
but the total action produces a linear voltage outcome.

And that is exactly the point, which you persist in trying to deny.
Although odd order non-linearity does remain in class A, AB, and B
amplifiers.

a pair of complementary pnp and npn SS devices do exactly the same
thing.

But its not unusual that the amount of V swing achieved in each
direction by each class B device is not equal so
some 2H will allways appear in the output.
But usually, the 3H dominates H distortion products.

You are simply stating that in the real world, devices, drive signals
and other parameters of the two sides of a PP amplifier may not be
perfectly matched, and as a result the even harmonic cancellation will
be incomplete, but partial cancellation still occurs reducing the even
order harmonic distortion, even when one or the other tube is cutoff.

The even order artifacts in PP are generated when the two halves of the
sine wave fundemental
are amplified to a different amplitude; ie, the transfer curve of each
tube isn't the same...


I am having trouble
reconciling this with your claim that cancelation can't occur.

You have to see the distinction between each tube in class A "sharing
the load"
and the summing actions once cut off has occurred.
One could say correctly that the severe current distortions of each
device in class B
are cancelled by means of the summing. In effect they are.

In class A, what one tube does with the load current affect the other
tube.
If you have one tube with high gm and the other with a low gm, then
the amount of class A power produced in each tube varies.
This in effect is because the load seen by each tube working as an SE
tube
varies, and where you have RLa-a = 10k, then the class A load of each
tube = 5k in theory.
Tubes ain't any more perfect than I am, and you will find that perhaps
one tube "sees" 4.5k,
and the other sees 5.5k. Careful measurements with 10 ohm current
sensing R in each anode or cathode circuit
will tell you about the current flow in each tube.

Not only will the load seen by each of the two tubes in a class A
amplifier vary depending on the relative gm of the two tubes, but even
with two perfectly matched tubes the relative load each sees will vary
over the complete audio cycle due to the dynamic changes in the tubes
characteristics over the cycle.

And you will get resultant intermodulations....

But class A PP remains basically more linear process than SE...

Also the
related observation that there is no even order distortion in a sine
wave which has had both the tops and bottoms of the sine wave
symmetrically clipped off.

A sine wave symetrically clipped usually has a pile of odd numbered H,
and not much 2H. Some 2H is usually there though.

There is no 2H if the clipping is symmetrical.

But only if the clipping is symetrical.
In many amps it is not perfectly symetrical, and 2H remains substantial.



Try measuring one of your AB amps at 1W, 3.2W, 10W, 32W and as clipping
progesses from mild to severe.

I'm not sure what you think this experiment proves?

Depends what you measure.

If its the amplitude of each harmonic one will get some surprising
results showing varying levels of each harmonic as outpt voltage is
increased.

For any given amp, its very difficult to predict and or calculate the
rate of rise in harmonics with output voltage.

Better you merely assume you'll get THD, and then measure it to know how
much.

Some empiricle assumptions can be made, but to know, you must measure.


All it proves is
that a voltage amplifier or driver stage is producing some 2H at high
levels of drive, or that the output stage begins clipping asymmetrically
at high output levels due to some circuit feature or imperfection, I
don't see what it has to do with the issue of even order harmonic
cancellation? If one wanted to go to the trouble I have little doubt
that given some effort one could design a class AB amplifier that
wouldn't exhibit the problems you are alluding to. You are simply
dragging a Red Herring across the path with this issue which is
unrelated to the issue of even order harmonic cancellation in class AB
amplifiers when the tubes cutoff.

And these sorts of problems aren't restricted to class AB amplifiers,
there are class A PP amplifiers that show a rise in even order,
including 2H, distortion even before clipping sets in.

Don't worry, be happy.

Patrick Turner.



Regards,

John Byrns

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