[west]

Class A sound, from what I know, is not really needed during most music
passages and its advantages are not fully appreciated during loud passages.
This leads me to ask ...
How can you take a basic 75w PP UL and run it at class A for, lets say, the
first 20 watts, and the rest at class AB? I can read curves but this is
somewhat dynamic and too far over my head, at least for now. IOW how do you
basically set up a "sliding" bias? Is there a disadvantage in trying to
accomplish this with a PP, UL? Thanks in advance for your comments.

Cordially,
west

[Ian Iveson]

west wondered:

Class A sound, from what I know, is not really needed during most
music
passages and its advantages are not fully appreciated during loud
passages.
This leads me to ask ...
How can you take a basic 75w PP UL and run it at class A for, lets
say, the
first 20 watts, and the rest at class AB? I can read curves but
this is
somewhat dynamic and too far over my head, at least for now. IOW how
do you
basically set up a "sliding" bias? Is there a disadvantage in trying
to
accomplish this with a PP, UL? Thanks in advance for your comments.

Several problems.

First, it's not clear what you mean. In terms of design specification,
Class A means that at full rated output, both valves are passing
current all the time. With AB, each valve stops conducting for part of
the time at full output, but for some of the time both are conducting.
Class B is when only one valve is conducting at any time, at full
output.

It follows that there must be an output level, less than full power,
at which a class AB amp is "operating in class A". Otherwise there
would be no time when both valves are conducting, so it would be a
class B amp.

So, if you must use class as a mode of operation rather than a point
of specification, then *all* class AB amps operate in class A at low
output levels. The level at which one valve begins to switch off can
be anything between very low and almost full power, depending on
design.

Second, what would you use to control a "sliding" bias? If you use the
signal, then it's not bias, is it? Perhaps you could rectify and
low-pass filter the signal and subtract it from the bias? Well, that's
actually what happens with simple cathode bias, as the even harmonics
drive up the voltage on the bypass cap. Whether this is a desirable or
euphonic phenomenon may be a matter of debate.

Thirdly, and the crux of the matter, a class A design should normally
have a higher output transformer impedance ratio and probably a lower
HT voltage. It will probably use a bias level halfway between cutoff
and grid current. Several other aspects may also vary from a typical
class AB design, to support those primary differences. The penalties
for this include low full power and hot idling. Advantages include
lower output impedance and greater linearity especially into a varying
load. In short, a class A amp is optimised for class A operation, and
a class AB amp is not. And vice-versa.

*You can't change a good class A amp into a good class AB amp, or
vice-versa, just by changing the bias*.

So I was interested to see that a new line of amps by Cambridge Audio
are labelled "class XD" for "crossover displacement"...which is
another way of saying "sliding bias".

Apparently, the output stage is set up for optimum class B operation.
It is dragged into class A by a constant current drain from the
output. Consequently, it is class AB, but the B part of the operation
is optimised, rather than the A part, so reducing crossover
distortion. Seems to me, incidentally, that it also means that the
class A part is SE, but they don't say that, and I don't know enough
about transistor output stages to be sure. To complicate matters,
Cambridge say that the constant current source varies with signal
level, so it is a form of feedback. It's all about cooking the
performance measurements, I think. Anyway, I put some thought into
doing a similar thing with valves, but it's silly...requires a gapped
tranny, a mostly-redundant output valve, and a stage that runs SE most
of the time but is optimised for class B...doesn't sound like a good
idea.

cheers, Ian

[Eeyore]

west wrote:

Class A sound, from what I know, is not really needed during most music
passages and its advantages are not fully appreciated during loud passages.
This leads me to ask ...
How can you take a basic 75w PP UL and run it at class A for, lets say, the
first 20 watts, and the rest at class AB?

Class A means that both negative and positive output devices conduct all the
time regardless of signal amplitude.

Class B means that one conducts, then the other. They 'cross over' at exactly
'zero power' output.

Class AB is similar to Class B but the output devices both continue to conduct
until a certain output power is reached, when they cross over.

What is it exactly that you want ?

Graham

[Yves Monmagnon]

"west" a écrit dans le message de news:
hIhrh.1190$R76.804@trnddc05...
Class A sound, from what I know, is not really needed during most music
passages and its advantages are not fully appreciated during loud
passages.
This leads me to ask ...
How can you take a basic 75w PP UL and run it at class A for, lets say,
the
first 20 watts, and the rest at class AB? I can read curves but this is
somewhat dynamic and too far over my head, at least for now. IOW how do
you
basically set up a "sliding" bias? Is there a disadvantage in trying to
accomplish this with a PP, UL? Thanks in advance for your comments.

Cordially,
west


Mmmmh, Google for QUAD 405 "current dumping amplifier".

Sure it was not a tubed unit, but is the idea of using a small class A
scheme directly driving the speakers at lo power, and a "big" class B
"buffer" for hi levels transposable ?
Even if I don't really know how !

Sliding bias looks difficult to acheive considering large and fast change in
output level.
In other words, what about this "pianissimo" just following an huge
orchestra hit ?

Anyway, most tubed amps using link caps in the grids of the power tubes
exhibit undesirable bias change (sliding ?) when over driven in grid current
region.

Yves.

[Patrick Turner]

west wrote:

Class A sound, from what I know, is not really needed during most music
passages and its advantages are not fully appreciated during loud passages.
This leads me to ask ...
How can you take a basic 75w PP UL and run it at class A for, lets say, the
first 20 watts, and the rest at class AB? I can read curves but this is
somewhat dynamic and too far over my head, at least for now. IOW how do you
basically set up a "sliding" bias? Is there a disadvantage in trying to
accomplish this with a PP, UL? Thanks in advance for your comments.

Cordially,
west

Perhaps if you could consider what i have said at my website at
http://www.turneraudio.com.au/loadma...p-triodes.html
http://www.turneraudio.com.au/loadma...mtetrodes.html

From all of that the load determines the amount of class A and AB.

Since speaker loads vary dynamically because at different F there is a
diferent speaker load,
and music contains many different F.
The amp really doesn't know what its doing much because of the waffly
loading and
ever changing signal, so it just settles for making the output signal
power a lot more than the input signal power,
and it hopes its owner allows for enough ceiling, simple, without having
to have its biasing fiddled
with during all this juggling with Motzart, or Led Zeppelin.

But for SLIDING BIASING of class A or AB amps the idle current can be
increased
for large amounts of power to keep all the music production by means of
class A rather than AB
if it is deemed that the load will force the amp to make too much B
power if the load is a low value.

This isn't what i would ever try to do, but essentially one produces a
positive voltage which is applied to the
grid bias circuit that is dependant on the output signal current of the
amp.
Thus the grid bais voltage becomes less negative and it is done in a way
so that this only occurs SLOWLY and sonically un-noticeably, and only
when the output signal current
reaches over a threshold. It means that where one may have Pda = 25
watts for a KT88 in a PP output circuit, the
bias Ia is increased so that it slides upwards so Pda = maybe 45 watts
as a result of some
large volume programme material, but it then slides slowly back down to
25 watts when the programme
becomes lower amplitude; This means that rock and roll with almost no
dynamic range would
raise the bias and it would stay high, but for classical the bias would
be continually rising or falling.
But never would the bias be where one wants it fast enough for sudden
crescendos appearing after a quiet solo passage.
But between bangs and crashes of rock and roll the tubes are dissipating
perhaps more Pda than one would want,
and it could be unreliable, and frankly, we don't need to increase
greenhouse gass in this
rather vain-glory manner.

There have been a number of sliding bias SS amps over the years where
the makers have claimed all sorts of things but really,
changing bias on an SS amp that is well made is often utterly inaudible.
I know, I have tried
adjusting the bias of a class Ab SS Cambridge amp while listening
between class C and heavy class A and heard exactly
no sound change, proving that biasing isn't important in high NFB
designs because the NFB
is what keeps the THD/IMD below 0.05% regardless of the bias setting or
programme level.
( BTW, the Cambridge amp did a harsh job on music regardless of what the
bias setting was )

In a tube amp the change of bias may be more noticeable because with
much lower NFB there is not such a capable mechanism to
make the amp perform electronic gymnastics in order to keep the thd/imd
low,
and in the case of a cathode biased PP class AB amp, the cathode voltage
gets higher as output signal current increases
and the dc power to the outpur stage increases.
When Ek rises, the bias voltage between grid and cathode places the
output tube bias into what is used for class B or even C
bias condition where they are cut right off during part of the cycle so
no amount of NFB
will correct the problem so you may get 3% thd with a sine wave at
clipping for the class AB power,
wheras there would be 0.2% if the tubes could remain in class A, or 0.5%
if the tubes were
operating with fixed bias.

But allowing the bias to become more positive when controlled by the
signal current is asking for trouble
with reliability, and no maker i know does this. It means you must have
current sensing R
and a bias voltage amp, low pass filter, all of which will be
embarrassing
if it fails.....

if you don't like using huge amounts of pure class A, using a current
dumping circuit
arranged so that there is at least a small class A circuit always
powering the output, and when the
loading and volume level threatens to force the little class A amp into
class AB operation,
then some tubes biased off are turned on to keep the "pilot" class A amp
working in class A.

Some attempts commercially have been made where a pair of EL34 in triode
make the "pilot" amp class A power,
but also connected to the OPT are a second pair of EL34 operating in
pentode and
in parallel which are operating in class B and biased
right off for low power levels, and this avoids the cross over
distortions at low levels.
But there is distortion at higher levels as the pentodes are switched on
along the cycle someplace,
between say 10 watts and 70 watts maximum.
The biasing of the pentodes sets the transition point.

The output resistance of the output stage thus configured is always at
least the Ra-a of the triodes
which is lower than a quad of pentode EL34 all in plain parallel and
equally baised for class AB.

Quad developed current dumping with SS devices in their 405, but there
isn't anything stopping
folks from using the same principle with simpler circuitry with tubes.

Electronics sometimes is like playing with russian dolls.
You remove the cover off doll No1, and there is doll No2 with a scowling
look,
and doll 3 smiles in a pink dress, while 4 appears naked and 5 is an old
hag; one doesn't
always know what one is to get the more you dig around, and the more
complex the amp becomes.

Quantum Mechanics is a lot worse though....

Patrick Turner.

[west]

"flipper" wrote in message
...
On Wed, 17 Jan 2007 04:35:25 GMT, "west" wrote:

Class A sound, from what I know, is not really needed during most music
passages and its advantages are not fully appreciated during loud
passages.
This leads me to ask ...
How can you take a basic 75w PP UL and run it at class A for, lets say,
the
first 20 watts, and the rest at class AB? I can read curves but this is
somewhat dynamic and too far over my head, at least for now. IOW how do
you
basically set up a "sliding" bias? Is there a disadvantage in trying to
accomplish this with a PP, UL? Thanks in advance for your comments.

If I understand your intent, Class AB *is* what you're talking about.
It operates Class A (the 'A' part of Class AB) to some power level and
Class B (the 'B' part of Class AB) for the rest.

You can pick your idle for how much you want in Class A vs Class B,
while staying under plate dissipation.

Will you show me how to do this using a real life example? Thanks.

west