I've found a textbook that uses an AB buffer stage mosfet, but in an
excercise a current in output transistor is always present for all
output current range to avoid distortion. I have a question: if
output transistors are always on, output stage is an A stage (push-
pull) and not an AB stage, is this right?
They are defined as:
1. Class A - The operational characteristics is defined such that current is
flowing at all times through the amplifing device(tube, transistor, etc).
2. Class AB - The operational characterics is defined such that current is
flowing more than half but less then the full electrical cycle.
3. Class B - The operational characterics is defined such that current is
only flowing when the input changes beyond is steady dc value.
4. Class C - The operational characteriscs is defined such that current is
only flowing for less than half of each cycle.
They are in some sense a shifted version. Assume the device can only conduct
one way. Then the distance then "dc" of the input from the input of no
current defines the class.
i.e., if the averaged "dc" input is much greater than the 0-current state
then it is class A. If it is close to but above then it is class AB. If it
is the same as the 0-current state then it is class B, and if much lower
then class C.
Mathematically, Suppose f(t) is the input signal with the "average dc"
defined as [f(t)] and 0 meaning the input with 0 current(usually 0V DC ).
Class A - [f(t)] >> 0
Class AB - [f(t)] > 0
Class B - [f(t)] = 0
Class C - [f(t)] < 0
(note [f(t)] << 0 would mean no ouput ever so would be useless)
(the >> and < are made more precise if one knows the max swing of f(t) but
can also be defined as above by saying the current is not flowing for a
minority, half, or majority of the cycle)
(also keep in mind it all has to do with the input that produces no current.
Most of the time this is an input of 0 but not always.)
Another way to think about it is:
Class A - the input is "shifted up" so that the amplifier is always
conducting current. (you must know the max signal swing for this to be
completely valid because it's possible that a class A could then be a class
B in some cirtumstances(but somewhat useless circumstances)).
Class AB - The input is "shifted up" only some so that part of the signal
"clips" when it goes below 0(since it can't go below 0 (unless push pull of
course which is different)). Remember, 0 is the no-current state so when the
signal gets "clipped"(or limited) at 0 we have no current for some of the
cycle. It shouldn't clip more than half a cycle else it is class B or C.
Class B - The input is not shifted and then only the "positive" part
conducts(the negative part gets clipped).
Class C - The input is actually shifted down so the signal conducts only on
part of the positive cycle.
(this assumes the input has no DC-bias... if it does then just remove it
first)
Another way to look at it is in terms of power. Class A is conducting all
the time and hence uses more power. Class AB uses less power because it is
not conducting between 50% and 100% of the time. Class B is conducting only
for 50% of the time and class C is < 50% of the time.
(although 49% would still be called class B)
Mathematically,
P(class C) < P(class B) < P(class AB) < P(class A)
All the discussion above only references single ended. For push-pull and
other stuff it is different. (but usually related)
