John Byrns wrote:
In article ,
Patrick Turner wrote:
John Byrns wrote:
In article ,
Patrick Turner wrote:
Cancelation of even order harmonics occurs in amps working in class AB
during that part of the wave forms which are in class A, ie, the bits
either side
of the zero crossing.
But once each tube moves into cut off, nothing is cancelled.
Patrick, I'm surprised to hear you say this. What are you trying to
tell us, that the even order harmonics are only cancelled during those
parts of the cycle when both tubes are conducting, but that the even
order distortion components reappear during those parts of the cycle
when only one tube is conducting?
In effect the even order distortions DO re-appear when the amp moves to
class AB.
Its because of the mismatch between the two tubes, and each has slightly
or greatly different gm, so each +ve and -ve half of the waves are
amplified
by a different amount, and hence you get even order generated in the
output from across the whole primary
or secondary of the OPT.
You are changing your tune a bit, you originally said "once each tube
moves into cut off, nothing is cancelled",
One cannot have distortion cancelling by one tube cancelling that in
another
when one is cut off. But the SUM of the joint action of each tube in
class AB with very non linear currents
gives a linear voltage outcome.
now you say "Its because of
the mismatch between the two tubes", which is a completely different
deal.
In class A, you often do not get perfect 2H cancellations, since tubes
are not matched,
so 2H is present with 3H etc.
The effect of missmatched tubes gets worse when the amp moves to AB.
more 2H with more of everything else. Summing of the output voltage
still occurs while little cancellation happens.
I stated in several of my posts that the tubes must be identical,
of course that is impossible in the practical world, but we want to get
as close as possible if we are intent on minimizing distortion. When
the tubes are mismatched, the problem is not just confined to class AB
operation, it will occur even in a class A amplifier with mismatched
tubes.
That is what I have said all along. PP isn't perfect.
Tubes and bjts and mosfets ain't perfect.
SS pn and pnp devices are often about as similar as using a 6L6 and EL34
in a tube PP amp; ie, not very well matched.
Doesn't madder, add piles of NFB and she'll be right....
The bias may be able to be adjusted to cancel the 2 nd harmonic
at some selected power output, but there is no guarantee that the other
even harmonics will also be canceled, even in a class A amplifier if the
tubes are mismatched, and if the power output level is changed even the
second harmonic can reappear with mismatched tubes. It is the
mismatched tubes that are at the root of the problem, not operating in
the class AB region with the tubes cut off over part of the cycle.
Matched tubes don't stay matched. One should not be too dependant on
matched tubes.
In a class AB amp with 50% of the power in class A before AB commences,
and with tubes with 10% gm variation, the amount of 2H in the first few
watts is way less than if the tubes were operating
in SE mode.
One does not have to worry if the TD in an AB a tube is below 0.1% at 2
watts.
The 2H and 3H will be dominant, and not cause undue IMD.
if the same amp makes 50 watts in AB, and THD is 0.5%, then its OK,
still not to bad,
but nodody will notice it.
I like having less than 0.05% for everything below average levels; if
for an instant during a drum beat
power rises to an instant 50 watt level, and the THD leaps to 0.5%, I am
not concerned.
Most of my PP amps have less than 0.25% at clipping and 0.03% at average
levels.
I have heard a VAC amp with 4 x 300B in PP for each channel to make 56
watts max,
and its usable with/without global NFB.
There is no difference to the sound that I am aware of when altering NFB
from zero to
having 9db applied. This means that if the THD is 0.1% with zero NFB,
applying 9dB to reduce it to about 0.03% makes no audible difference to
me.
Patrick Turner.
Regards,
John Byrns
--
Surf my web pages at, http://fmamradios.com/