> Here is an explanation
It raises some points which could be seen otherwise.
Page 5: 'point B must never be over 100%'..... this can be violated, sometimes quite wildly. Moreover point B is not special, or not even very important, because sine-signal touches B for only an instant. The tube spends ~~40% of its time on the line from B to almost A, and is *over* the 100% line all this time (touching 50W Pdiss). The main thing is that the "off" half-cycle (not shown) tends to *zero* dissipation, and the 30 Watts is "averaged over an audio cycle", so over the "on" half-cycle we can swing quite far over the 100% line.
On the other hand... a specific "feature" of push-pull is that, with ideal devices, we can idle the amplifier at *zero* current. Think of the savings of heat, and tube life! So why do we never do that? Because real devices have zero gain at zero current. Small signals come out under-sized, "crossover" distortion.
So we idle the devices at some non-zero current, where they have some gain.
Now we have both devices working on small signals, but on the loudest signals one device cuts-off on the peaks and the other device carries the peak. If each device's gain were constant, gain would drop-off on the peak, another form of distortion. However we already know gain varies with current. If we idle the devices so the gain at idle is about half of the gain at the peak, gain is fairly constant from small to large.
Gain drops-off slower than current. If you assume that vacuum tube gain varies about as square-root of current, you can see that we want the per-tube idle current to be about 1/4 of the peak current, to get fairly constant gain over the whole cycle.
The loadline in the McCaul paper peaks near 350mA. So we might like to idle at 88mA. However 88mA at 450V is 39 Watts, far past 100% dissipation.
This will be true for most high-voltage power amps. In experiments on a clean 7189 amp, I liked it best with the tubes cherry-red. However turning-down well below Pdiss didn't sound a lot different. working *far* below Pdiss did.
And in real life we don't know what the peak current is. Say a Corgi SEx1300 comes in the shop. We don't know the loadline or the power output. Sometimes we are only paid for "a quick lookover"; rarely an extended analysis and listening-test. About all we know is the power-bottle type, from which we can find max Pdiss.
We have to pick a compromise between idle dissipation and low distortion. 100% Pdiss is hard on the tubes. 70% Pdiss is much less hard on the tubes. 70% of Pdiss is easy to find, and will almost always work about as good as any other.
Topic: Bias setting?? (Read 5371 times)