Had a fellow amp builder ask me this question the other day regarding a well functioning, fixed bias P-P amp plan: What happens to the bias voltage when a signal is going through the amp? Meaning, if you were measuring, say, -40'ish VDC on the control grid of the power tubes with no signal what would it read with a signal.
...
He reports that it drops to -55'ish VDC when throwing a well struck power chord at the input.
I couldn't get my head around that!
Hey Dave!
You and a friend are atop 10ft ladders, on wither ends of a 40ft rope. You stretch it essentially tight between you, and the rope remains a consistent distance from the ground. That's your d.c. bias. The difference is you and your friend are physically above ground, while with respect to output tube bias, the electrical situation is mirrored and "below ground."
Now you and your friend shake the ends so that you develop a wave in the rope. The rope, at various points, rises above and drops below the height you maintained when the rope was stretched tight. That's your d.c. bias with a.c. superimposed.
Imagine (for whatever reason) you and your friend shake the rope so hard and far that the peaks of the rope's wave run into the ground. This
might approximate a d.c. bias with a superimposed a.c. signal which is large enough to drive the grid beyond 0v and draw grid current. Since our scenario is a mirror image, the rope is running into a boundary downward, while with our tubes the 0v grid boundary is "upwards" from our negative bias.
One possible reason bias voltage can deviate from its static value is driving the grid positive, such that it draws grid current, while having a signal source which cannot deliver this grid current (i.e., deliver voltage
and current, or [/i]power[/i]). The grid current has to flow out from the grid through the grid reference resistor, and tends to develop a charge on the coupling cap. The charge wouldn't exist if the source of our signal could deliver power, and supply an equal current to offset the charging of the cap.
The charge on the cap tends to add to the bias, resulting in a more-negative d.c. voltage at the grid. Large enough, or prolonged overloads seems to develop a big enough bias to "shut off" the tube momentarily. That's called "blocking distortion".
Your friend was probably driving the output stage hard enough to cause distortion, which probably also means it was driven hard enough to drive the grid slightly positive. That resulted in the apparent shift of the bias voltage. Depending on whether he used a guitar or a signal generator, he may not have realized there's an upper limit of reasonable signal strength for a test (when using the generator). If he was using a guitar, he might not have enough control to avoid transients causes the shift. Or he just didn't know test conditions effect the test results.
There are other reasons that bias shift can happen; it often goes the other way, appearing to reduce the bias. That's called "rectification effect" and is described in the section on power amplifiers in RDH4. In a nutshell, tube distortion components are not all the same phase. Depending on the relative amounts and phase of each harmonic frequency, a resultant d.c. current happens which may be an increase or decrease from the idle value. Very often, it causes the apparent d.c. to rise with signal. This is more easily observed in a cathode-bias amp, where you can monitor voltage across the cathode resistor. Aiken says the difference in voltage across the AC30's cathode bias resistor, as indicated in the schematic, is proof the amp is not class A. The AC-30 may or may not be class A, but even a class A stage can have voltage variation across the cathode resistor due to screen current and/or rectification effect (or distortion).
Topic: Bias voltage measurement (Read 3707 times)