#1 · Aug 19, 2008 17:42 UTC
OK... here's where the "challenge" comes into play! I want to provide some additional thoughts on power amps, negative feedback, psuedo-Class A operation and some other things. Things to "chaw on" as they say.
psuedo-Class A Operation
I want to re-quote something already quoted in another topic, as it directly applies here, too. This comes from Mullard - Tube Circuits for Audio Amplifiers. The section is Pentode Push-pull Stages, located on Page 18 of the book. Keep in mind this was a "definitive" guide, created and published by Mullard in 1959 to help tube purchasers better understand the application and integration of the tubes purchased by Mullard. Similar to the data sheets and Application Notes created, today, by companies like On Semi, TI, etc.
The conditions for Class AB operation normally recommended and published by valve manufacturers are based on measurements with continuous sine wave drive. The bias under zero-drive conditions and the anode-to-anode load resistance are so chosen that the optimum performance is achieved when the working points of the valves are displaced under driven conditions. This displacement is caused by the effect of increased anode to screen-grid currents on the cathode-bias circuit. For a typical oiutput stage working from a 310V mains supply and using EL84s, the rise in cathode current - and thus in cahtode bias voltage - with a sinusoidal signal voltage is about 40% at full drive...
When such a stage is used in the reproduction of speech or music, however, the operating conditions are different. The mean amplitude of the input signal is now very small compared with the peak values which occur from time to time, and thus the mean variation in the cathode current is also very small. Because of the relatively long time constance of the bias network, even under peak signal condition, the displacement of the working point is small enough for the stage to be considered as working with fixed bias...
There is something, here, that's important to note... The tubes are biased, in the power amp, for Class A, AB1 or AB operation. Tubes can be biased using one of two different methods: 1) Adjustable bias in which the grid voltage is controlled by a pot/power supply and balanced. In this configuration the cathode is, typically, connected directly to ground, or to a very small resistor (1 - 5 ohms max) used for "current sensing". 2) Cathode bias (self-biasing) in which a cathode resistor is placed between the cathode and ground. This resistor will vary in value based on the "idle current" required by the stage.
The cathode bias is rather nice as the feedback generated by the cathode resistor will keep the tube operating at the "correct" bias no matter how the tube degrades over time. Fixed bias must be adjusted, through the life of the tubes, in order to maintain the proper bias/balance as the tube(s) degrade from use. Typically, this adjustment should be made about every 50-100 hours of operation.
No matter how the stage is biased (fixed or adjustable) the bias setting sets the tube grid at a specific voltage level. Now, let's say that the output, from the preamp stage, has a positive, and negative, voltage swing that is LARGER than the grid bias voltage, by some amount. Whenever the signal swings negative one of the tubes will "shut off" (go into cut-off) BEFORE the other tube goes into saturation. Whenever the signal swings positive the same will happen, but for the opposite tube.
Class AB1/AB operation is configured, and tested, at MAXIMUM SIGNAL INPUT TO THE POWER AMP TUBES! Thus, under these conditions, one tube will cut-off and the other will approach, but not get to, saturation. In the other direction the same occurs. The "transfer" occurs at various points in the signal swing, based on the bias voltage set at the grid of the tube.
Let's use an example, here... let's say the input voltage, to the power amp, is set to 3V peak-to-peak, 1.5V peak and grid voltages are set up to -1V on the power tubes. That means for .5V of the input swing (30% of the time) one tube will be "biased off" by the signal and the other will continue to trek toward saturation. In the reverse, the same thing will happen, but the tube(s) impacted by this will be reversed.
Now, what happens if the signal is at, or below, 1V? Each power tube, in this example, will work through the ENTIRE swing of both the positive AND the negative portion of the signal acting more like a "Class A" amp than a Class AB/AB1 amp! WHAT???? It's true... if this were a TRUE Class B amp, then each tube would ONLY work on its portion of the wave and the output would be HORRIBLY distorted.
So, in this example, a Marshall/Fender output stage might "extend" about 70% into the next tubes region. In the VOX AC30 the extension/overlap is closer to 80%... slightly further than most AB1 topologies. By the way THESE NUMBERS ARE NOT ACCURATE THEY ARE JUST INTENDED TO ILLUSTRATE THE POINT! DO NOT CREATE AN AC30 MYTH FROM THESE NUMBERS ;D
However, in Class AB and AB1 topology each tube amplifies PAST the 0V point INTO the region of the other tube. Class AB1 push-pull amps move FURTHER into that region than a Class AB amp but BOTH move into that region by some amount. Most guitar amps are biased in Class AB1 range and work for at least 50% of the other tubes signal swing, or more, depending on the bias set by the output stage designer.
This means you can pull all but one output tube and STILL get a perfectly undistorted signal! PLUS, the feedback voltage is near "0", because of the decreased output voltage from a single tube and not the full complement of tubes. This "loosens" the output stage and decreases damping causing the stage to be more "loose" and "flabby", a bit more "bluesy". Put the tubes back in and the feedback increases in correct proportion and "tightens up."
If you'd like to "loosen up" the stage, you can increase the value of the LARGER of the two resistors in the feedback loop. This will decrease the feedback voltage, loosening the stage up. You could, even, open this resistor up and reduce negative feedback to "0". So long as the amp is designed without too much phase shift it should handle this quite nicely.
If your amp is too loose, you can INCREASE the value of the smaller resistor in the feedback loop. This will increase the amount of feedback voltage, decrease gain as a function of that AND "tighten" up the amp a bit.
These are changes that can be made to the amp to change its response and performance, overall.
Enjoy... ask questions if you have them.
Dar
psuedo-Class A Operation
I want to re-quote something already quoted in another topic, as it directly applies here, too. This comes from Mullard - Tube Circuits for Audio Amplifiers. The section is Pentode Push-pull Stages, located on Page 18 of the book. Keep in mind this was a "definitive" guide, created and published by Mullard in 1959 to help tube purchasers better understand the application and integration of the tubes purchased by Mullard. Similar to the data sheets and Application Notes created, today, by companies like On Semi, TI, etc.
The conditions for Class AB operation normally recommended and published by valve manufacturers are based on measurements with continuous sine wave drive. The bias under zero-drive conditions and the anode-to-anode load resistance are so chosen that the optimum performance is achieved when the working points of the valves are displaced under driven conditions. This displacement is caused by the effect of increased anode to screen-grid currents on the cathode-bias circuit. For a typical oiutput stage working from a 310V mains supply and using EL84s, the rise in cathode current - and thus in cahtode bias voltage - with a sinusoidal signal voltage is about 40% at full drive...
When such a stage is used in the reproduction of speech or music, however, the operating conditions are different. The mean amplitude of the input signal is now very small compared with the peak values which occur from time to time, and thus the mean variation in the cathode current is also very small. Because of the relatively long time constance of the bias network, even under peak signal condition, the displacement of the working point is small enough for the stage to be considered as working with fixed bias...
There is something, here, that's important to note... The tubes are biased, in the power amp, for Class A, AB1 or AB operation. Tubes can be biased using one of two different methods: 1) Adjustable bias in which the grid voltage is controlled by a pot/power supply and balanced. In this configuration the cathode is, typically, connected directly to ground, or to a very small resistor (1 - 5 ohms max) used for "current sensing". 2) Cathode bias (self-biasing) in which a cathode resistor is placed between the cathode and ground. This resistor will vary in value based on the "idle current" required by the stage.
The cathode bias is rather nice as the feedback generated by the cathode resistor will keep the tube operating at the "correct" bias no matter how the tube degrades over time. Fixed bias must be adjusted, through the life of the tubes, in order to maintain the proper bias/balance as the tube(s) degrade from use. Typically, this adjustment should be made about every 50-100 hours of operation.
No matter how the stage is biased (fixed or adjustable) the bias setting sets the tube grid at a specific voltage level. Now, let's say that the output, from the preamp stage, has a positive, and negative, voltage swing that is LARGER than the grid bias voltage, by some amount. Whenever the signal swings negative one of the tubes will "shut off" (go into cut-off) BEFORE the other tube goes into saturation. Whenever the signal swings positive the same will happen, but for the opposite tube.
Class AB1/AB operation is configured, and tested, at MAXIMUM SIGNAL INPUT TO THE POWER AMP TUBES! Thus, under these conditions, one tube will cut-off and the other will approach, but not get to, saturation. In the other direction the same occurs. The "transfer" occurs at various points in the signal swing, based on the bias voltage set at the grid of the tube.
Let's use an example, here... let's say the input voltage, to the power amp, is set to 3V peak-to-peak, 1.5V peak and grid voltages are set up to -1V on the power tubes. That means for .5V of the input swing (30% of the time) one tube will be "biased off" by the signal and the other will continue to trek toward saturation. In the reverse, the same thing will happen, but the tube(s) impacted by this will be reversed.
Now, what happens if the signal is at, or below, 1V? Each power tube, in this example, will work through the ENTIRE swing of both the positive AND the negative portion of the signal acting more like a "Class A" amp than a Class AB/AB1 amp! WHAT???? It's true... if this were a TRUE Class B amp, then each tube would ONLY work on its portion of the wave and the output would be HORRIBLY distorted.
So, in this example, a Marshall/Fender output stage might "extend" about 70% into the next tubes region. In the VOX AC30 the extension/overlap is closer to 80%... slightly further than most AB1 topologies. By the way THESE NUMBERS ARE NOT ACCURATE THEY ARE JUST INTENDED TO ILLUSTRATE THE POINT! DO NOT CREATE AN AC30 MYTH FROM THESE NUMBERS ;D
However, in Class AB and AB1 topology each tube amplifies PAST the 0V point INTO the region of the other tube. Class AB1 push-pull amps move FURTHER into that region than a Class AB amp but BOTH move into that region by some amount. Most guitar amps are biased in Class AB1 range and work for at least 50% of the other tubes signal swing, or more, depending on the bias set by the output stage designer.
This means you can pull all but one output tube and STILL get a perfectly undistorted signal! PLUS, the feedback voltage is near "0", because of the decreased output voltage from a single tube and not the full complement of tubes. This "loosens" the output stage and decreases damping causing the stage to be more "loose" and "flabby", a bit more "bluesy". Put the tubes back in and the feedback increases in correct proportion and "tightens up."
If you'd like to "loosen up" the stage, you can increase the value of the LARGER of the two resistors in the feedback loop. This will decrease the feedback voltage, loosening the stage up. You could, even, open this resistor up and reduce negative feedback to "0". So long as the amp is designed without too much phase shift it should handle this quite nicely.
If your amp is too loose, you can INCREASE the value of the smaller resistor in the feedback loop. This will increase the amount of feedback voltage, decrease gain as a function of that AND "tighten" up the amp a bit.
These are changes that can be made to the amp to change its response and performance, overall.
Enjoy... ask questions if you have them.
Dar