Nonetheless, if you used a fistful of electrolytic caps, resistors, and three IC's (two low voltage regulators and an Op-Amplifier, each of which houses an additional fistful of transistors and resistors) in place of a good ground connection you would, quite rightly, be judged an audio menace. However, do effectively the same thing at the B+ side of the amplifier and you now have a feature for the marketing people to seize upon.

               323.1k = 150k + 62k + (100+1)1.1k
which equals a voltage division of 0.54 at the plate, which is the output of this circuit. This means that whatever disturbance occurs at the top of this plate resistor shown in figure 2 will appear 0.54 times as big at the other side at the plate of the tube. Thus, whatever sonic contribution the regulator might add at the top of the plate resistor will appear 0.54 times at the output. If this had been a Cascode circuit, the math would have been much worse: basically, whatever appeared at the top of the plate resistor would have appeared at the bottom of it, as the Cascode suffers from virtually no PSRR.

What type of sonic contribution by the regulator are we worried about?
   At the very least, noise. Unfortunately, many three-pin IC voltage regulators are fairly noisy. This noise results from compromises made in the internal construction of the device; all IC regulators have within them a voltage reference, which serves as an electrical yardstick against which to measure the output voltage. This internal voltage reference is invariably inferior to an external discrete voltage reference, if for no other reason than that it cannot be bypassed by a large capacitor. The next obstacle in deriving a quiet and accurate voltage reference is cost. The price of a high quality external voltage reference can be ten times that of the IC regulator. The result is that a cheap and poor performing zener diode is what is used in many fixed IC voltage regulators. And a zener diode makes a great noise source, if noise is what you need.
   Static noise is obvious on an oscilloscope trace. Dynamic noise requires more scrutiny. As varying current demands are made on the regulator, the regulator can be induced into erratic behavior. It might start oscillating or momentarily shut off. Beyond this gross distortion, there can be a more subtle alteration of the output of the regulator. Maybe some range of frequencies provokes the regulator to follow them sympathetically or buck them, either of which will alter the frequency response of the circuit being fed. Or perhaps just the first half cycle of a current waveform provokes some aberration at the regulator's output, which quickly disap-pears. Such an aberration is unlikely to show up in a sweep test of the regulator output, but might be obvious to the ear while listening to the amplifier being fed by the regulator.
False Conclusions
   Will this webzine be regulator free in the future? No. Absolutely not. Forsaking any technology, even transistors, IC's, or pentodes, is needlessly limiting and ultimately foolish. All we wish to achieve is a wider eye look at how a regulator fits in an amplifier and how it may influence its performance. Your views on this subject are welcome.
                                     // Editor

Figure 2. High-Voltage regulator

    The circuit in figure 2 is the result of creating a feature by turning the circuit in figure 1 on its head. All the current that had to pass through the PNP transistor into the tube in the virtual ground circuit now travels from the tube into the emitter of the NPN transistor. Both circuits are just as much part of the amplifier as the other, as both circuits connect the amplifier to the power supply.
   Perspective is relative in electronics. What is called ground in one circuit might be called B+ in an other; for example, a +100 volt power supply versus a -100 volt power supply; ground is wherever you want to be. Beyond this semantic flexibility, there a mental flexibility; for example, we think of a light bulb as only a light source, but its filament can also serve as a coil in an oscillator. (You can use this concept to your benefit if a circuit eludes your understanding:  try redrawing it, since a different rendering means a different perspective.)

Low Zo
   As for the argument that since the low output impedance of a high feedback IC-based regulator is such a small percentage of the total impedance of the circuit, its contribution must practically fall out of the equation, the math, quite simply, is being stood on its head. The plate resistor and the rp of the tube added to cathode resistor's value times the mu of the tube plus one equals the total impedance of the tube circuit.
               impedance = Ra + rp + (+1)Rk
or in this case:

pg. 2

< PREVIOUS   Copyright © 1999 GlassWare   All Rights Reserved