ohms? Would that not be as insane as saying the thickness of the wax on my bald head makes me seem taller?"
   The rebuttal begins with the observation that since the regulator is in the same current path as the audio signal, it can contribute to the final sonic signature of the amplifier just as much as any other component in the same path. A general rule to follow is: if the audio signal could possibly flow through a device under scrutiny, worry about its effect on the signal.

Welcome to Issue Three
    This month's issue continues the noise reduction theme: one article on a noise reduction technique for fixed-biased single-ended amplifiers and another explaining the how and why of  voltage regulation. Even this month's Design Idea, a para-feed Dynaco ST-70, furthers the noise reduction theme, as the greatest sonic advantage para-feed holds over the more conventional designs is a very good PSRR.
    A little bit of personal history: I used to design all my tube circuits with a regulated power supply. I took pride in the complexity and performance of those regulators. If the B+ could not be regulated, then at least the heater supply or the negative power supply for the grid bias would be. The results were mixed; most projects worked well and sounded better than most, a few were very unreliable, and one or two sounded worse with the regulation than without.
    Discovering the sources of unreliability took some sleuthing, but was never unsolvable. For example, if an IC manufacturer specified a mediocre capacitor in its high voltage regulator design, it wasn't that it was being con-sistently cheap. Sometimes a cheaper part effectively comprises an unspecified component. In one case, the capacitor's ESR defined a needed resistance crucial for non-oscillatory operation. Replacing the cheap capacitor with an expensive polypropylene capacitor resulted in transforming the regulator into an oscillator. 
    Finding the cause of poorer sound with a regulated power supply was much more elusive. My reasoning was that the regulator hadn't regulated enough, that if the output impedance were lowered from one tenth to one thousandth of an ohm, the sound would have to improve. It might have, but I doubt it.

Regulators and the signal path
   Recently, old tube regulator circuits have been resurrected as an alternative to the often fragile and poor sounding solid state regulators. "Poor sounding?" I can see the eyebrows crawling up the bald head of the aged audio reviewer. Can't you hear him saying, "Voltage regulators don't sound at all, period. No more than the LED on the front panel does. How could a modern device with an output impedance of 1/100 of an ohm, bypassed with a large electrolytic capacitor influence the sound of such a high impedance circuit as one made up of a 100k resistor and a tube with an rp of 60k

    The circuit in figure 1 helps demonstrate the point made. Here an artificial ground is being derived from voltage dividing equally between a plus and minus 15 volt power supply, which yields 0 vdc at the midpoint. Then this 0 volts reference pseudo ground is capacitively coupled  to the Op-Amp's positive input. The emitter of the PNP pass device

Figure 1 Virtual ground circuit.

yields a very low output impedance and the IC's low input noise could possibly make for a quieter "ground" than two feet of poorly laid out wire which hooks up to a screw on the back of the chassis. What we have here is a ground amplifier; its output is a copy of what exists on the actual ground. This circuit constitutes a virtual ground. The current path as shown in Figure 1 begins at the -15 vdc power supply rail and travels through the PNP transistor, then through the cathode resistor, the triode, the plate resistor, and finally ends at the +300 vdc power supply.

4
6
8
10
11

Lowering the SE Amp's Noise  -part 2-
Tube Voltage Regulators
Design Idea: Para-Feed Dynaco ST-70 
E-mail
Publishing Information
Past Issues: March  April
Glossary of Audio Terms 

NEXT >