the greater the noise at the output. If the transformer sees half the power supply noise, then the output stage will have a PSRR of -6 dB, or 50%; in other words, very poor indeed.
    Here's the rub: the low rp, which proved wonderfully advantageous in lowering the amplifier's output impedance, now proves to be a hindrance to low noise operation. (Of course, feedback will lower not only distortion and Zo, but noise

How an SE Amplifier Works
   The SE amplifier can only operate in pure Class A, as the whole output waveform must be entirely handled by the single output device (if parallel tubes are used, each still sees the entire waveform). This arrangement makes for a more linear output from the tube, as it requires that the tube be run at a high idle current comfortably up from the non-linear portion of the tube's operation near cutoff. However, this arrangement also makes for a poorer PSRR figure, as the SE amplifier does not benefit from the noise nulling within the output transformer as the push-pull amplifier does. Furthermore, there is an implicit tradeoff between a low Zo and a high PSRR figure.

as well. Still, the lower the noise prior to the application of feedback, the lower the noise at the output after feedback.)
   High impedance devices like pentodes yield much smaller amounts of noise to the speaker because the primary impedance is so small compared to their extremely high (nearly infinite) internal impedance.
    One solution then would be to use a pentode in the output stage and wrap a healthy amount of feedback the output to lower the Zo and the distortion of the output.  While this

Triodes vs. Pentodes
    In the simple Grounded Cathode amplifier, the plate resistor defines the load into which the tube works. If there is any noise at the power supply connection, some of it will make itself known at the output, as the plate resistor and tube define a voltage divider. The lower the rp of the tube, the less noise at the output. Low rp is one reason why triodes can be much quieter than pentodes. For example, a 6BX7 has an rp of 1300 ohms and it is rare for any

pentode to have an rp of less than 50k; in fact, most are closer to being ideal current sources, i.e. near infinite rp. This low rp of triode is a great boon, when the circuit is a Grounded Cathode amplifier and the load is a plate resistor. But in a transformer loaded SE amplifier, the low rp now

approach would work, it hardly appeals to those who dislike pentode sound and/or feedback. (Here is where the parallel feed arrangement works well to mitigate power supply noise from the output, as the output transformer's primary has one attachment to ground and the other at the tube's plate: it only sees the noise present on the plate, which is buffered from the power supply noise by the high impedance of the loading choke.  Of course, this means an additional choke and a coupling capacitor. If the output transformer spans the B+ connection and the plate, nothing will be gained from the parallel feed configuration in terms of noise reduction.)

proves to be a detriment to low noise operation.
   The signal voltage that develops across output transformer primary, once it has been stepped down through the transformer, is what the speaker will see at the secondary. The greater this signal is, the greater the output will be. So too, the greater the noise voltage developed across the transformer,

The Trick
   What we are hoping to achieve here is to cheat the system: to use low rp triodes in the output stage and still enjoy low noise playback. How can noise be reduced from the output?
   The brute force approach would be to eliminate the noise at its source: to use huge power supply chokes and reservoir capacitors, to filter and double filter, and even possibly, to

< PREVIOUS

NEXT >