Both circuits lessen the noise by injecting power supply noise into the circuit. The first version is preferable as it allows for precise nulling of noise from the output of the entire circuit. The second circuit requires one less capacitor and is conceptually elegant.
  The circuit shown below is also a hybrid and it too uses the poor PSRR to good advantage. The FET first stage has a PSRR figure of zero, but as its decoupling capacitor is referenced to the negative power supply rail, the noise from that rail is relayed to its drain through its load resistor (2k). And then this this noise is further relayed to the grid of the bottom triode.
  The zener diode only serves as an emergency voltage clamp to protect the FET from the +100 volt rail and to prevent the FET's power supply voltage from dropping more than 0.7 volts at turn on, when the capacitor has yet to charge up to normal voltage. Once again a little paranoia goes a long way in saving the voltage sensitive solid-state devices from blowing. Thus, the diode that bridges the top triode's cathode and grid not only protects the triode's grid from seeing too great a positive voltage at start-up when the tube is cold and not conducting, but it also serves to keep the output voltage close to ground potential, should the tube be removed from its socket during use.

   The amplifier shown above has three gain stages, but only one capacitor in the signal path. The real twist, however, comes from the grounded grid of the output tube. Normally, the cathode is grounded and the grid is driven with the signal. Here the cathode receives the signal and the grid is locked at ground voltage. This twist gives the amplifier several real advantages.
  The first is that the Miller effect capacitance does not obtain, as the plate's voltage movements are shielded from the cathode by the grid. Consequently, the amplifier will exhibit much more extended high frequency bandwidth than would usually be the case. Of course, in most tube power amplifiers, the output transformer limits wide bandwidth.
   A second advantage is that the grid can never be driven positive relative to its cathode, as the P-Channel MOSFET cannot go anymore negative than the ground. In fact, it is doubtful that it will even get close, as the 6DJ8 must have some voltage across its cathode and plate to conduct. If greater negative voltage swing is desired, a small negative power supply of 25 volts could be added to increase the headroom.

pg. 7

<PREVIOUS

NEXT >