One important note is that both the single-ended and the push-pull circuits that use auto-bias must run the output tubes in strict Class-A mode. The Op-Amps will strive to maintain a constant average current flow through the output tubes and only the Class-A amplifier draws a constant average current flow. A second note is that this circuit will not work with tubes like the 6L6 or 6V6. These tubes require a much larger grid voltage swing. In order to use these tubes, an Op-Amp which can withstand a higher power supply voltage differential will be needed along with an increased negative power supply voltage. An alternative approach might be to use the lower voltage ICs  in a cross-coupled configuration.

The Usual Way: T>SS
    The common approach is use tubes for voltage gain and MOSFETs or transistors for current gain. This arrangement appeals to our sense of fairness: tubes are happiest with voltage and woefully current limited; solid-state devices are current ebullient but fearful of over voltages. An exaggeration this, but not without a kernel of truth. Surely, you have read the advertising copy: the Mega X837 can put out peaks of 100 amperes! As its power supply holds 40 volt rails, the implication is that it can put out 2000 RMS watts. It can't, unless you have a .4 ohm speaker. 100 amperes is of little importance for those own 16 ohm speakers, which can only see a peak current of 2.5A into 40 volts (i.e. 50 watts into 16 ohms). This situation finds a complement in the ad copy for some tube OTL amplifiers: The Ultra-Mega Blue Swan puts out voltage peaks of 140 volts. Impressive, but of little importance to those who own 4 ohm speakers, a load that runs into the amplifier's 2A maximum continuous current output limit all too often (8 watts into 4 ohms).
   (If you are thinking that the world needs a good 1 ohm and a good 100 ohm speaker, you are right. A compromise might be to make loudspeaker drivers that held four independent voice coil windings each with an 8 ohm impedance. All four windings wired in parallel gives a final impedance of 2 ohms; all wired in series, 32 ohms; and all placed in a series-parallel combination, 8 ohms.)

Transistor or MOSFET or IGBT?
     Which solid-state device should be used in the output stage? The transistor wins in theory, but the MOSFET wins in practice is the quick answer. The transistor has a much greater transconductance which makes for a greater gain, which in turn allows for more feedback, which ultimately lowers more distortion. Yet, the MOSFET is used in most professional, high-power amplifiers because the MOSFET does not blowout as readily as the transistor does. Furthermore, the MOSFET is almost always used in hybrid designs due to its reputation for being more tube like in sonic character and because its gate represents a high impedance that is theoretically easier to drive than the transistor's base. (I do not think that MOSFETs amplifiers sound anything like a 2A3 or even a triode connected pentode like the EL34. And I find its high gate capacitance a true nuisance.)

     Beyond auto-biasing, a further advantage that a solid-state front-end holds over the vacuum state equivalent is less noise. While a power amplifier is not as noise sensitive as a phono preamp, lowering the noise floor can only help make the qualities of the output stage more apparent. Simple thermal noise can be overcome by paralleling many tubes; microphonics can be lessened slightly by isolating the tubes from the chassis and by damping the envelopes, but microphonics is only really eliminated by finding low microphonic tubes, which is not easy. Double mica spacers are certainly a move in the right direction as are steel clamping-clips internal to tubes such as in the 5751. Still, vacuum tubes too much resemble tuning forks in construction: one end fixed, the other end free to resonate. DC power supplies for the heaters also help. But ultimately, the solid-state device wins. So why not use the solid-state device for those small fragile signals until they can be passed on to a tube? A single FET input stage perhaps? I understand that even Jean Hiagra uses a solid-state pre-pre-preamp (this datum is for the benefit of those who insist on designer labels).

< PREVIOUS

NEXT >