Page Title

Simple MOSFET SRPP

The circuit to the right is a class-A push-pull amplifier that forgoes using a separate phase splitter or a balanced input circuit. Instead, the current the bottom MOSFET draws through the 0.65-ohm resistor defines the drive signal for the top MOSFET. This resistor should in theory be equal to the inverse of the top MOSFET’s gm, but published specification is unlikely to match the actual gm at the idle current, so some experimentation is required.

The output must be capacitor coupled because of the uni-polar power supply, but with sufficient bypassing, the capacitors should not harm the sound and will actually offer a nice safety margin. Once again we are back to a low input impedance, so it is worth looking into the trick of using a negative power supply rail to allow a larger set of input and feedback resistors.

More ð

Simple Hybrid Push-pull Amplifiers

The simplest hybrid amplifier, in terms of low part count, might be to take an existing solid-state push-pull amplifier and drive it from a single tube gain stage. Adding a feedback loop gives the tube control over the solid-state amplifier. The circuit below could form the basis of an excellent computer amplifier, as only one tube and two ICs are needed, all of which could easily fit into an aluminum removable hard drive holder. The LM12 is a robust four terminal op-amp that can deliver 10 amps of current with full short-circuit protection.

Alternatively, this technique could be applied to a wimpier power amplifier for headphone use. In fact, a 6GM8, a ±12V DC-DC voltage converter, and a dual power amplifier IC could be held in an Altoid tin, while an external battery pack could hold a lead-acid 6V battery or four D cells for those times when a wall wart isn’t available. The 6GM8 is a twin triode that looks like 6DJ8 on the outside and the inside, as it too is frame-grid triode. It differs from the 6DJ8 in that is has a specially treated cathode that can emit a strong electron stream under only 10 volts.

Of course, many would as soon rather drink red wine with fish than use an IC in their audio equipment, even if that IC were as slick as an LM12. So something more difficult (I mean “discrete”) is needed. The amplifier below uses tubes, transistor, and MOSFETs. The triode works into a current mirror that reproduces the current fluctuations in the triode in the PNP’s collector, which is loaded by the 5k resistor. The signal that was inverted at the triode’s plate is then inverted again by the transistor. The gain is much higher than you might expect, as the triode is not loaded by the 5k resistor, but by the 200-ohm resistor, which means that the triode loses little of its transconductance, which means the triode undergoes fairly large current swings, which in turn the transistor relays into the 5k resistor, much as a cascode circuit would. In other words, because the 5k resistor is not in series with the triode, the triode’s transconductance is not reduced; more transconductance means more gain. This amplifier uses no internal coupling capacitors, but it does require an output capacitor.

It looks like the 5k resistor serves as a feedback resistor, but the amplifier’s gain is far less than the gain the 100-ohm and 5k resistors would imply. If the transistor were loaded by a constant current source, then the 5k resistor could attach to the output and work nicely as a feedback resistor. The diodes are used as voltage drops and the string could be replaced by a voltage reference or, gasp, a zener. One problem with circuit is that its PSRR isn’t all that good. Using a pentode or a cascode in place of the triode would help, but there is a better way.

The next circuit retains the current mirror and uses a bipolar power supply for the output stage. (Note the lower rail voltages.) The bottom triode effectively equals a 10k resistor, which allows a good deal of gain to be created. The bottom triode also allows a path to the negative power supply’s noise, which will then counter the positive power supply’s noise and cancel at the 250-ohm resistor. Nice trick. All in all, not a bad design.

Here are some possible embellishments. Feedback can be easily added from the top triode’s cathode to the output. A DC servo could be added to keep the output centered at 0 volts. A lower mu triode, such as the 5687 might be fun to listen to. And finally, the PNP transistor could be replaced by a P-channel MOSFET.