www.tubecad.com           Copyright © 2003 GlassWare           All Rights Reserved

     Now, let’s look at some bad ideas. A trick used in cheap amplifiers was to forgo the expense of an input transformer or phase splitter circuit that used an extra triode, which makes a circuit more than casually similar to the SRPP. In the amplifier at the right we see the top EL34 being driven by the input signal, but the bottom EL34 receives its drive signal from the signal developed at the top pentode’s screen. What’s wrong with this amplifier? The degree of balance between tubes varies with the load impedance and both the first pentode’s distortion and bandwidth limitations are just cascaded into the bottom pentode. Furthermore, power supply noise does not cancel, as it would normally. Still, while the results were not perfect, they were good enough for many applications, such as cheap phonographs.

Text Box: The Journal of Tube Audio Circuit  Design
Text Box:  < Back
Text Box:     Next >

More ð

     The last amplifier design also carried the limitation of requiring the use of pentodes. On the other hand, the amplifier at the right can use either triodes or pentodes, as the bottom tube’s drive signal comes from a tap in the primary winding. This circuit is an even closer equivalent to the SRPP than the previous circuit. But it too shares the same shortcomings. (In a perfect world that held perfect tubes, transformers, loads, and power supplies, these circuits would work much better. Sadly, that’s not the world we live in.) The only advantage to these two circuits is that they force a class-A operation on the amplifiers.

     As a tube push-pull amplifier must see a balanced drive signal to work well, it looks like using the input transformer or a balanced line signal are our best bets.

    Or are they? The next amplifier requires a floating power supply, as it shifts its reference point from ground to the middle of the two 47k input resistors. This shift in reference creates a phase splitter out of the two resistors, allowing a single-ended (unbalanced) line amplifier to be used. This phase splitter is about as pure as it gets, as resistors do not limit frequency response or add the phase shifts of the inductively based phase splitters. But the demand made on the line in terms of needed voltage swing and gain will be great. In addition, any parasitic capacitance between the floating power supply and the grounded chassis can upset the high-frequency phase splitting (insulating washers might counteract this).

Simplest MOSFET Amplifiers

     Like the tube version, the simplest MOSFET amplifier should hold only two MOSFETs. And like the tube version, the MOSFET amplifier can make use of input and output transformers to keep the part count down to a minimum. Unlike the tube version, however, the MOSFET amplifier doesn’t require such large winding ratios or such high voltages, as the MOSFET has several magnitudes of extra transconductance over the. 

    The amplifier at the right uses an input transformer to split the phase of the input signal and to develop the required voltage gain. The two output MOSFETs are configured in a source follower topology, which provides no gain, but does provide a wide, a low output impedance, and clean output swing. The low winding ratios in themselves make for a better transformer and this amplifier’s output transformer can be as low as unity wound, 1:1, which would greatly lower the transformer’s imperfections. The only obvious problem with this amplifier is the high input capacitance. Once again, the input capacitance would be magnified by the input transformer. A small quibble might be had with the need for two power supplies, a main and a bias power supply.

    Adding three resistors eliminates the need for a bias power supply. In the amplifier at the right we three resistors in place. Note that the input transformer’s secondary is being loaded by the two 280k resistors in series. This is actually a benefit, not a determent. Each transformer has its optimal load impedance, which is easily found with a square-wave generator, a potentiometer and a scope. As 560k is unlikely to be the right value for every (or any) input transformer, an extra loading resistor can be placed directly across the secondary. Also note that the single 10k resistor allows the idle current to be set by varying its value and that this resistor could easily be replaced with potentiometer or, better still, a potentiometer and a resistor wired in series.