In an obscure corner of the Radiotron Designer's Handbook there is a brief mention of a rather interesting concept: phase inversion in the power amp through common-cathode coupling. Despite eliminating an entire stage from the amplifier, which sounds really cool, the idea just never caught on. It has, after all, two major drawbacks for audiophiles: the power amp breaks into distortion at much lower volume levels and the lack of a separate phase inverter increases distortion.

Achieving full power at lower volume sounds downright appealing to those of us who love big power tubes but don't play the football stadium circuit. As for distortion, well, guitar players aren't known lose sleep at night over a few extra harmonics. So perhaps this circa 1953 concept deserves another chance.

push-pull power amp without a phase inverter

Take the concept of a long-tailed-pair phase inverter and merge it into a push-pull power stage and you've got "phase inversion through common-cathode coupling." It's terribly inefficient, but it's an order of magnitude greener than a dummy-load soaking up 50 watts of wasted heat. Moreover, it eliminates one whole stage from the design, which makes it architecturally efficient.

The key concept is the DC design: more voltage between the plates and cathodes creates more power. More voltage across the tail creates more balance. To put some hard numbers on it, let's see what we can achieve with the venerable 6L6 beam power tetrode.

DC Circuit Design

We'll use a relatively small 125-0-125 power transformer to create VPP = 177 volts at the plates and a choke to keep about the same voltage for VSS. The screen voltage measured from screen to cathode is going to be substantially less, because we'll put most of the available supply voltage between the cathode and ground.

Here are the average transfer characteristics for a 6L6GC, a big tube we'll task for this relatively small job:

6L6GC average transfer characteristics

These curves are for a plate voltage of 250 volts, but sufficiently accurate for our purposes because screen voltage plays a much more important role. If we set the idle screen voltage to 50 volts and use the 50-volt curve to set the tube approximately halfway between cutoff and saturation, we get the DC operating point shown in blue, where the plate current is 10 milliamps and the grid voltage is -3.5 volts. The total plate current through the tail is double: 20 milliamps. By Ohm's Law this means that the cathode resistor value is RK = 175 ohms. (For my bench tests I use 200 ohms).

A signal amplitude of 3.5 volts measured between the grid and cathode drives the amp to full power. Twice as much is needed between the grid and ground, so the preamp needs to provide at least a 7-volt signal amplitude to send the power amp into overdrive. This is substantially less than the 50-volt requirement for a typical 6L6 guitar amp operating in Class AB.

There is 50 volts from plate to cathode and 3.5 volts from cathode to tail, leaving 124 volts across the tail itself. To get the tail resistor value we again consult Professor Ohm and get RT = 6.2k. (I use 5.6k on the bench.) We need to worry about exceeding the maximum cathode-to-heater voltage, but for the 6L6GC it's 200 volts, so we've got about 70 volts to spare. 124 volts and 20 milliamps represents 2.5 watts of heat, so I use a 5-watt tail resistor and give it plenty of air.

We can use the zero-grid voltage curves to determine the optimum output transformer primary impedance, but we're operating way down at the lowest curve, so there is not much resolution. I'm estimating a peak current of 18 milliamps and a minumum plate voltage of 18 volts for this load line (the tiny line in blue through the knee of the 50-volt curve):

6L6GC average plate characteristics

The voltage swing from idle is then 50-18 = 32 volts, for an impedance of 1.8k. We're operating in pure Class A because each tube conducts plate current through 100 percent of every cycle. The plate-to-plate primary impedance for this load line is thus double: 3.6k. Max power is 576 milliwatts for a square wave, 407 milliwatts for a sinewave. (4k works well. On the bench I use a 10k plate-to-plate transformer for less overall volume in overdrive. At these power levels the impedance isn't critical.) Here is the final configuration:

push-pull power amp without a phase inverter final values

Some of the smoothness of this amp in severe overdrive, I believe, is directly attributable to its pure Class A configuration and high gain from input jack to 6L6 grid. Even Tony Iommi might want to back off on the volume control bit. Langford-Smith couldn't have imagined what "common-cathode coupling" might lead to...


Here is the schematic of the complete amp for my bench tests:

Low-Power 6L6 Guitar Amp

The voltages shown are for my particular tubes. Here are the data sheets:

6L6GC datasheet
EF806S datasheet