The first DHT amplifier that I built was a 71A. It was a bear to tame. The low power output (0.7 watts) dictated extremely high efficiency speakers, which were in turn very susceptible to AC filament hum. The 71A did me no favors there as it operates at twice the filament voltage as its descendants, the 45 and 2A3. Despite its shortcomings, there was something special about the sound. I loved listening to it through the most efficient speakers that I had at the time – a pair of Coral Fe103s in a large infinite baffle setup. It sounded better and different than anything else I had built to that point, but in the end, it just didn’t have enough ‘oomph’. I ended up re-wiring it for 45s before passing it on to its next steward.
In the years since, I have thought about that amp a lot. With the Granit Prime Circuit turning into a Rube Goldberg Masterpiece, I recently decided to take a step back and give the 71A a second look. Two things would have to be overcome: Filament hum, and the anemic output power. DC filaments would be the easy answer to the first, and the tube was designed for battery operated sets in the first place. As for output power, there is only one way to squeeze more power output from the tube – operate it in Class A2 by allowing the grid to be driven positive with respect to the cathode. I can find no examples of this having been done before with the 71A, but it has been done on both the 45 and 2A3 with good success. Peeking inside, the internal construction of the grid looks substantial enough that it should be able to dissipate a fair amount of power without evaporating itself.
Starting with the RCA datasheet curves, I plotted out my best guess at what the gridlines would look like for +10, +20 and +30 volts. I’m sure they aren’t exact; they may not even be close. Next, I super-imposed the maximum plate dissipation curve (5 watts) onto the graph. Plotting a 5KW load line, I snugged it up right against that maximum dissipation curve. A lower load impedance would give more power, but at the cost of linearity. With a grid swing of +/-50V at a peak value of +30V yields a nearly symmetric plate swing of 230V – good enough for 1.32 watts! that sounds even bigger when you call it out in the vernacular of the original RCA datasheet: 1320 milliwatts! The plate curve with these operating parameters is shown below.
Now the question becomes – Is this accurate? And, if it is, can it be achieved? To get the ground truth, I wired up a NOS 71A Sylvania Coke Bottle tube with the basics. AC filaments, Tamradio output transformer, HV power supply, and a reworked version of my Skylark drive circuit. Adjusted to the idle point shown above and with +/- 50V of grid drive. Maximum clean power output came out even higher than predicted @ 1430 milliwatts. In the curve trace below, the top crest of the waveform is the positive grid peak. The flattened bottom trough of the wave indicates the distortion is coming from the negative grid region where the plate curves start to compress together. Harmonic content at this output power is still a very clean 1.28%. For comparison, a 45 driven to its maximum class A output with the same load impedance and grid swing predicts 1320 milliwatts @ 1.9% THD. I am going to give an emphatic ‘Yes!’ that the 71A can absolutely be up-gunned to match any SE 45, and probably some of the 2A3s out there as well.
Oh – yeah, about the sound. Even though it is mono, and sloppily wired, it is magical. All of the sweet sound that I remember, but with the punch that was missing before. I can’t wait to get this built out right and in stereo. You can follow the design and build progress on my ‘Oncilla’ page – replacing Granit Prime.
Foxtrot 9 Audio 