This page is an extension to the large page on Vacuum Tubes, containing additional experiments and information as they develop.
The type 31 power triode is one of the last examples of a type of amplifier tube that was replaced almost completely during the 1930's by the power pentode. This type was designed for use as the power amplifier driving the loudspeaker in small battery-powered radio sets. Nevertheless, it has the same general characteristics and was used in the same way as more powerful tubes, such as the 2A3, a larger power triode. While the 2A3 could supply up to 15W as a push-pull amplifier, the 31 was generally used singly, with an output up to 375 mW.
Power triodes are characterized by a low μ, of 5 or less. They were intended for power amplification, not voltage gain, and had to be preceded by a driver stage that would produce the necessary grid swing, usually a medium-mu triode like the type 30. Both tubes could be replaced by a single pentode, which had high gain as well as power output. Nevertheless, power triodes had certain advantages as amplifiers, among them low distortion even without feedback, and low Miller effect. They are still prized by vacuum-tube audiophiles, and all power triodes are now very hard to find and quite expensive. Even the 31 was a bit expensive, although not used in hi-fi amplifiers, but a sale price made it more accessible.
The 31 was designed for receivers using lead-acid cells as a power source. A vibrator and rectifier could be used for the B+ supply, while the filament was supplied directly from 2V DC. The oxide filament was very efficient, taking only 0.13A for a filament power of 260 mW. The type 30 medium-mu triode required only 0.06A, since the plate current requirement was less than for a type 31. This should be compared with the normal 6.3V at 0.3A, or 1.89W, for a heater-type cathode. The 31 was made in later days with the graceful ST12 envelope, but retained the typical small 4-pin base traditionally used for filament-type triodes. The type 30, with a similar envelope and base, was modernized as the 1H4-G and given an octal base. The type 31 never received this treatment, since it was already obsolete and used only for replacements.
The basing is shown at the right. The two large pins, 1 and 4, were always for the filament, pin 2 for the plate, and pin 3 for the grid. This basing was standard on all filament triodes with a small 4-pin base. There was a type 11 with a different basing, but its 4-pin base had only one large pin (2, the plate; the filament was on pins 1 and 3.). The similar type 12 had normal basing. Maximum plate voltage is 180. At a grid voltage of -30V, the plate current was 12.3 mA. The large grid voltages are characteristic of low-mu triodes.
The first thing I did was run some characteristic curves. A carefully preset variable DC source was used for the filament. In the dark, the faint red glow of the filament could just be seen as an inverted V, looking through the top of the tube. You have your choice of which end of the filament to use as the cathode connection. Which choice you makes changes the grid bias by 1 V either way. I made pin 4 +2V and used it as the cathode connection. This makes the average grid bias 1 V below the cathode-grid voltage, which I did not take into account, merely labelling the grid voltage by the grid-cathode potential. I used grid voltages of 0, -2, -6, -10 and -14 V, with plate voltages from 10 to 120 V. Both voltages are limited by my tube-testing apparatus, but provided sufficient range for good results. I did not let the plate current reach 20 mA, so dissipation was always under control. In fact, the plate appears to be unblackened, shiny nickel. The coarseness of the grid winding was noticeable; this is what gives the low mu. I found an amplification factor of 3.75 and a transconductance of 1.050 mS, both almost spot-on the specifications. This is remarkable for a device that is probably 60 years old, and shows the very high point that vacuum tube design and manufacture had reached by the 1930's.
Next, I wanted to make a power amplifier. I chose a plate supply of 180V, and a load line that passed through 30 mA, corresponding to a load resistance of 6k. The quiescent point was determined by the grid bias of -14 V, which gave a plate current of 11.6 mA at a plate voltage of 111 V. For a bias of -14 V at 11.6 mA, a 1200Ω cathode resistor (1/4W) was required. I did not bypass the cathode resistor. Using a load resistor, a 28 V swing at the grid (the maximum without driving the grid positive) gave a 70 V swing at the plate, a voltage gain of -2.5, and a power output of 204 mW. Had I used a transformer converting an 8Ω speaker to a 5k load, and bypassed the cathode resistor, I probably could have obtained a larger power output. At any rate, the amplifier worked very well.
Composed by J. B. Calvert
Created 18 April 2003