This is my first attempt at making a triode. It was only a crude attempt at making something that behaves as a triode, as there are a number of tools and components I don't have yet that would be required to make a truly functional triode (namely a high vacuum pumping setup, spot welder, annealing oven, induction heater, nickel sheet metal). For the most part, construction of the triode resembles that of my initial attempts at making incandescent light bulbs (see Homemade Light Bulbs (failed initial attempts)). However, the seal between the outer envelope and stem was made by soldering the two parts together with an indium alloy. This method was used because I wanted to see if it is possible to make good vacuum seals using indium without too much effort, and this seal is also the most difficult to make, and is very prone to cracking. Since I do not have an annealing oven to fully relieve the stress, this makes the seal less risky (I'm using lead glass, which has a high thermal expansion).

The lead wires consist of a short section of 0.35 mm borated Dumet wire (roughly 5 mm long) butt-welded to 26 AWG (0.0159 in / 0.405 mm diameter) nickel 205 wire on the vacuum end, and 22 AWG stranded copper wire to the outside (the wire is tinned, so it looks silver). To butt weld the wires, a 32,000 μF capacitor is charged to 20 volts, and shorted out with the ends of the two wires touching end to end in a glass capillary. 32,000 μF at 20 volts was somewhat arbitrary, as I just pulled a large capacitor off the shelf.

Once made, the ends of the four lead-in wires are twisted together, such that the Dumet sections are spaced evenly apart for sealing in the glass. The end of the lead glass stem is heated at the end of a highly oxygenated flame until soft, then it is squashed between two graphite pads, sealing in the Dumet wire sections, as well as a small portion of the non-Dumet wires for mechanical support. The glass to metal seals are then heated until the glass glows orange to ensure a good seal to the Dumet wires.

The cathode is a coiled tungsten filament, which is supported by the nickel portion of the feed-in wires. The grid is simply a spiral of nickel wire surrounding the filament. Both of these can be seen more clearly by clicking on the images below to view them in their full resolution. The plate is made from 0.001 in aluminum foil, which is folded over one of the nickel lead-in wires for mechanical support and electrical contact.

An indium alloy was used to seal the outer envelope to the stem. Before joining the two parts, the ends of each were coated thoroughly inside and out with the solder in order to improve the chances that a good seal is located somewhere all around the joint. Once both parts have been pre-tinned they are joined, adding liberal amounts of solder to ensure there are no small openings. For more on using indium to solder glass, see Indium Solder Glass to Metal Seals.

Before evacuating the triode, a section of the 3mm tubing is thickened such that the inner diameter is reduced (keeping the same outer diameter), which will later be heated for the vacuum seal. The reduced inner diameter and thicker build up of glass allow for a stronger seal that does not get sucked inward by the vacuum.

Ideally, to pump down the triode a high vacuum pump such as a turbomolecular or diffusion pump would be used. To speed up pump-down the device can be purged with Argon, which is a heavy gas that is easy to pump, as it will push out contaminates as it is pumped away. To outgas components inside the tube, the outer envelope can be heated in an oven (not possible with this device due to the indium seal), the plate can be heated with an induction heater (not possible with the aluminum foil in this device, as it would melt), and the filament should be flashed to burn off contaminates. It is useful to have a getterer in the tube to eat up any contaminates that remain after tip-off.

Other than flashing the filament during pumping, I did none of these things, so the vacuum is without question soft. The pump used was a Welch DuoSeal 1400. It has an ultimate pressure rating of 10-4 Torr. Because the pump is brand new with a fresh charge of oil, it is hopefully getting down to the 10-4 Torr range in the tube after 30 minutes of pumping (I don't have a gauge yet to check this).


While the device does behave as a triode, performance is far from impressive. The filament was heated with 25 VAC at 390 mA. Plate currents are very low, and are detailed in the table below.

VAnode (V) Vgrid (V) IAnode (nA)
150 -20 3.2
150 0 4.2
200 -20 4.2
200 -10 5.4
200 0 6.5
200 +10 7.2
200 +20 7.9