When we asked people about the rugged WW2 American metal valves, we received mixed comments on how they were made. So we planned a dissection.
The museum had three specimens of the metal 6J5 triode. So the one in worst condition was chosen for sacrifice.
Please note that breaking the vacuum on any valve is a potentially dangerous action. We took special care with closed containers, gloves and goggles. If in doubt - don't.
Although metal-envelope octal valves were certainly widely used in American military equipment during WW2, the metal type was actually developed by RCA for domestic valves and was well established by 1935, at least 6 years before the US entered the War. In fact, RCA based their metal valves on the British Catkin types pioneered by MOV in 1932-34. The RCA types, however, used a metal envelope which was not connected to the valve anode. This enabled them to make metal versions of two-anode valves such as the type 5Z4.
The majority of equipment of the time used large classic glass envelopes. The International Octal base was introduced by RCA for their metal valves.
The octal base was soldered onto the wires from the envelope, but we chose not to use heat in case we damaged the vacuum seal. The plastic octal base was fixed to the envelope by 4 pinched-over lugs on the outer skirt. Opening the lugs released the base. At first we wondered if a glass envelope would slide out from the metal envelope. It had been considered that the metal envelope was just a protective screening can.
Nothing came out with small force applied. As the base rotated slightly it was expected that the wires from the seal went to the base pins in the same way as glass octal valves of the period. The latter would be of the pinch construction.
To remove the plastic base, force from a screwdriver blade twisted between base and metal skirt was applied. This was done gently at four points round the base. Some slight splintering sounds emerged but the wires broke quite easily.
As the base was removed we found our first surprise. The wires emerged from a 17 mm flat glass base. Miniature valve size, nothing like the 1930's octal valves with their tall pinch construction. Inspection with a magnifying glass showed that the base was sealed into a metal tube, that itself appeared to be sealed into the envelope. The possibility was that we had concentric metal envelopes.
To find out what was inside required cutting into the main envelope. The vacuum seal was a glass tube that emerged from the centre of the glass base and fitted into the octal spigot. Taking great care the seal was scored with a file and snapped open. The sharp edges were immediately smoothed for safety.
The envelope was gently cut open with a hacksaw. The blade being moved round the circumference to ensure that the envelope was only just cut and that no internal structures were damaged. During this process it was found that two rod-like supports touched the envelope. It was also clear that the envelope was a single thin mild steel shell. No glass lay beneath the shell.
With the circular cut completed, the envelope cracked into two. The main envelope was plain inside without any obvious coating. At the base end the view was obscured by the mica supports. The rugged construction could be seen from the rod clamps from the base to the top of the electrode assembly. These formed an 'A' frame and were quite rigid.
Cutting the supporting rods was accompanied with cutting the connection to the box like anode. The anode enclosed the rest of the electrodes on four sides. The space for the other electrodes being rectangular.
Before the anode was removed, the top mica support was gently pulled off. It was a surprise to find that the Cathode tube came with it. The tube was about 1 mm in diameter and the white oxide coating was clearly visible.
As the anode box was removed the grid and heater became exposed. The grid is a fine wire helix of oval cross section held on two vertical rod supports. The fine grid wire is held in place in notches in the wire supports.
The heater looks to be a long coil that is bent into a hairpin shape and then gently twisted together. Being coated in insulating powder the actual structure is hard to determine.
The exceedingly flexible nature of the heater was very apparent. After taking the pictures the re-assembly of the cathode tube onto the heater was much like threading a needle.
Although the RCA design of metal-envelope valve was far more versatile, and commercially far more successful, than the British Catkin type, it had a serious limitation in that the metal envelope had to be kept at a voltage below that of the cathode. Otherwise the envelope would behave as an anode and collect electrons. For most purposes this limitation did not matter since the envelope was normally 'earthed' to chassis and hence to the negative return of the HT voltage. However, professional equipment, including radar, etc., sometimes required circuits in which valve cathodes carried large signals and had to float at voltages well below that of the chassis. Glass-envelope '-G' versions of common types were therefore required. Initially, these '-G' versions were larger than their metal counterparts but with the introduction of pinchless, slimline '-GT' types 5Z4GT the advantage of the compact metal types was reduced to the point where RCA decided to phase out production in favour of (slightly cheaper) all-glass types.