[Amps] failures of tubes with handles

[John T. M. Lyles]

I just finished a grueling week working with a team of 6,  rebuilding 
two large amplifiers which were horribly damaged due to a bizarre 
failure mode in the controls and operator mistake.

Destroyed were two Thales TH555A tetrodes (250 kW dissipation each) 
and two Marconi/E2V BW1643J2 triodes (400 kW dissipation each) and 
the associated IPA and FPA stage which operated at 2800 KHz. Also one 
250 lb filament transformer, all water cooling hoses, and portions of 
associated grid resistors, anode feed choke, tube socket, etc.

Cause was a one point failure in the interlock system, which causes 
all 34 interlocks (water, air, temp, doors, and many more) to fail 
closed, in other words, bypassed. I was not the designer of the 
control system so I can critique it now. (Engineer is long gone) I 
had been suggesting a redesign with an Allen Bradley PLC or similar 
since 1998.

A very strange latching relay system had been devised, which used a 
diode steered reset pulse to latch the relays for 5 sec when reset 
button pressed. If an interlock then opened later, and LED turned 
red, and it remembered the fault until reset was pressed. A 1N4005 
diode shorted on 2/25, forcing the system to ignore interlocks. Then, 
an employee turned off the water pumps, which cool the anodes (and 
filament on the triode).  The HV and the RF had been turned off, but 
not the filaments. After 30 minute or less, the 7 kW filament power 
of the FPA triodes and the 4 kW filament power of the IPA tetrodes 
cooked the amplifiers, without water cooling to transfer away the 
heat. The paint was blistered off the tubes, and one tube acutally 
smouldered (paint) for some time.

It was a sad day to toss those tubes, and also to have to completely 
rebuilt the amplifiers with spare parts.
Cool your filaments well, especially when your tubes have handles or eye hooks.