Appendix G: The UHF Power Amplifiers

Apart from the fact that they used klystrons rather than more usual triodes or tetrodes, the amplifiers themselves were of straightforward and sound construction. The klystrons themselves were a novelty - each 25kW unit used two, one for vision and one (heavily under run) for sound. The reason behind using identical klystrons for sound and vision was solely to reduce the spares stock holding required. Also, when a vision klystron became 'tired' and incapable of further service, it could be happily run for many thousand extra hours in a sound position. 'Tiredness' often took many years to arrive: klystron lifetimes of 50,000 hours were by no means uncommon.

The one shortcoming of the design revolved around the klystron cooling. Because these were very early high-power devices, the collectors had to be cooled by a pumped liquid coolant circulated through a cooling jacket around the collector. Later designs exclusively used much more efficient vapour cooling. The trouble was, there was a lot of heat to get rid of (early klystron designs were only about 35% to 40% efficient in power conversion terms, so a 25kW klystron could be generating nearly 50kW of waste heat), and this design specified heat exchangers mounted external to the building to remove the heat. Due to the long piping needed, pump pressures had to be high to get the required flow-rate - about 90psi. Because the heat exchangers were mounted externally, plain water couldn't be used as a coolant, else on a cold night it would freeze solid. What was used was a fairly strong ethylene glycol mixture ('antifreeze' to you motorists, except the stuff used contained special inhibitors and arrived in 55 gallon drums).

Anyone who has used glycol solution knows that it has amazing 'creep' qualities, and will search out the minutest leaks with great efficiency. The klystron cooling was connected up with 'quick release couplers', a fertile source of such leaks. When the coolant was pumped through them at 90psi, the thing became just too easy. The result was that every coupler dripped coolant: leak trays made from old cut-up oil cans swiftly appeared. These had to be emptied on a fairly regular basis. If the liquid dropped on the vinyl-tile covered floor, anybody who stepped on this liquid in rubber-soled shoes would swiftly be on his back, staring at the ceiling through a haze of stars. It was astonishingly slippery stuff.

The Black Heart of the Exciter

This early drive used an unfamiliar form of modulator - the 'absorption modulator'. The basis of this was that if a valve diode were enclosed in a suitable tuned cavity, then by varying the DC bias on the diode, the effective RF resistance of the cavity could be changed. If RF is thrown at such a diode, a proportion of the energy will be reflected, since for most values of bias the diode will not look like its nominal 50ohm design value. If a pair of such diode cavities are connected to two ports of a 3dB directional coupler, and RF is fed into the coupler, then the amount of RF coming out of the coupler will depend on the DC bias to the diodes. If to this DC bias you add an AC video signal, the result is that the RF output is modulated with the vision signal - just what we need.

In practice it was never as simple as this - there were too many variables present. The tune-point of each of the cavities had to be correct and matched to one another. The diodes themselves had to have identical characteristics, and keep them so as they aged. In practice, thermal variations in the drives, and normal aging phenomena (these diodes did not have very long lives, in practice), meant that both daily and long-term variations were inevitable.

The matching of the diodes was a constant pain. The closeness of matching needed was not appreciated in the earliest days, and was one of the factors that led to the 'closed-doors' fiasco. The diodes arrived from the manufacturers as unmatched items, and much time was spent using a testing-jig to find pairs of valves whose emissions matched closely enough to be usable - these pairs would then be taped together in their boxes, and if a replacement was needed, a new matched pair would be fitted. The valves themselves were made by GEC-Osram (a special item, coded A3012 I think) and were basically a DET24 disc-seal triode with the control grid left out. Their characteristics could vary alarmingly from one specimen to another, and there was not even a guarantee that two matched items would age at the same rate.