The Uvicord II has a pretty fascinating way of running the UV lamp: The light source is a sealed quartz glass tube with a little zinc or mercury in it (depending on the desired spectrum), but it has no leads! Instead, it is externally heated at one end (see below) and the energy is transferred by high frequency capacitive coupling into the tube. Like when you bring a fluorescent lamp into a microwave oven or similar if you move it close to a charged old fashioned CRT :
Gas at low pressure + alternating high strength electric field -> ionized gas -> light!
Fig. 3 shows the reconstructed schematic of the lamp driver: It generates a high frequency signal which is capacitively coupled into the light source tube. The power is supplied through P1 and the HF current is coupled into the gas tube with metal rings connected to the ends of the coil (depicted as P2), they are located at the reverse side of the PCB (Fig. 2).
The easyeda file: uv-driver_easyeda
The NPN transistor, originally a BSX53 (30V, 100mA, 130V, 300MHz, Ic/Ib=75..750, of which it was quite difficult to find a datasheet, the PCB is from 1987, almost 30 years ago and the board still reads “LKB”) was supposedly and apparently dead (I tried to measure its hfe with my DVM, it turned out to be 0, i.e. no response). So I replaced it with a 2N2219A which has similar electrical characteristics (UCE 50V, 800mA, 0.8W, 300MHz) and the same case (TO-39, to re-use the heat sink).
The gas tube is heated at one end by a BD179 transistor (30W, UCE 80V, IC 3A, IB 1A) mounted to the lamp holder as heat sink.
Notes & Acknowledgements:
The schematic was created with the free SAS online editor at easyeda.com.