70MHz - GS31B (YU1AW)                                                                                        GW4RWR
	Why go to the effort of putting a largeish triode on a band where the power limit is 160W? It's a fair question. Perhaps it's just for my own amusement. Most cars could easily exceed the speed limit. But they're not driven around country at maximum speed. I'm just hoping to see a fair bit of gain, so that I could achieve a clean, legal limit signal.without needing too much drive.  Who knows, perhaps the power limit will be raised in the future. The GS31B is one of the cheap Russian valves sold as 'new old stock'. I think I paid about £35 for this one. Apparently it's electrically very similar to the GS35B, but has a smaller anode cooler, dissipation is quoted as 1kW.  Not many amplifiers are designed for this smaller tube, but it's substantially cheaper, and may have longer life expectancy. So it may offer better value?  The series apparently have perhaps 3dB less gain than Eimac's 3CX series, and run at substantially higher grid currents. The filament runs on a convenient 12V. This cabinet was welded together from four 3mm aluminium blanking panels,with a thiner plate for the floor, welded about 1/3 up from the bottom. Base and top plates have folded in edges to give a tight fit, with self tap screws holding them firmly in place. A diecast box encloses the cathode, and a copper box takes care of the anode side. A directional coupler can be seen in the third photo. This has four coupled outputs, two are used for measuring forward and reflected power, two others protrude through the rear panel and can be used independently for feeding to a spectrum analyser. Results: 50mW drive from a 2N3866 gave 2.5W out, so about 17dB gain at low level. 15-17W from a 2N5519 gave 400W out, with 4kV on the anode.  So roughly 13dB gain. 120mA grid current with cathode bias of 22V. 8W drive gives 180W out, so just right for our legal limit.
The 470pF 20kV blocking capacitor is overkill, but was all I had available. The insulating 'wall' had to be added between valve and copper pipe anode line to support the blocking cap weight. The brass tune capacitor has insufficient spacing and will arc across beyond 400W, but is perfectly adequate at the 200W level that I run this amp at.
	The control board became a bit overgrown! I ran out of space since the input and output relays need at least 27V and 48V respectively. An extra transformer was fitted between the control board and the switches, with a voltage doubler to reach 50V. The cathod bias is unorthodox.The only thing remotely suitable in the junk box were 2N3055. Emitter is connected to cathode, with base and collector tied together. One transistor drops 22V in this mode, and the other drops 13V. Front panel switching between 22V and 35V (both devices in series) is available. You can make out two white 22k 12W resistors in parallel, ensuring that the valve is biased off during receive.
	The input variable capacitors are visible at top left of the diecast box, below. A computer fan blows air out of the side of the diecast box's side holes, powered from the filament transformer.
	The input coil links to the cathode via an aluminium strap. I couldn't cut a thin enough strip to strap around the filament terminal, especially since I wanted to avoid blocking the 'ventilation' holes around its circumference, so one of the filament wires is soldered directly to the valve.
	The recessed meters are 1mA fsd for Grid current and 100uA fsd for power. The existing scaled card was scanned, edited using 'Paint' and then printed on a laserjet. I deeply regret not having allowed sufficient space for a third meter in the column, to display fwd/rev power simultaneously. There may appear to be space above the switches, but in practice they extend too far up towards the lowest meter. A meter at column top would look out of place, being too near top of cabinet. Or?