Although not as well known today as the Williamson, and the many Leak, Quad and other hi-fi push-pull valve amplifiers that followed in the 1940s and 1950s, this amplifier is the grandfather of British high fidelity amplifiers. First published in 1934 the designs were revisited many times and until January 1946. See that final design here - Wireless World Quality Amplifier.
See also Rethinking the Quality Amplifier.
See also my notes of designing and building a pre-amp for the Wireless World Quality Amplifier.
Output valves. Directly heated British triodes such as the PX4 and PX25 used in the Wireless World Quality Amplifiers are rare and expensive. American alternatives, with 2.5V filaments and different bases, are cheaper, especially the modern reproductions.
The economy "Wartime" version of the amplifier - schematic above - used American 6V6G tetrodes rather than the then unobtainable PX4 or PX25. For my 2008 interpretation of the amplifier I chose the Mazda PEN45 beam tetrode. This valve was developed in the 1930s and has a 4V heater and Mazda Octal (a.k.a. British Octal) base Marconi had a similar valve - KT41 with the old B7 base. After WWII 6.3V heaters became standard everywhere so this valve was changed to a 6.3V heater and International Octal base and then known as 6P25 - equivalent to the EL33 pentode and KT61 tetrode.
Another other option is to use beam tetrodes (or pentodes) strapped as triodes, Williamson did this with KT66 for the famous 1947 Williamson amplifier. Williamson notes that these provide equivalent characteristics to PX25 triodes. For my 2010 version I chose KT44 beam tetrodes strapped as triodes.
Driver valves. The original design called for MHL4 triodes. Again these are scarce and expensive today. Fortunately ML4 are more easily found and are perfectly usable with a different cathode resistor. In the last published design for the Wireless World Quality Amplifier, alternative American octal valves with 6.3V heaters were also suggested for the driver triodes. The options given were of 6J7 (or 6SJ7) pentodes strapped as triodes or 6C5 triodes. Later valves with very similar characteristics are readily available today, e.g. EF37A pentodes or 6J5/L63 triodes. An interesting option might be to use ML6, the 6.3V version of the ML4, for its classic look with British base and quite likely as good, if not better, performance.
Power supply. For the 2008 amplifiers the mains transformers are from Marconi RF signal generators. Secondaries are 4V, 4V, and centre tapped HT, no load r.m.s. volts are 297V and 233V. The signal generator only had 3 triodes (all ML4) and one rectifier (MU14) but was seriously over engineered - the manual gives a 40W figure for power consumption (the ML4 has a maximum anode dissipation of 5W and a 4W heater, the MU14 has a 10W heater, and dropping 100V at 50ma is another 5W, so 40W seems about right if the valves are run hard).
For the 2010 amplifier I built a power supply capable of powering two quality amplifiers with triode strapped KT44 output. This was possible as I acquired a vintage Elstone mains transformer. The chassis, chokes and paper in oil capacitors came from a vintage power supply - probably from a cinema amplifier.
Output transformers. The original 1934 Wireless World design had the output transformer off the chassis. This was common at the time, with output transformers often mounted on the loud speaker. I've tried a few different output transformers, at present I'm quite happy with a pair taken from an Armstrong 222 amplifier.
Chassis. I've used aluminium chassis. In the late 1930s this was usual for home built equipment, and even today with cheap power tools it's still much easier to work with than steel.
It's worth noting that British B5 (or B4) valve sockets fit the same hole and mounting bolts as IO sockets. There are no doubt exceptions but this is true for typical Celestion bakelite chassis mount types.
Anyone else considering building this or a similar valve amplifier might like to consider the following - punching holes close to the edge of a chassis is harder work than in the centre. Next time I'm tempted to try a configuration more like the Jefferson chassis shown here http://www.roger-russell.com/jeffmcpg.htm. Though I'd probably put the valves slightly off centre to give more space for transformers.