See also my notes of designing and building a pre-amp for the Wireless World Quality Amplifier.  

Function	Type	Alternatives	Notes
Rectifier	U14	NU3,MU14,UU4	NU3 is an equivalent (directly heated perhaps).   The others should be OK.  Have examples of all of these.  N.B. All 4V indirectly heated but cathode connected to heater.  Thus suitable for use with indirectly heated output valves.
Output	PX4	( how about 2A3?)	See below.
Driver	MHL4	ML4, 41MTL	Not a direct equivalent; but I've already got a few. See here for how to calculate component changes.
Phase-splitter	MH4	41MTL, 354V and others SP41 strapped as triode. Or none - as in the 1934 version.	41MTL doesn't seem common, but I have them.  Seems to be a direct equivalent. It seems SP41 was a popular professional audio valve, much as the EF36 would be a few years later - both saw use as pentode or triode.   In the end I went with none!  I'll feed the amp with a balanced signal.

    

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.  Another other option is to use period  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. 

The economy "Wartime" version of the amplifier - schematic above - used American 6V6G tetrodes rather than the then unobtainable PX4 or PX25.  For my 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. 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.

Using tetrodes/pentodes in a high quality amplifier isn't such a bad thing to do, indeed this is just what manufacturers were doing at this time. For example some high quality RDG wireless sets moved from push-pull PX4 output to PEN45 in the 1940s - see http://www.rgd.org.uk/rgd_exh_pg_3611.htm. Of course ordinary wireless sets had been using single ended pentode output since the mid 1930s, generally of very modest output - 2 or 3 watts. Really expensive radiograms continued to use push-pull triode output throughout the 1940s. 

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. 

Iron. 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.  I'll aim to run two ML4s at 2W each) . 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.

Chassis. I've used an 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.

Layout options.  Here are pictures of some variants of the amplifier from 1934 onwards -

1934
1936
2008

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.