Hi all,
to give a slightly different twist to this story:
To start the EB91 was introduced within Philips/Mullard in 1946. The two TV sets designed that year (the 363A and 563A, where the middle 6 indicated the design year 1946) both used the EB91. See the full story
here.
As already suggested and discussed, the 90-series seems to have originated from the UK Mullard valve design group, first order for local UK marketing only.
As to the valve naming, and knowing the Philips way of thinking from the inside, I'm pretty sure there was a kind of rule "keep the valve name as short as possible given the available coding scheme". So a double diode was B, with the EB91 succeeding the EB4. Because there were no codes for triple diodes the triple diode-triode had to be the EABC80.
But then all of sudden around 1950 other companies (Telefunken?) introduce a double diode as EAA91, although the specs were roughly identical. Strictly spoken not a violation of the coding system, but clearly not in line with the Philips coding methodology. Because the Philips Radio Valves division had the policy to offer all main valves sold in their markets, they required an answer to this EAA91. One option was to start a campaign explaining that the EAA91 and EB91 were identical, but the EB91 the best of the two. A pragmatic alternative was to relabel the EB91 also to EAA91, and offer both in parallel. Although there was also the corporate rule not to issue unnecessary product codes (every code cost money for administration), in this case having both codes was probably deemed the most cost effective one.
With the result that both the EAA91 and EB91 co-existed, although as far as I can see no EAA91s were used in Philips sets, only the EB91.
Cheers, Pieter