Mullard released the EF91 and the EF42 at about the same time. For example both were mentioned in WW 1947 November.
The EF42 was the high-slope pentode “TV valve” in Mullard’s then-new Rimlock domestic receiving series. As far as I can determine, the EF91 was intended more for industrial applications (which were essentially out-of-scope for the Rimlock series), along with other new B7G issues such as the EC91, ECC91 and EL91. Also, it would appear that the EF91 was more-or-less a clone of the Osram Z77, as were the Mazda 6F12, Cossor 6AM6 (Cossor did the American registration) and Brimar 8D3 (later redesignated as the 6AM6).
The Z77 was Osram’s high-slope pentode “TV valve” in its new miniature B7G “77” series, and the one valve served both domestic and industrial applications. Evidently Cossor and Brimar both adopted the same “one high-slope pentode for both domestic and industrial markets” approach. On the other hand, Mazda followed the Mullard pattern in offering the 6F12 primarily as an industrial valve and its Rimlock 6F13 as a domestic “TV valve”.
Circa 1949 Philips adopted the B9A base for its definitive range of TV valves, and also for new radio receiving valve issues. Thus its enneode FM demodulator, which started life as the EQ40, quickly became the EQ80. The EBF80, I think introduced when it was found that the EAF42 did not work as well as expected, out of necessity used the B9A base. At the same time, Philips also strongly embraced the power-transformerless, live chassis form for TV receivers, which necessitated series-string heaters, with 300 mA being the chosen heater parameter, and which implied a lowish upper limit to the available HT, as mentioned by G8HQP Dave. This outruled the EF42 (or a B9A copy thereof), which had a 330 mA heater and worked best with 250 V HT. Thus was developed the EF80, first mentioned in 1949, in which some slope was given up to allow the use of a 300 mA heater (whilst staying at 6.3 volts) and 170 V HT. The B9A base allowed two cathode pinouts, considered to be beneficial at VHF, whilst retaining separate pinouts for the suppressor and internal screen.
At about this time Mazda released the 6F1 as an improved 6F13, with two cathode pinouts. However, as Mazda had retained the Rimlock base, this required that the suppressor and internal screen shared a pinout.
Osram simply cloned the EF80 as the Z719, in its B9A-based “7n9” series (which had started with the X79).
Through all this the Z77 and its various clones such as the EF91 continued unchanged, and were widely used in industrial equipment – and some domestic equipment – through to the end of the valve era.
Thus I think that one could say:
1. The EF80 was the direct replacement for the EF42, which had been a contemporary of the EF91.
2. The reasons for that change were fourfold: to conform to 300 mA heater current; to work with 170 V HT; to provide improved performance by having two cathode pinouts; to use the B9A base.
3. Like the EF42 before it, the EF80 was intended to run alongside the EF91, and not to replace it. Whilst they overlapped, there were differences, and the EF91 evidently was a good fit for many industrial applications.
Cheers,