The direct heated cathodes turned on fast enough
that
one didn't have a noticeable delay. It wasn't until the
indirect heated cathodes that things changed. That was
because it was too hard to build an AC radio with the
directed heated tubes. There was both a hum problem
and it was more difficult to get each tube biased correctly.
Once they had selenium rectifiers, direct heated became
practical for low power applications ( like the Zenith TO's ).
They still had to do some tricks to get the biasing
working right for each tube. Some clever engineering.
I have a UK battery/mains radio that uses valves throughout (even the PSU
rectifier) and directly heated valves in the signal stages. The filaments
are in series (carefulyl ordered to get suitable cathode bias voltages),
for a total of 7.5V. The batteries were a 7.5V LT and 90V Ht pair. On
mains, there is a mains transformer, the output of which is full-wave
rectified (EZ40 IIRC). That provides the HT (B+) directly, and the
filament supply (A+?) via dropping resistor.
One think to be wary about. There is a capacitor across the filament
string, after the capacitor. If the filament string is open-circuit for
any reason, then that capacitor will charge to about 90V. It normally
withstands that (although it's not rated to do so), but if you then
replair the filament strign, there's enough energy in that capacitor to
burn out one of the filaments. I always make sure that capacitor is
discharged before fiddling with the valves.
-tony