In article , Peter John Lawton
writes:

A slight advantage of vacuum tubes in capacitor-input rectifier
circuits was that the very high initial turn-on surge isn't there;
a vacuum tube diode literally turns on slowly as the filament
warms up. In cheaper tube designs that was offset by the
higher heat dissipation of tube rectifiers creating a local hot
spot much higher than with semiconductor rectifier diodes.
[typical heat dissipation of a 5Y3 common dual-diode in 100
Watt units was 15 to 20 Watts all by itself]

I seem to remember the 5Y3 as directly heated which means it warms up
before the rest of the valves in the set so producing high HT value
intitially. Need to take this into account with electrolytics.

By comparison to semiconductor diode rectifiers, any tube/valve
diode type with hard vacuum is sloooooow to turn on. A solid-
state diode starts conducting as soon as forward conduction
voltage is achieved. That causes an enormous (relative) surge in
charging current for the input filter capacitor which, before turn-on,
had no charge at all. Given no load resistance at this point in
time, all of that charging current is flowing in that input filter
capacitor. If the primary power is applied at or near the peak of
AC voltage, the only thing limiting the current surge value is the
transformer total inductance and winding resistance, diode
internal resistance, equivalent series resistance (ESR) of the
input filter capacitor, and wiring resistance (negligible by
comparison). That can be one #$%^&!!! large current pulse.
For any sort of longevity, the solid-state rectifier diodes need to
be selected for their "safe" forward current surge ratings.

For a hard vacuum (as opposed to gas, like mercury-vapor types)
diode, even directly-heated, the initial diode "resistance" goes from
near infinity to near full conduction conditions in many cycles of
the AC input voltage. There is nowhere near the slug of current
at initial turn-on as with solid-state diodes. Tube diode rectifiers
can be classified as "gentle" towards input filter capacitors, so
much so that this surge current was seldom mentioned in older
texts. The conditions you highlight are a slightly different one...

In a tube rectifier, yes, a directly-heated cathode will apply the
full peak voltage quicker than the rest of a radio using indirectly-
heated cathode tubes. The initial load resistance is very high
and the input filter capacitor will thus reach the equal of the peak
AC input voltage. That's the major reason of having a Working
Voltage rating of minimum 1.5 times the AC RMS input voltage
(allowance for 10% over-nominal line voltage). Once the rest of
the tubes warmed up, the load resistance dropped, and now the
steady-state condition of the entire supply circuit could be
assumed...and the DC output voltage dropped to design levels.

The major cause of failure in millions of 5-tube BC receivers
other than tube filament burn-out, was the cheapness of the
input filter capacitor. To improve profits of a very economical-
to-buy unit, designers brought the working voltages of the
(generally) cardboard tube cased electrolytic filter capacitors
down to the minimum value. Few of those millions of BC
receivers were designed for good heat dissipation...the total
filament string dissipation was 17 1/4 Watts plus another 5 W
loss in the half-wave diode tube rectifier voltage drop. The actual
useable power (the "B+") was 15 W. That would be at least
38 Watts of heat dissipated in a small plastic enclosure. Those
waxed cardboard tube cased electrolytics could only dissipate
any internal heating (from ESR) through their wire leads.

More robust tube equipment used aluminum-can-cased
electrolytics and large areas of aluminum chassis metal to help
distribute the internal heating and dissipate it better. Designers
used greater safety margins and increased the working voltage
ratings of electrolytic and some other capacitors. With higher
sales values of such specialized products, they didn't have to
skimp on the parts in order to stay market competitive.

In more recent simple power supplies, you will still find some
cases of input filter capacitor failures but the incidence of diode
rectifier failures has also risen due to designers selecting diodes
with lower surge current ratings (not necessarily due to economic
reasons...more a case of not paying attention to it). The focus of
failure reasons has changed over the years but it is still a rather
more complex set of causes involved and not just in circuit
design.

One thing I've observed over quite a while of pushing electrons
around is that "old" tube-based equipment design talked overly
much about voltage levels and neglected current flow things.
[I used to be guilty of that but grew out of it due to smoking...of
the equipment, not me...:-) ] In more "modern" electronics in
general, BOTH voltage and current have to be considered equally
and basic circuit considerations can't get by with simple rules of
thumb for tube circuits.

Len Anderson
retired (from regular hours) electronic engineer person