In article , "Tim Wescott"
writes:

Your solid state devices will either be really light-weight or will have
significantly higher capacitances than the tubes, so it would be very
difficult to get your oscillator to track.

I've got an RF-1 which I've contemplated upgrading. Before I went to solid
state I'd give the filaments a regulated supply (like budgie's post) _and_
put in a regulated plate supply. A solid-state wideband buffer would
probably be a quite good idea; a 0-50MHz broad-band amp into 50 ohms should
be fairly easy.

Going to all-regulated supplies for a vacuum tube thing turns out to
be rather easy. I've just finished breadboarding that for a 1960s era
HF SWBC receiver to improve its frequency stability. Tube oscillators
need a stable "B+" or plate supply that, in older days, was enabled
by gas shunt regulators (OA2 or OB2, for example). Heater voltage
variation is less of an effect on frequency stability but such does
occur when the AC primary voltage varies to the extremes of its +/-
10% possible range.

On checking the HV supply of this little receiver, the OB2 strike
voltage had increased beyond manufacturer's spec so it did not always
strike and start regulating with power-on. The lone OB2 in spares,
also 40 years old, was just the opposite, striking below spec. That
prompted looking into a series regulator for the whole B+ using an
ST (Fairchild also makes them) TIP47 series pass NPN with a 250
V breakdown and a 2N3440 250 V NPN as the error amplifier. Not a
problem to have it regulate everything at +120 VDC plate supply,
using only half of the original filter capacitor quad.

Regulating the filaments just barely works with a conventional 7812
positive IC regulator. My version is a combo series-parallel string of
6.3 and 12.6 V nominal filaments, the whole taking 0.6 A. With less
than 10% ripple on full-wave rectification and 12.6 VAC RMS nominal,
the 2.0 V minimum of series drop necessary is maintained even at
105 VAC primary transformer input. The filament voltage is maintained
constant at 12.1 VDC within 1 percent at all conditions of AC primary
voltage. To get a regulated minus, a 7912 (if it can be found) will do
with rectifier and capacitor polarity reversed, handy for biasing other
circuits with a negative voltage.

With a 6.3 V filament arrangement, that would be on the hairy edge
of regulating at -10% AC line voltage. Filter capacitor needs to be
high enough for only 5% ripple and rectifier diodes can't be much more
than 0.7 V forward drop to fit the minimum 7806 regulator 2 V head-
room. Note: The original AC filament voltage was made within certain
(seldom specified) limits; if that is on the low side also, the 7806
may poop out during droops in ripple voltage. 7906s were originally
made in the series for negative voltage regulation, also good for
providing a minus bias voltage. Tube filaments for indirectly-heated
cathodes don't care about DC polarity. Slightly-lower than nominal
voltage will improve heater life.

Higher voltage NPNs for series-pass applications are getting scarce.
So are higher voltage zeners. TIP48 has slightly higher breakdown
than TIP47, can work for higher B+ values although there's a need to
check voltages for +10% AC line conditions rather than -10%.
Exceeding breakdown voltage can be catastrophic to the transistor
junction in just a single AC cycle.

Len Anderson
retired (from regular hours) electronic engineer person