I assume you are talking about temperature induced drift, as opposed as
frequency lability due to hand effect, changes in regen setting etc.

I have experienced such stability over time that SSB pitch would stay
dead on for days (run on and off) in the shortwave range with the radio
tuned on a volmet station half a continent away. In this instance, I
was running a 1-tuber with direct heated filament, battery powered, and
cap-throttle regen control, but the lessons apply in general.

The main tricks in my book are quite simple:
- keep oscillator power as close to zero as humanly possible
- keep voltages stable
- select LARGE components and hookup wires
- ensure ventilation!

In a word, make sure the oscillator basically runs at ambient
temperature.


If you were referring to frequency lability caused by other factors, I
have good news in the hand effect department, and mixed news in the
regen control area.


Hand effect can be more or less got rid of by proper shielding or
decoupling. You can find an interesting discussion he
http://tinyurl.com/dr936 .

In a word: keep your body away from ANY capacitance related to the
oscillating tank circuit - and that mainly refers to the ANTENNA, not
the variable capacitor or the coil. If there is a sizeable
antenna-case capacitance, then a change in body-case capacitance
affects antenna and inevitably the tank circuit too.

So the oscillating tank circuit must be properly decoupled (usually by
RF amp/buffer stage ahead of the regen detector)
OR
it must be well enclosed + the antenna coax fed
.... or both.

I would like to experiment with cap-decoupling by putting a
slitted-shield around the tuning coil, and putting a separate antenna
coil next to it, but never got around to doing so. Even theoretically,
the idea of such a trick gives me a headache.


In the regen control department, the mixed news is that capacitive
throttling is considered the "best performing" way to go, but can't
avoid to affect the tuned frequency.

[ In theory you could, of course, what with a small co-shafted VC that
removes capacitance from the main LC as the throttle cap adds it, but
it must be a pretty complicated law, and I don't know of anyone who
tried it ]

You can instead turn to the gain-control way, DC-controlling one of the
following:
- second grid voltage in a pentode
- plate voltage in a triode
- base voltage in a transistor
or some such like.

All of the above minimize frequency change especially in tube designs
(where varicap effects are basically nonexistent), but can suffer from
annoying hysteresis, in my experience more than cap throttle. Not sure
where the "high performance" advantage of cap throttle is though...

Another trick I wish I had tried is changing a *stabilized* voltage, as
opposed to using a simple throttle pot. That could be done with a
variable regulator IC and might remediate the hysteresis.

Another trick is to use a couple of effects working in opposite
directions, such as a resistive throttle that adds gain but shunts more
RF to ground, such as in the Charles Kitchin "Beginner's Radio".

See www2.arrl.org/tis/info/HTML/regen/
with many derivatives linked in this article of mine:
http://www.geocities.com/filippo_cat...sert-ratt.html .

I think the same trick may be in use in the MFJ-8100 regen, but I can't
tell what's trimmer and what's main control pot.
(hand drawn schematic may be he
www.diagram.com.ua/list/mogik.shtml
sorry Martin!).

So, there's hope.