Tiny ovens

[Steve Jones]

I have seen this debated/discussed here and in the
archives, but I'm not sure I've ever been able to pull
out any consensus (what else is new 8^).

The old problem: oscillator drift due to temperature
change.

An assumption (however reasonable): no 726's.

An oft proposed solution: little transistor ovens.

Is the following feasible/reasonable?

I'm tentatively planning my VCO.  I'm thinking of using
a transistor array; one of those including a diff pair.
I'm considering epoxying an inverted (physically) power
transistor and a thermistor to it's top.  The thermistor
would be part of a voltage divider feeding a comparator
(w/ hysteresis).  This comparator would then be used to
drive the heater-transistor.  The threshold would be set
to a level just hotter than the hottest normal running
temperature of the array.

Is the epoxy heat conductive enough?  Am I oversimplifying
this or should this work?

steve