Before I modified this one, I made some very surprising measurements on the stock stabilizing method. The manufacturer uses a PTC thermistor, standard practice on most of their LO's. I had a spare one from one of their kits, and I was curious, so I stuck a sensor on it and discovered it settled out at 50C with a 9v supply.
Next, I dropped a sensor through the adjustment hole in the LO can, and measured the environment in there; it was 55C.
What this told me was that the PTC thermistor, which is supposed to stabilize the crystal at 50C, was unable to do that...it was already hotter than that in the LO compartment.
Next, I disconnected the PTC thermistor, and measured the environment again...no change, still about 55C.
Then I relocated the 9v regulator to the chassis, removing it from the LO compartment completely. The environment in there dropped to 47C; I didn't think this was enough of a drop to warrant moving the regulator, so I put it back.
At that point, I knew that the controller retrofit had to get the LO environment pretty hot, probably at least 65C if any decent control was to be maintained.
Actually, it has to be around 70C. After many measurements on the response of the crystal, I discovered that it had two turn points on it's temperature curve; one was around 49C, where it discontinued moving down (as it got hotter), rapidly reversed direction and moved up, and then rapidly reversed direction and moved down again. In fact, the reversals were more like jumps (I noticed this same behavior in my 222 and 902 transverters, both of which are Down East units using JAN crystals).
The second turn point was found at about 68C, where the crystal made a smooth slow-down in it's downward direction, stopped, and began to move up even more slowly than before. This transition was slow and smooth, unlike the rapid gyration it made around 49C.
Thus the decision to set the sensor up close to that turn point.