Kai-
to answer your question regarding the frequency - on my board the lamp excitation oscillator runs at 96.7MHz when the lamp is warm and everything's stationary. Initially it's one or two MHz higher. That's bad luck with your lamp exciter transistor, these beasts are not particularly cheap and quite awkward to replace.
Meanwhile I've been able to remove the SRD network from my sick '101 - it's been possible to get the anode pin of the SRD free using a desoldering station and after that, just careful but ample application of heat at the ground flange of the RF feedthrough permitted to "thread" the whole network through the hole in the back wall of the cavity.
I was amazed to find that the feedthrogh is actually a capacitor of 100pF, so external installation of the SRD would have been completely useless. Moreover, so far no suitable replacement SRD is in sight so I've got a plan that may sound quite mad - but there's nothing to lose anyway: The SRD is incapsulated in a glass envelope, just like the ancient germanium diodes. Internally, the chip seems to be connected via a (molybdenum?) sheet metal spring and some kind of a small "pip" on the die itself.
If I measure the SRD at certain times / moon phases
I can actually get "some" continuity readings in the passing direction but never in the blocking direction. This makes me assume that the crystal itself is still intact.
What I now intend to do is to make a small fixture to place the diode between two opposed metal shafts with tiny holes drilled centally to accomodate the terminal wires of the SRD. These shafts can be rotated by two synchonizing gears so I don't put any torsional stress on the glass envelope. This whole arrangement will be mounted vertically so the weight of one of the shafts lies on the diode body. My approach will now be to heat the glass evenely while rotating the diode, just enough to soften it slightly in order to increase the pressure on the molybdenum spring. I've got other diodes in the same mechanical package so I can figure out the required temperatures / times.
I'm quite curious if this repair approach will work -- and as said before, there's noting to lose, only experience to gain and maybe even win!
Please see three microscopic photos of the SRD network. There are the 100pF feedthrough, a 3 ohms low-inductance resistor and the SRD itself. The closeups of the SRD show its internal construction more or less clearly...
Cheers,
Thomas