Greg,
Tube cooling is significant on tubes with a glass temperature in excess of 200C. This applies to output tubes, transmitting tubes, and generally anything with more than about a 30 watt filament.
There are some small tubes which run very high temperatures, because their dissipation is very high. Examples are the excellent and almost unobtainium 5687, which has a merry life, but a short one. Other examples are the small, grunty Rusky tubes, like the 6H23pi, which run VERY hot. For these tubes, I recommend tube coolers, but not for those of more moderate ilk, particularly the lower current octals, like the 6SN7, which run cool as cucumbers........
Tube envelope temperature is also dependent on operating point. You can run 'em to the rail, or you can run 'em light. You choose, and this affects life, sound, and temperature. I run my designs around 2/3rds of Continuous Commercial Rating, and this assures them of long life. It is not unusual, for example, to pull the tubes on a sixties Tek CRO (there are lots of 6DJ8s in them) and find that most of them are still OK, with emissions above 90%. Tubes, particularly the smaller ones, can be extremely durable, particularly those made at the end of era when competition with SS was fierce and the principal drawcard was reliability.
The hottest running tube I've seen is the 6C33C, a Rusky tube with 42 watt filament and 65W plate dissipation. I know the CX 250 GE transmitting tubes for radio stations run MUCH hotter again, and are forced air cooled. The 6C33C tube was the series pass element in the MIG Jet fighter, and runs a 260C glass temperature. It will withstand 10G, however, so is very robust. A tube cooler on this tube would be an impressive piece of thermodynamic design.........
Lastly, microphonics are really only an issue on tubes used for voltage amplification purposes, and are largely determined by the fragility of the plate, grid and cathode elements, the mu of the tube, and the mounting configuration (some sockets are rubber mounted). In a cathode follower application like the GK1, microphonics are not a significant problem, because the output at the cathode is tolerant of considerable variation of mu. However, if you use a medium mu tube like a 6DJ8 as a phono amp in cascode configuration, tapping the envelope with your fingernail will almost certainly produce microphonics, even while the phono is operating. But this is a high gain application, highly susceptible to microphonics.
Transistors suffer from microphonics too, but far less than tubes as their active electrodes are very small and tightly attached to each other on a rigid die. Most of the tweaks concerning resonant platforms in electronics are concerned with the microphonics intrinsic in transistors and capacitors, though this is rarely specifically mentioned.
Cheers,
Hugh
