Thought I saw you mention and I tried this a couple days ago with a good result. Will revisit with a clamp & pass through next round.
Cool, it wasn't until I twigged the pid display works differently, and there actually was some considerable heat oscillation from any change in conditions (e.g. using it!) that doesn't display on the pid. All I can really say is for me it seemed to produce more repeatable results, and whatever the real physics of the thing, in the end that was what I wanted (as well as a satisfactory hit!
).
And I'll bet (even real money!) that it also depends on how each person uses it - draw technique, temp setting, bowl packing, blah blah. So all advice is good, but needs to be taken as only one persons experience - not right or wrong, just one way of doing of many.
The clamp thing I also picked up on, the slight mismatch in the connection when the weight of the handle pulls down. I also thought about a catch/hook opposite to the handle, built in to stop it, but then removal after the hit could be a problem if it stuck or caught on removal (accidents pulling over a water filled bong and dropping a red-hot heater in your lap all at once!
). If the mating is so close it doesn't get pulled, then expansion's likely to make it stick (I even think Cal mentioned this some time ago, but I may have unknowingly made that up!
).
My solution, if you can call it that, was to just allow it to sit unassisted for the first warm up as the thin Ti is such a good conductor that over that time period enough heat passes over quickly, and I'm really just taking the cold off, not trying to hit an exact temp, then the second soak I hold the handle so the mating is 'perfect' for the 10/15 seconds (another reason why the pre-soak helps, not coming up from room temp), and then keeping the heater aligned I pull (aim being to get as much in one hit, more important than total amount - hence I found half bowls work better).
But I know I work in a fairly specific mode and not everyone likes to do things in that way, each to our own, be so so boring if everyone agreed on everything! But if that gives a possible framework to try, hope it helps.