...
I don't see how an external resistor would help, it would get hot inside the caboose as well. I prefer the diode for this task.
Well, just the addition of an external resistor of ANY value lowers the smoke for a given track voltage. So for a given track voltage, the caboose will get less hot AND there will be less smoke. But smoke is still proportional to track voltage. And if 8.5V to the smoke unit is "ideal", you still have the issue that a higher track voltages you may overheat the smoke unit.
Agreed, the 3A diode method cuts smoke power about in half. Skipping the nerdy math, cutting power in half means reducing the voltage by about 30%. So with the diode, if you want the smoke unit running at 8.5V, this corresponds to a track voltage of about 12V. Well, 12V can move an engine-alone just fine but couple some cars and a caboose and I suspect you might need closer to, say, 15V track voltage to keep the freight moving. 15V on the track with the diode presents 10.5V to the smoke unit which apparently causes melting. And, like the resistor method, smoke is proportional to track voltage.
Hence, that's why I suggest a voltage regulator.
I just quickly found the two examples above; I realize you only need one of each depending on which method you go with but gives an idea of what they at least look like!
The regulator method presents 8.5V to the smoke unit irrespective of track voltage. I'd think this is what you want FOR A CABOOSE. That is, for an engine, I can see how you might want smoke to increase as you increase track voltage since the engine is presumably going faster, working harder, etc.