PhilD13 wrote:Since the current from the adjustment terminal (IAdj) represents an error term in the equation, the LM317 is designed to control IAdj to less than 100 uA and keep it constant. To do this, all quiescent operating current is returned to the output terminal. This imposes the requirement for a minimum load current. If the load current is less than the minimum, the output voltage will rise.
That totally makes sense, and thank you for the explanation.
PhilD13 wrote:You can get LM317 in 500mA in a TO220 package which will probably fit the space and provide a reasonable margin if your concerned about the warmth of the T0-92 package.
The TO-220 package will fit, although it's a bit of a squeeze. I was actually planning to mention that the TO-220 package should work in the xmas-icetube Rev. D board (even though the silkscreen outlines a TO-92), but I was waiting until after I tested Rev. D.... so you must be something of a mind reader!
EDIT: I prefer a TO-92 package for aesthetic reasons, but using a TO-220 package in the upright position on the xmas Rev D board doesn't look bad. So if AC-to-DC power supply robustness is your priority, the TO-220 package is the superior choice. The
only reason I suggest a TO-92 package on the xmas-icetube board is for aesthetic reasons. For hackers who would rather be extra safe, the TO-220 package is fine!
Assuming the Rev. D is my final revision for a while, my (as yet unwritten) documentation will recommend a TO-220 package for use with an unregulated wall-wort or any other AC adapter that could provide significantly more than 9 volts DC.
PhilD13 wrote:[T]he ON Semiconductor has a better datasheet to use for reference.
Thanks for the ref! Although the datasheet you linked is for the TO-220 package, the ON Semiconductor datasheet for the
TO-92 package is indeed much better and makes me more comfortable with the decision to go with a TO-92 package:
The IV-18 datasheet specifies a filament current of 75-95 mA at 5v, which is consistent with the tubes I've played with. (If I had never seen the IV-18 datasheet, I would have naively guessed a narrower range centered at ~85 mA.) With a regulated 9v switching power supply (like the one that comes with the Ice Tube Clock kit), I would guess the worst-case voltage drop across the LM317 would be 3 volts, and the worst case current would be 100 mA, as the resistor divider only consumes ~1.3 mA. The ON Semiconductor datasheet gives a thermal resistance of 160 degC/W. In the absolute worst case, I can imagine the internal temperature of the clock reaching 50 degC, which would place the regulator temperature at 50degC + 3v * 0.100A * 160degC/W = 98degC. According to the ON Semiconductor datasheet, the maximum operating temperature of the regulator is 150 degC. I am comfortable with that margin of error.
EDIT: The calculations above were for the Adafruit kit. My xmas-icetube board schematic specifies D1 as a 1N4001--a cheaper part which will lower the voltage drop across the LM317 by ~0.6v and further increase the margin of error.