Edit: Added practical observations at bottom.
I need to provide power to an ESP8266 12E NodeMCU board, and my "raw" power is a 24 volt "wall-wart" PSU. My first thought was just to use a 7805 with a heatsink, but I have doubts about the suitability of this.
The current drain for the ESP device has been measured as about 80 mA under "normal" conditions, but with a peak during startup of 350 mA. It's not clear to me how long this startup current persists, but the "boot time" of a typical program from power up is less than a second.
So, the peak power dissipation seems to be (24-5)*0.35 or close to 6.5 Watts. If I read the data sheet correctly a 10C/W heatsink will support this (continuously) up to an ambient closing in on 50C (I live in Colorado, so in the Summer that's "good enough" but not plenty). I have, on hand, some no-brand extruded aluminum heatsinks that are just a little wider and taller than the TO-220 package, and have four fins on them that are perhaps 1/4" deep. They look similar to things that actually do claim to be 10C/W.
As a side note, I need the 24 volts for other things in this particular project, so getting a lower input voltage isn't an option in this case. ed So I have three questions:
Am I missing something obvious that might be a better way of doing this?
Is my interpretation of the data sheet close enough to correct?
Does my heatsink seem adequate?
Well, given that I had the parts to proceed with a 7805 on hand, and the buck-converter (I admit I'd never heard that term before, but it seems to be what I have called a "switch-mode", yes?) won't show up for a day or two, plus the observation that sometimes these switch-mode devices might not always be totally happy/stable at very low output current, I went ahead and constructed the proposed configuration.
With the heatsink in place, with no insulating layer, but also with no thermal grease, and the CPU doing some very minor ticking over, the heatsink warmed up fairly substantially, in a few minutes, it was running around 110~120F based on an infrared thermometer. Of course, that's the "outside" temperature, so I have to assume the silicon inside is running quite a bit hotter.
From this, I conclude a few things, first that the heatsink I have probably isn't as good as I'd hoped (though it's also not nothing). Second, that for this much drop, the proposal to use one of these buck converters is clearly far more satisfactory. I built my (strip/vero) board layout so that I can jumper the 24v supply off the 7805, and instead put the new converter between the two. That way, I'll still have the low-current-stable linear regulator, but be feeding it with perhaps 7.5v instead of 24.
Bottom line, not such a good plan to try to drop this much voltage, even at very low current over a 7805 (nor, presumably, it's siblings).
Appreciate all the input.