On a Duemilanove, Uno, and Mega2560, I recently discovered (?) that if you provide power via USB, experimentally there seems to exist a load-bearing (i.e. 10k to GND holds up) 5V at the V_IN shield header pin, even without any power supplied via the barrel jack. Technically I think it was more like 4.2V or so, but I'd have to verify. Reviewing the schematics and Eagle files for both boards, I couldn't figure out how that could be the case electrically. Is this behavior by design and if so can anyone explain to me the mechanics of how 5V from USB gets to V_IN? My impression was that V_IN was merely designed as a way to pass the auxiliary power from the barrel jack up to shields for consumption, or for shields to supply auxiliary power into the Arduino in lieu of its barrel jack.
I was surprised to find a boost supply I setup to use V_IN (about 300mA at 4.3V) got power when I connected the USB, I had been using the barrel jack, and did not expect it to power up, but it did. The "mechanics" are that the linear regulator has a reverse diode that allows USB power into V_IN.
ron_sutherland:
The "mechanics" are that the linear regulator has a reverse diode that allows USB power into V_IN.
Ron, that's the conclusion I came to internally as well - it must be a "feature" of the particular linear regulator part used by Arduino (MC33269D)? In my own designs I've used LD1117D and USB power definitely does not seem to feed back through it from USB. I'm not sure what to make of that... I wonder if people expect the Arduino hardware behavior (I, like you, obviously did not), or is it more likely something that most people don't ever encounter / need to worry about? Perhaps more importantly to this thread, I'm wondering whether how widespread reliance or accounting for this behavior is in popular shield designs?
I've been using NCP1117 and on its datasheet under heading for "Protection Diodes" it is saying that the device has a diode between Vout and Vin that can withstand a peak surge current of 15A, and "Normal cycling of Vin cannot generate a current surge of this magnitude. Only when Vin is shorted or crowbarred to ground and Cout is greater than 50uF, is it possible for device damage to occur." So if the LD1117 does not have a reverse diode then these things are all over the road, which I guess sounds about right.