Mega 2560 Power from 5V pin

I recently dug out an old robotic platform that I worked on a few years ago, but I can't find my documentation.

In it I have a 12V NiMH battery that connects to a Power board of my own design with regulators on it. I have pins for 6V, 5V and 3.3V.

My memory was that I was able to power the Mega 2560 using the 5V pins on the power board and plugging it into the 5V on the Mega 2560, but the current arduino.cc website makes it sound like the 5V pin on the arduino board is output only, and the Vin needs at least 7V to run. I see passing references on the forum about powering the mega from the 5V pin but no real documentation that deals with situations that would arise if I did power the board from the 5V output.

Does anyone have any pointers of how I was able to do this? Should I just rework my power board with a 8V 1A power regulator to power the mega using the Vin pin?

Powering the Mega through the USB connector is basically connecting 5volt to the 5volt pin.
There is only a polyfuse and a conducting mosfet in between.

I use a hacked USB lead to power a Mega I have.
Leo..

Why mangle a USB cable, rather than just pull the 5V to the 5V pin?

There have been many threads about backfeeding the onboard 5volt regulator.
I suggest you do a search.
Leo…

Wawa:
There have been many threads about backfeeding the onboard 5volt regulator.
I suggest you do a search.
Leo…

I have , hence why I said "I see passing references on the forum about powering the mega from the 5V pin but no real documentation that deals with situations that would arise if I did power the board from the 5V output. "

Most of what you get is people saying the took a USB cable and fed it 5V, but nothing the really gets into why that seems to be the main solution rather than backfeeding the 5V regulator. I was looking for something more than what I found doing a search where one person says they do it just fine, someone else jumps in and says that you risk damaging the board, and a their person jumps in and says that they modified a USB cable and fed it 5V that way.

Looking at the schematic, I don't see how powering it off the USB connector is any different than powering it directly from the 5v pin, from the perspective of the regulator (as long as you don't connect it to any other power sources, ofc)

This topic
http://forum.arduino.cc/index.php/topic,82046.msg758170.html#msg758170
Reply #32.

Apparently the regulator in use is not tolerant of being back driven. Doing so can wreck the regulator, and somehow there is USB chip damage in there too.

I recommend a diode from 5V (anode) to Vin (cathode) to limit how much the regulator is back driven.

See page 10 of the attached for diode discussion and more.

NCP1117-D-81326.pdf (238 KB)

This whole thing doesn't make sense - and that 3 year old post does not seem to clarify anything...

Look at the schematic - If you power it off 5v... the output of the regulator is at 5v and the input is floating.

CrossRoads:
This topic
http://forum.arduino.cc/index.php/topic,82046.msg758170.html#msg758170
Reply #32.

Apparently the regulator in use is not tolerant of being back driven. Doing so can wreck the regulator, and somehow there is USB chip damage in there too.

I recommend a diode from 5V (anode) to Vin (cathode) to limit how much the regulator is back driven.

See page 10 of the attached for diode discussion and more.

The datasheet seems to indicate that there's no need for a diode here, unless one were to short Vin to ground when the output caps were charged (and the output cap is 47uF anyway, so it should be okay). That implies that there's no need for a diode with the NCP1117...

from the datasheet (emphasis added:

Only when Vin
is shorted or crowbarred to ground and Cout is greater than
50 F, it becomes possible for device damage to occur.
Under these conditions, diode D1 is required to protect the
device.

Similar story with the AMS1117 which IIRC is used in some official boards instead (even though the datasheet suggests that a 47uF electrolytic cap is not an appropriate output cap):

Unlike older regulators, the AMS1117 family does not need any
protection diodes between the adjustment pin and the output and
from the output to the input to prevent over-stressing the die.
Internal resistors are limiting the internal current paths on the
AMS1117 adjustment pin, therefore even with capacitors on the
adjustment pin no protection diode is needed to ensure device
safety under short-circuit conditions.
Diodes between the input and output are not usually needed.
Microsecond surge currents of 50A to 100A can be handled by the
internal diode between the input and output pins of the device. In
normal operations it is difficult to get those values of surge
currents even with the use of large output capacitances. If high
value output capacitors are used, such as 1000µF to 5000µF and
the input pin is instantaneously shorted to ground, damage can
occur. A diode from output to input is recommended, when a
crowbar circuit at the input of the AMS1117 is used (Figure 1).

Seems to be the same with other brands as well - they note the potential for damage if the input is shorted to ground while the caps are charged, but not otherwise.

I've had no problems powering the output of regulators without powering the input, with no diode (on 1117-series regulators from several manufacturers)

If you want my theory, a few people accidentally shorted out the power jack (easy to do - they're big exposed pins on the bottom, just takes a foil gum wrapper on your desk to short it out) with power from 5v connected, and damaged their board (probably in part because it bypassed the polyfuse on Vin), and posted about it at around the same time, leading to people believing there was a problem with the board.

Could be. But the MC33269 used on the Duemilanove and Megsa1280 didn't seem to have any bricking issues like the Uno and Mega2560 had, and there is no mention in its datasheet of needing protection diodes like the 1117 types have.

That 3 year old thread seems to deal with running the mega 2560 with 5V on the 5V pin and having the USB plugged in at the same time and killing the USB controller (I can see the USB 5V being +1-2% and the 5V pin being -1-2% and causing s quick spike in current killing the USB controller, haven't spent anytime looking at the USB data side of things to try and explain it.)

Looking at the data sheet for the NCP1117 I don't see any problem in running off a 5V externally regulated, as long as you avoid a situation where you have 2 power supplies (just a good idea in general.)

jodosh:
That 3 year old thread seems to deal with running the mega 2560 with 5V on the 5V pin and having the USB plugged in at the same time

Well that'd do it...