The FDN340P (P channel MOSFET) is installed backwards in the Arduino Uno schematic.
The P channel source, which is tied to the body, should be pulled high. In this case USBVCC.
In the FDN340P's current positioning in the schematic, the body diode is forward biased if USBVCC is alive. This means USBVCC will continue to supply +5v minus the FDN340P's body diode drop, regardless of output of the comparator.
The only thing separating USBVCC and the +5V regulator is a body diode.
In the following image, you can see that despite Vin=12v and the output of the comparator being +5v, which should turn the FDN340P off (I know I'm using a ZVP3306), the USBVCC is still supplying 4.6v.
I believe the purpose of that PMOS is to prevent the 5 volts, which is derived from the power jack, via a regulator, from back feeding USBVCC. This 5 volts, which you have not indicated, appears at the source of the PMOS. Thus the original circuit is correct.
Thanks for the response. I think what you wrote is absolutely correct, however, I feel it "might" be important to show
what you mean.
In the case where a computer USB is connected to the UNO, and not supplying power for some reason, we can
see calamity that would occur if you reversed the P channel. I took the liberty of loading USBVCC with 1K ohms for visualization.
The top schematic shows the P channel as it is drawn in the Arduino UNO schematic. Notice only a small
amount of leakage current is getting past the P channel to USBVCC (~1.25 uA).
The bottom circuit shows what happens if you reverse the the P channel. What you get is "follower"
action from the output of the comparator, via Vgs of the P channel, to USBVCC. This means the 5v regulator
can now supply current to USBVCC based on USBVCC's loading, which could be significant, and this is very bad.
It looks like you have got it. This, that is power handling, is one of the big differences between the Uno and the Nano.The Uno uses this relatively complex circuitry to prevent the 5v from Vin backfeeding the USB. The Nano uses a crude circuit using a schottky diode. This works but introduces a volt drop.
I think the Schottky may be a good option for the Nano. When the Uno is powered by USB +5V, the source-to-drain voltage drop of the Pchan MOSFET (Uno's FDN340P) is around 400mV because of how the Pchan's source is oriented. This is why I initially questioned the positioning/orientation of the Pchan. The MBR0520 Schottky Diode has a forward voltage drop of 375-440mV. Using the MBR0520 seems reasonable compared to the Uno's power circuit, while using less circuitry. I also like the Schottky because it has significantly less junction capacitance, and therefore less susceptible to power spikes feeding through to the USB power. Thank you for pointing out the power handling of the Nano! I'm making/designing an Uno right now, and I think I'll implement the power handling of the Nano on the Uno. Thank you!
