Powering C4 via header pins

[campbell]

Is it possible to power the C4 at 5V via pins 2 and 4 on the J2 40-pin header, the way the C2 and other Pi form factor boards can be powered? Are there any board peripherals which would be disabled or unreliable when doing so?

[joerg]

It is not possible. The 5V rail is supplied by an step down converter. You cannot feed power back over it to supply the other step down converters for 3.3 and 1.8. I checked this already when I designed my interface board for in car use.

[tobetter]

An alternative way to supply power is:
https://wiki.odroid.com/odroid-c4/appli ... ower_input

[campbell]

It is not possible. The 5V rail is supplied by an step down converter. You cannot feed power back over it to supply the other step down converters for 3.3 and 1.8. I checked this already when I designed my interface board for in car use.

Is it safe to power the board via its 12V barrel jack while it is connected to something else that provides power to the 5V pins?

[mctom]

It is not possible. The 5V rail is supplied by an step down converter. You cannot feed power back over it to supply the other step down converters for 3.3 and 1.8. I checked this already when I designed my interface board for in car use.

Is it safe to power the board via its 12V barrel jack while it is connected to something else that provides power to the 5V pins?

Directly connecting another power source to the 5V pins seems to be quite dangerous.
To make sure, show us your power connection diagram.

[campbell]

Is it safe to power the board via its 12V barrel jack while it is connected to something else that provides power to the 5V pins?

Directly connecting another power source to the 5V pins seems to be quite dangerous.
To make sure, show us your power connection diagram.

Pins 2 and 4 are connected to the output of an AOZ1353DI-01 power switch, whose input comes from a MAX20077ATCB switching power supply. The board was designed to hold and power a Raspberry Pi. We are looking for ways to disable the AOZ1353DI-01 when an Odroid C4 is connected and separately powered (or assure ourselves that we don't need to worry about it), while still allowing the board to be used when a Pi is connected instead.

[mctom]

Attaching an extract from C4 schematics, a converter from DC input to 5V. This 5V is exposed on GPIO pins.

Of course, no 5V regulators output exactly 5V, but a value really close to that.

If your external 5V source will happen to have voltage slightly higher than onboard 5V, this Buck controller chip may try to lower it by force, using its internal low side MOSFET.
If it tries too hard, it will burn.

This is the most common failure mode in RasPis from my experience - by shorting pin 1 and 2, you're shorting 3.3V and 5V. As a result, 3.3V converter gets 5V on its output and commits suicide trying to lower it back to 3.3V.

The same may happen in your setup, if you supply 5V that is "higher" than onboard 5V.
How much "higher" is bad? I honestly don't know.
But to make sure no such thing would ever happen, one could for example lower input voltage to 4.9V. Raspberry Pi will still work fine, because it needs 4.75V minimum, and only if you plan to use USB. I ran a stack of Pis at 4V and lower with no problems. The CPU runs at 3.3V anyway.

Or just cut 5V pins on C4, lol.
Or modify R103/R104 divider, so C4 runs at some 5.2V instead. Shouldn't cause any trouble, USB has 5% voltage tolerance.

[mctom]

Tonight I slept a bit better than last night so I got some new insight.

True, ensuring that C4 works at higher voltage than your mysterious module will protect C4's internals... but it may backfire, quite literally, and kill your MAX20077ATCB converter in exactly the same way.

What could be done about that? A diode in series with inductor in your design. This way, the converter won't be able to absorb current coming from the outside.
However, this may impact the Buck converter stability because diodes are nonlinear, so such solution ought to be tested. And that may be a significant power loss compared to the whole Buck converter on its own, with all output current flowing through a diode.

Another possibility is MOSFET-based ideal diode between your buck and switch, or even behind the switch. With such low losses I wouldn't worry about voltage drops, and thus wouldn't bother adding that to the regulation loop.

One more thing, removing pins from C4 isn't that daft idea after all.. It may be possible to remove them with a soldering iron hot enough so it would melt both tin and header plastic. I've never done that on purpose, but have certainly proven it could be done unintentionally