The mythical (?) POE mezzanine for SweetP looks like it must be sending 5V to the board through a couple pins on the 40 pin header. The schematics are less than helpful to me, with cryptic notations about what’s present and what’s omitted in various configurations. So can I get confirmation that this is a safe/approved way to power the board? The USB connector is fine for bringup and testing, but wouldn’t fit into my project.
Also, the USB input doesn’t appear to communicate with a power brick that conforms to the USB power delivery specs. At least I assume that’s why the board resets if I plug an SSD bay into one of the ports. The brick is rated for 3A at 5V, but as per spec it limits the current available if the attached device doesn’t signal it.
After much sifting through other topics, I stumbled across this nugget of information:
Since I have also decoded the schematic enough, I know that backpowering through USB is, aside from a 2A fuse, just the same as powering through the VCC5V pins on the header, so that answers my original question, I guess. But of course that leads to new questions about how it behaves when so powered:
“Unable to reset power via regulator”? What’s that mean in plain words?
I have some inkling about the SoC watchdog, but from what I’ve read it causes other annoyances if enabled, no?
I’m still not sure what happened the first time I tried to powerup through the VCC5V header pin. Optimism plus laziness had that done without the serial adapter connected to the console port. Something was wedged for a while, and all I got out of the console through some fiddling was “GXL:” (and no HDMI output). After some of that, I pulled the power brick (had been using a switch between it and the SweetP), let things “rest” while I double checked all the few connections, Back to basics, but with console monitored, plugged it in and… it just worked.
poweroff seemed to work, and it rebooted after pressing the boot switch, so I’m still in the dark about what… oh, “unable to reset”. Let me try that… nope, systemctl reboot also works just fine. So no clue what “Unable to reset power via regulator” really means, nor what the watchdog might be needed for. Unless the disabling of the watchdog (mentioned in some other(?) topic I read while researching (viz., wandering aimlessly through a twisty maze of topics, all disorganized) was reverted in later images.
I’m not certain whether this was part of the odd and varying results observed previously, but after having gone back to a minimal setup in order to run the SPI flasher, I realized I was probably plugging the external SATA SSD into the OTG port. At least so far it seems to be acting sensibly when in one of the hub ports. Confusion was aided by the fact that which port is the OTG is somewhat obscure unless you trip over a passing reference to it (which I did, but I’d be lying if I could tell you what topic I found that in now).
…makes sense that early boot would handle the OTG port differently, since there’s that emmc flash over USB mode…
Something was wedged for a while, and all I got out of the console through some fiddling was “GXL:” (and no HDMI output). After some of that, I pulled the power brick (had been using a switch between it and the SweetP), let things “rest” while I double checked all the few connections, Back to basics, but with console monitored, plugged it in and… it just worked.