Is there a shell command to output the internal voltage of the Pi that is used for the undervoltage warning? I believe this is triggered about 4.63V and there are a few Plugins out there that will display a graph of an input. It would be handy to combine these 2 facilities and so attempt to fault check an occasional historic undervoltage warning.
The closest I have found from the web is for the 4 internal core voltages but these don't really help.
This question seems to have been asked a few times on various RPI forums but there isn't an answer.
anyone managed it?
Last I checked those core voltages were the only thing reported, and that it was a physical circuit doing the measuring so the Pi does not know what the voltage is but only that it is under 4.63v. Not entirely sure but that's the best I've heard.
To measure the voltage you can put a meter across the 5V and GND GPIO pins.
Yeah that's what seems to pop up all over the net. It seems strange that it can't tell what the voltage is if it's able to make the decision that it's under 4.63.
Oh well, thanks.
Given the set-up of the Pi, it's tremendously unlikely that my problem is with the power supply, cable or connectors. I currently have a voltmeter hooked up like you suggest and sitting on the corner of my desk but unfortunately it only gives a real-time voltage. For the past few days it has just sat there on 5.18V but I have twice had the historic warning (solid undervltage symbol) so at some point it's gone under 4.63V. One of these times was when no print was active and the Pi just at idle.
A graph would therefore be really handy. I might have to buy a new multimeter just to help solve this!
You should be able to see from dmesg exactly when the voltage went below if that helps.
Ah good point. So this will be in the syslog rather than in either serial or octoprint .log?
A time stamp might help to figure out what it's actually doing at the time
OctoPrint polls every 5 mins when everything is OK, and every 30 secs when there is undervoltage to know when it ends. So it is not accurate, what is logged in the octoprint.log. Syslog/dmesg is accurate as far as I know.
Wild guess, but perhaps voltage is being measured on the 3.3V side of the regulator and the >4.63V is inferred.