Has anyone observed a higher power consumption after upgrading to 23.05.2?
With a WAX202, I noticed that a solar/battery system ran out of power during the 2nd night after upgrade. Unfortunately for the timing of this issue, it's also winter in the northern hemisphere, and the family won't like me taking the router offline - given it's designed to have indefinite power ...so it's currently on a mains charger every other day until the passing of the solstice, or I discover the issue.
It also appears there's increased CPU usage, but I cannot identify a single culprit (although, I notice more kworker and NAPI entries in top than I previously recall). I also notice the load is in the 30's instead of the usual 10's and 20's on the LuCI overview page at login. From the cheesy charge controller, it appears I'm consuming .3 instead of the average .2 amps. I will attempt to connect a better ammeter if/when I have an opportunity and report.
I don't have comparison figures with previous versions, but these are my power consumption stats for an idle EAP615 accesspoint with 23.05.2 according the PoE stats from Netgear GS308EP
This should be pretty easy thing to scientifically proof.
With a simple external power meter or some way to externally measure the power. And measure the power with the same configs and hardware setup with the last firmware installed and with 23.05.2 installed.
Not really, as this battery is a moving target - with variable charging input, variable battery usage and even variable battery health (temperature, discharging due to partially obstructed solar panels, cell degradation). There's no way around actually measuring the router itself with a good power meter (be it an ampere meter sliced into the 12V line or a slightly above average quality -dedicated- power meter on the mains voltage side; a smartplug probably isn't good enough to measure switching power supplies and very low power draw correctly).
EDIT: and even usage isn't really constant either.
As a mips SOC (mt7621), the wax202 does not really have advanced power features, in other words the power draw is rather constant between idle and full load, what will still make a difference is the number of connected ethernet ports (and their throughput negotiations), as well as the activity of the wireless radios.
I was gonna add on an edit - I have a laboratory supply, so if necessary, I can set a steady input, charge, etc. I was also gonna mention weather differences.
Yea, I have a few, but those won't work in this troubleshooting. This is a DC system. Only the laboratory supply used as backup is connected to the mains. That output is known - in DC (you dial it in on the screen).
Maybe I can switch the router back to AC and measure. But that removes the battery from the scenario. Also it adds the unknown of the inefficiency of the transformer, etc. I need to wire the meter into the direct line to the router for best accuracy (or invest in that clamp-on meter).
Batteries work also. But the thing is that whatever energy supply in use the problem is to know exactly how much energy that goes in to the actual device under test.
Lead batteries have a discharge voltage profile that looks like a forward leaning S and how fast it drops in the end is this days capacity. But the current is also changing all the time with the voltage. So if batteries is the source you need to measure and log with 1s timestamps both voltage and current and then post process the data to calculate the power consumption every second.
So the measurement gets a lot easier if using a lab supply that takes the voltage variable out of the calculation. Then we only need to focus on the current. But this only work if it runs on a external PSU so it can be directly powered by DC.
...so it's currently on a mains charger every other day until the passing of the solstice, or I discover the issue.
But that removes the battery from the scenario. Also it adds the unknown of the inefficiency of the transformer, etc. I need to wire the meter into the direct line to the router for best accuracy (or invest in that clamp-on meter).
I may have to hurry before the family sees (lol)!