I’ll be keeping using dayenergy with the dayenergy=1 flag then - don’t know why others are having problems, but mine has been flawless using DAY_ENERGY since the system was turned on a month or two back.
I struggle to understand how using PAC can be more accurate when accumulating the samples into an energy figure for the day, when PAC is just a moment-in-time measure. If the 5 minute sample is taken right after I switch on the kettle for 30 seconds, that level will be taken as the energy for the whole 5 minute block, and PVO will have no concept of how much the power use has flickered up and down in the time until the next sample. Using PAC, the accuracy of the accumulated energy for the day will be greatly dependent on the sampling frequency, whereas using the DAY_ENERGY figure directly from the device I know the accumulated energy will be correct even if I only sample hourly rather than every 5 minutes, and even if a few PAC samples are missed due to a Internet outage.
(I’d really really like to be using the Tesla PW2 ‘energy_imported’ and ‘energy_exported’ counters rather than the ‘instant_power’ counters for the same reason. Much much more accurate to have an accumulated energy counter as the primary data source, and then calculate power for the time interval from that, than going the other way calculating energy accumulation from power samples - for those devices that keep accumulated energy counters internally).
Thanks for head-up.