Actually I haven't done much. The main problem on the pulse meter was solved by adding a capacitor on the signal out pin to act as a hw debounce

@monte if you can share, pls do so

at that price it is worth to switch to rfm69

Last time I used the FTDI cheap clones I have it was about 2/3 weeks ago and they were still running. I haven't tried them lately so I can't confirm if any new update broke the drivers

SD cards fail faster if you have programs constantly writing to them, I am using dietpi as it has been thought to rely mainly on ramdisk for log and temp files

If you don't want to rely on the RPI, you can still use node to node communication and have a node sending temperature to the relay and when it gets a temperature lower than a certain limit will trigger the heating. You can also use a small attyny as an HW watchdog on the pi that will trigger a hw reset if it doesn't detect a certain condition

It could also be the article is not very new and lately battery technology allows for a bit higher standby voltage... who knows
I don't think 0.1/0.2V would make a huge difference in UPS runtime while on battery

I am no expert, but I suspect it is not charging the battery at full capacity to preserve it for a longer time as it will stay charged for most of the time. Also fast charging on a UPS it is normally not needed, so a slow charge at 0.3A is fine

@phamchien1112 maybe you can also consider this https://shelly.cloud/product/shelly-4pro/ , maybe a little overkill for your needs

Yes, in fact I bought the DPS5015 with its metal box and BT/USB interface and I can control it from the computer. Data logging of course is limited to the values on the screen