This is on my I-9PSL board running firmware 3.1.4. The other boards are all DIY PRO v3.3 PCB running the Beta version for v3.7 that includes Home Assistant.
I had an ISP / Internet outage last night and the VOC index has the weird reading that lines up exactly with when the Internet went out to when it came back online.
When the device can’t connect to the AirGradient server, it reboots after some amount of time, 5 minutes maybe.
When it reboots, it resets the VOC index to 100 and then adjusts from there.
So I think this would be the expected outcome.
Thank you for the explanation. Weird that it only affected the I-9PSL and not the older DIY Pro boards, and also for the I-9PSL the inconsistencies are only for the VOC Index readings and no other measurements (CO2, Temp, Humidity, pm 2.5).
I have on my desk in view but I guess I didn’t notice it rebooting throughout the day. Also they stayed in Night mode past the normal time but that is expected with no Internet to tell them to change to Day mode.
The older DIY board didn’t have a hardware watchdog chip on them, while the new model does. If the chip isn’t refreshed each time that it sends data to the AirGradient servers, it will reboot after the ~5 minute interval (Unless you configured the device in Offline mode)
Only the VOC Index reading is impacted due to how that chip does readings and defaults to a 100 baseline on boot and goes up or down from there, depending on how air quality improves or degrades from that point forward. No other sensor in the AG works like that.
I would have expected you to see the bootup screen if you were watching though.
It’s off to the side of my desk and since it was still in Night mode maybe I didn’t notice the screen change, or it just boots up really fast 
Again thank you for the quick reply and clarifying that its the watchdog chip. I was poking around the Dashboard but don’t see an option that we can change the Watchdog reboot interval.
Leaving the default at 5 min but having the option for longer like 15, 30, or 60 minutes would be nice. I guess that the AirGradient Dashboard doesn’t care if it is rebooting every 5 minutes because it is not getting the data anyway in an Internet outage scenario, but as you can see it really affects any local data collection.
I actually don’t think it’s related to the hw watchdog but the WiFi manager trying to connect to WiFi and rebooting after 180 seconds to retry.
The OP mentioned their ISP had issues, not that their internal WiFi was down, so I expect that it was connected to a Wifi network, but had no internet connection.
If it was not connected to Wifi at all, the display would have mentioned that. If it didn’t have a connection to the AirGradient servers, it would say “Server N/A” at the top.
Last time I dug into the code, it only refreshed the hardware watchdog after a successful POST to the API unless configured for Offline mode. In this scenario, with Wifi connected, but no Internet connectivity, it would keep trying to POST but never successfully communicate, so it wouldn’t refresh the watchdog resulting in a hardware reset.
At least that is my understanding.
Correct that WiFi and local LAN was not affected, just the Internet was unavailable.
I believe I did see the “Server N/A” message a few times, but again it is off to the side of my desk and I wasn’t really staring at the screen all day.
