AirGradient Forum

Air Gradient Outdoor goes offline frequently

My AG Outdoor unit with firmware 3.3.7 continues to go offline, sometimes for several hours. I updated the firmware to 3.3.7 a week or so ago and it stopped for a day or two but then it has resumed flipping to offline mode.

In offline mode does it continue to monitor and store values or is all that data lost?

How do I determine why this unit is flipping offline?

It is installed on a post in a mostly-shaded location close to 2m from ground level. It is powered over USB from an inverter on my solar system. The inverter has two USB-A ports. One of those uses a short USB-A patch cable to connect to the AG USB-A/USB-C cable and the AG Outdoor unit. The USB connection is rated to 2A, 5V or 10W so it should have plenty of juice. I store energy from the panels in an array of 2 - 100Ah LiFePO4 batteries.


The white cable coming into the red inverter supplies power to the AG outside the greenhouse on a post.

The AG is shielded from the sun by using some louvered panels screwed to the post.

This was the original installation back in April. I have since moved it to a post on the opposite corner where it is more shaded during the day. That necessitated using a short patch cable at the inverter since it was about 3’ further to that post.

At the time that I moved the unit, I brought it inside and did the firmware update hoping to solve the offline problem.

My signal levels show to average -88dB at the site. There is no obvious correlation with offline mode and signal level.

Anyone have suggestions for a meter that can log USB power usage and save the data either to SD card or internal log in CSV format?

As far as I know it is not going to store any information locally when it cannot connect to the Internet. I know it has been requested and maybe some progress, but I don’t recall seeing it in any of the releases.

AirGradient does have an “Offline Mode” but that is about not trying to send the data to the AirGradient servers and I think an additional option if you didn’t want it to connect to wifi at all.

-88 dB is incredibly low signal, with -90 being basically no connection, so I think you are dealing with signal issues
Wi-Fi Signal Strength Basics | MetaGeek
Surrounding the devices with metal plates is certainly going to impact how well it can communicate over Wifi. I would start with removing the shields and see how it changes.

Do you have more than one Wifi Access Point? I’ve had some trouble with the ESPHome firmware when my device was connected to a further AP that it was sticking with, and rebooting the APs temporarily solves it.

Most of the other solutions I remember reading about were storing data locally over MQTT or something with Home Assistant, but since it looks to be related with overall wifi signal, those won’t really help. I’m not aware of a direct way to log the data to a USB/SD card short of another computer that could be connected over USB to gather the logs and then write to another file.

Oh geez. Am I dumb. I can’t believe I used those aluminum vents.
I did remove them a few minutes ago so I will see how that affects the signal level.

The average has been 86 dB instead of 88 dB so not much of a difference. Hopefully it will improve. The best I have seen is only around 78 dB since I installed in April but it has had those shields up the entire time.

This should be a real world example of the effects of using a metal shield on WiFi signals. There were two layers of thin gauge aluminum so that had to be a barrier.

I do want to be first in line for any solution that gets on-board logging so I don’t lose data. I am trying to correlate this with other sensors I have out here and holes in the data don’t help.

If there is a guide to setting up MQTT or Home Assistant I would check that out for sure.

I have a new router to set up a mesh network here to push a stronger signal out there but I wanted to check other things first to be sure that the signal was the issue and not a power issue. I have to take down and reconfigure my whole network, switches, etc to get this new one in the mix so I haven’t done that yet.

Okay. It has been one day, 24 hours, since I removed the aluminum louvered vents from around the AG.

Before I did this the 24 hour WiFi signal level showed an average of -86 dB on a 5 minute graph.

After they are removed the 24 hour average on a 5 minute signal graph shows -83 dB. That’s a 3 dB improvement, or a doubling of signal level at the AG if I remember correctly. It is still 3 dB below what is considered to be the minimum level for reliable connectivity at -80 dB.

With that in mind, the signal improvement has resulted in a 24 hour period with no notifications about the sensors being offline. Since being restored about 26 hours ago it has been online continuously.

I can see from the signal graphs that the signal level for most of the last 24 hours has been between -80 and -82 dB for about 12 hours and has been below -82 dB for about another 12 hours with only a few readings exceeding -85 dB. Overall signal levels are higher than they have been over the last 30 days.

It is safe to say that whoever put those aluminum louvers around that AG was thinking more about shielding it from the sun so they could get better temperature and humidity readings than they were spending worrying about WiFi signal levels.

Information about my setup - The AG is about 125 feet (38.1 m) from the router and is at a lower elevation than the router by about 2.5 feet (0.76m). The signal must travel through at least two interior walls with wood framing and gypsum sheetrock plus the exterior wall which is also wood frame/sheetrock inside and brick veneer on the exterior. The AG is mounted 6.67 feet (203 cm) above ground level on a post as in the photo.

You are still on the very low signal side, so if you can get it improved more, then your shielding may be able to go back up.

For whatever it is worth, I have my outdoor unit maybe only 25 ft away, but has to go through 1 interior wall, then an exterior wall and it is showing -52 dB, which is in the good quality range.

But at least you are above the minimum to keep the device able to communicate with the AG servers.

One option may be to have an external WiFi signal transmitter.

This can be a transmitter / repeater (I forget the exact term) either:

fed direct from the router via a LAN cable - this is the best way as the transmitter will be transmitting the maximum wifi signal strength

or

a repeater for the existing wifi signal that amplifies the signal at the repeater - how strong a wifi signal you get will depend on the strength of the signal reaching the repeater

The downsides are that you will need to have the transmitter / repeater outside the house, so will need a mains supply and a LAN cable from the router (if connecting to a LAN). Mounted on the house wall is fine, and needn’t look unobtrusive.

The other option is to have a similar arrangement inside the house, but as near to the external wall a possible.

For the moment I am pretty happy about how this has worked out. Removing the aluminum louvers resulted in a 4 dB gain from an average of -86 dB (24 hours @ 5 min graph) to -82 dB over the last 24 hours. As a result, in spite of the signal being below what is considered to be the minimum for a stable connection, I have not seen the AG go offline since removing the louvers. Signal levels have been in the range of -79 to -85 dB since removing the louvers 42 hours ago with signal levels below -84 db the exception.

I do see a curious pattern in the data graph. Signal levels are consistently higher during daytime by about 3 dB. The lowest signal periods are mostly after dark and highest signal periods during daylight.

My plan is to add a mesh router in a room on the exterior wall. This will place a router about 30’ (9.1m) closer to the AG and should be a permanent solution to the signal issue. I have the gear but adding it requires a full take-down of my network and at the moment there is still a need for constant connectivity due to work-from-home requirements.

The intended effect of the aluminum louvers was to prevent having the AG in direct sunlight during the last hours of the day (the hottest) when the sun will be in the part of the sky where there is no shading available. The louvers allowed air flow around the AG while reflecting sunlight (and WiFi, LOL). In the temperature data I can see a temperature increase in the late afternoon since removing the louvers so I will be adding some shielding on the west side of the post where the AG is mounted, probably made of wood painted white. Since the WiFi is on the east side of the AG it will not affect anything.

I have another environmental sensor package inside the greenhouse shed to correlate temperatures/humidity and the two units have been closely tracking each other since the AG was deployed. The one inside the shed is shaded all day so it serves as a tattle-tale for the periods when the AG may be affected by sunlight since the temp curve inside is stable. The sensors are less than 9 feet (2.74m) apart and both see the same airflow conditions though airflow inside the shed is more static in low wind situations.

I will update this as I make changes. For now it is working well enough for me to get back to other existing projects.