Hi guys, we have boron LTE+ BNO085. Approximately once every 20 minutes, but it happens more often, the BNO compass indicators jump. For example, the device works steadily showing us the azimuth in the direction of increase or decrease as we need. After 20 minutes, the azimuth value jumped by 100-200 units and is now counted from the new value. We thought that the problem was disconnecting/connecting the device due to a weak connection, but disconnect is not always visible, and the change of mobile operator is also visible in console.
There is a keep-alive ping that occurs 23 minutes after the last publish, subscribe, function, variable, or the last ping. There will be a small burst of data transmitted and received. It's quite likely that if the device is stationary there would be no indication of this occurring in the console.
That shouldn't have that much of an effect on the compass, but it's possible if the cellular antenna is very close to the sensor.
We send it once every 1 minute. But if there is a disconnect or poor connection and the device switches to another operator, the compass data changes. It feels like something is affecting the compass, an anomaly or the device is initiating a module restart. I don't understand
Attached screenshots
In previous experience, I've used a sensor that was very sensitive to changes in voltage fluctuations. Are you powering it from the same power supply as the Boron? If you poll the sensor immediately after reconnecting, perhaps there's something still some load leftover that could affect the measurement. Maybe try adding a few seconds delay after reconnecting to reporting the sensor data.
That said, I would expect a change of a few percent and not have it point in a completely different direction. Could it be resetting to some default state that's different from your fully initialized state? Such as how temp goes between 100 and 34.6, that's pretty close to the conversion for °F to °C.
Are you able to get data from the device directly and get a better sense of when exactly the data changes?
Ya, we were thinking that it might be a psu. Gonna check it again. Same about reset to a kind of default settings, the question is why ?
It is also possible that the thermistor is connected, or the psu is partially out of order due to high temperatures. The device operates in a confined small space and can heat up to 65+ degrees.
Yes, we tried to track down some logic behind the crashes, and it seems to happen more often when the device is connected/disconnected from the network.
We will soon be able to test devices with the new psu and connection scheme. But for example, we tried the Photon + BNO on the table and after reconnecting it looked like the same problem. Value jumped.
It seems that Boron/Photon initialises or checks the modules when reconnecting to the network or changing operators.
Perhaps someone else has encountered this problem about BNO.
The module reacts to the movement, shaking, but when it goes smoothly and slow without sudden movements(like we need for our 360 turn in 24hr) it does not react to changes in any axis and sees it as drifting. It feels like BNO requires significant acceleration.
In short, I increased the sensitivity to zero, which is the highest sensitivity. Nothing...
If you tap your finger lightly on the device itself, you can see the reaction of all the sensors. The compass was calibrated. We use both the rotation vector and the gyro stab vector, nothing ....
We even found a register where BNO is not allowed to sleep. To exclude this problem. Nothing...
Adafruit library, sparkfan and custom library. Nothing...
Something more or less similar is when you reset the bno every second, there the indicators are stupidly messy, but on average something like the truth.
