iNavFlight
erratic altitude from GPS ?!
Last Friday 2025 October 10, I went to the usual country side, to fly the (usual) wing.
I flew in manual mode and everything was normal.
Only at home, looking into the blackbox and the recorded video, I noticed the GPS altitude had been very erratic during whole flight.
GPS location was accurate, instead.
Can you figure out why (!) this happened? (first time I see such a problem in that model or other ones and I used this release since its delivery).
I attach dump and blackbox log.
I attach also a chart I obtained from the bb log, that outlines the difference between GPS altitude and baro.
Moreover, you can see the recorded video at https://youtu.be/7Q3IHl320ok , showing the problem (see e.g. at 2:00).
While this isn't exactly accurate technically, it's a reasonable way to look at it:
If I told you that I'm south of Oslo, I could be almost anywhere in the world. If I tell you that I'm a little south of Oslo AND Berlin is to my south, further way, you know much better where I am. Having the two reference points on opposite sides makes a big difference.
For east-west, we have satellite references to the east and to the west, and can compare. Similarly, you have satellites north of you and satellites south of you and the unit can compare the distances. For altitude, GPS can tell us that you are somewhere lower than the satellites. It can't say "I'm higher than satellite A and lower than satellite B".
GNSS vertical accuracy is much worse than its horizontal accuracy. The best modules in our aircraft vertical accuracy is 3-4 times worse. But older modules can be a much bigger difference than this.
For example, a u-blox M10 has a horizontal accuracy of 1.5m CEP. Which means that 50% of the time, the position will be within a 1.5m radius. Its vertical accuracy is 4.5m in optimal conditions.
For M8, you're looking at 2.5m and 25m for horizontal and vertical accuracy respectively. The vertical accuracy is 10 times worse than its horizontal accuracy.
Currently I have not much to add to this issue. I agree with all comments, but ... what I am not ready to agree is that this is just normal GPS behavior. My sensor is a M10Q-5883 by Matek. Normally, accuracy is good: about 90% times, altimeter is so reliable that I could safely land just looking at it only. Then comes the time that altimeter is no use at all. For instance, since I reported this issue, it happened another day. Considering that my flights last less that 10 minutes and stay in a zone of about 300 m radius, I would be happy with the barometer altitude. But if I disable GPS altogether, I lose localization and the RTH feature. One day I shall have a comparison device ready when the issue happens so I maybe understand the impact of either: GNSS signal, interference, bug of sensor or INAV computation. (BTW, situation at time of arming determines the rest of the flight: here, navPlot[2] is also plotted. Colors are different here)