Altitude oscillation and crash

Could you tell me how to proceed in order to identify what is wrong with my drone?

I’m experiencing the following problem:
I started in altitude hold mode,
then, since there was some wind, i switched to loiter.
The drone stopped on that gps coordinates, then some wind arrived on the drone and the drone goes down of about 3 meters, and then it returned to the loiter height.
The drone did this for 3 times.
The third time it crashed.

he maintained the stabilization during these “power loss”, infact it landed in the correct position, even if the crash caused it to reverse at the end.
Note: the propellers never stopped completely to spin but the impression was that there was a power loss because of the oscillations in air.

Note: the drone has ESC calibration, radio calibration, accelerometers calibration, compass calibration.

It is a octocopter.

I supposed a damaged current sensor, but maybe in that case the apm should poweroff since it is powered by a ESC BEC.

Here annexed my logs, where you can see that the current has 3 oscillations. The last correspond to the crash.

Could you tell me how to troubleshoot this problem?

Please post a picture of your setup and describe the crash more accurately? Straight down descent until it hit the ground?

I can tell you that it looks like something is very funky with your barometer. Is it exposed to sunlight?

Please upgrade to 3.1.3.

Here i detailed my setup: http://ardupilot.com/forum/viewtopic.php?f=80&t=6525#p12693
why you ask about the sunlight?
the drone was inside my car under the sun for about 1 hour (without batteries)
While using the drone it was midday.
Do you know of problems of the barometer?
If there is a such kind of problem, are we able to detect the problem in any way by watching the logs?
You still watched my logs for this?
I’m really inexpert on logs analysis and sometimes things happens on my drone and i remain without words and without ideas because i miss how to understand the logs (or better, which signal to plot and compare with a “reference curve”).

Thank you for your continuous support on the forum.

Graph CTUN.BarAlt vs GPS.Alt

That will compare altitude measured by the barometer to altitude measured by the GPS. Baro has 80m spikes.

The barometer is sensitive to sunlight.

Are you powering your setup with the 3dr power module or some other way? Your Vcc seems a touch on the low side.

It could also be said that you probably would’t have these problems if you weren’t buying shameless clones.

i agree with you…

[quote=“jschall”]Graph CTUN.BarAlt vs GPS.Alt

That will compare altitude measured by the barometer to altitude measured by the GPS. Baro has 80m spikes.
[/quote]

I’m watching the barometer log and the gps altitude log, as suggested.
I numbered my questions:

  1. I never been to more than 20 meters height from the ground, so there must be something really wrong in the barometer…
    I suppose that the barometer should measure the height from the sea level… am i correct?
    Or are you expecting to read the height from the ground?

  2. I also opened the generated google earth file, and the google earth altitude graph is NOT coherent with the apm gps altitude log:

    It seems to me that the altitude in the kmz file is different than the the gps log and different than the barometer log… and i can see height jumps on google earth graph… how is these data obtained?(if it’s complicated to explain, don’t care this quesiton)

  3. Anyway, let’s back to the logs: do you usually see the gps altitude coherent with the barometer altitude on a working setup?

  1. Do you usually cover the APM with something dark? is the barometer sensitive to light, to electromegnetic wave generated by the sun or to heat generated by the sun? i think that electromegnetic wave can not be stopped by a cover. Did you see that my APM has the plastic cover on it? it is not completely transparent. Any suggestion on this point?

5)How could i deeply investigate on this barometer problem?

  1. A newbye question: the total number of samples is about 4000, which is the delay from a sample and the next one? how many milliseconds?

[quote=“jschall”]
Are you powering your setup with the 3dr power module or some other way? Your Vcc seems a touch on the low side.[/quote]

  1. no, i’m using a clone of the 3dr power module and it doesn’t include the BEC, for this reason i’m using a BEC of one ESC to power the APM. what do you mean for “a touch on the low side” ?
    I’m watching the CURR-Vcc and it seems to me to be constant at about 4,7V.