Arducopter 3.0 RC2 Crash (Suspected Bug with outputs to ESCs)

  • Installed Arducopter 3.0 RC2, had verification error which I have seen several times without problem, re-calibrated quadcopter and went to the field
  • Plugged in battery at the field and the ESCs started beeping an error which translate to "input signal abnormal"
  • I whipped out the laptop connected it via USB to downgrade back to 2.9.1b
  • Then the ESCs stopped the beep error and made the "ready song" so chose not to downgrade
  • I unplugged the APM from USB and from Battery to validate APM was stable
  • Connected battery, Beep-ReadySong (Previously the copter ESCs would be Beep-Beep-ReadySong)
  • I believed the copter was ready to go though GPS was not locked and took it up for a test flight and flew it for a few minutes noticing it was acting a bit slower (less aggressive in stabilize mode) than I remember but all in all normal flying
  • Then as I was hovering it and clicked a switch on my Tx (non channel, just to start timer) it lost all power and dropped 30-50ft and well... broke.

I reviewed the tlogs (and tried to interpret the values) and believe I eliminated the correlation of hitting my timer switch and rather it seems like all the ESCs\motors at the same time stopped consuming power.

You can see in the attached JPEG where the decent happens and I go full throttle trying to recover but to no avail.

Can someone please have a look at the tlog to identify if it is a bug in the code or somewhere else.

PS, reflashed to 2.9.1b and the anomalous beep issues disappear

Views: 1231


Reply to This

Replies to This Discussion

Im pretty disappointed community! 94 views! Where's Randy?

when I did switch from 2.9.1b to 3.0 I did reset&erase the APM and install a clean firmware.

This makes all previous settings from 2.9.1b to be lost. Then I did redo all calibrations again, ESC, motors, TX, accel (level) and compass (dance).

Not sure if is mandatory to reset/erase, but I did feel that better and I did not have any issue with 3.x.

i think randy has a life outside arducopter, so let s him the time !

Do I understand right that you uploaded the new firmware, got a verification/checksum error and still kept on and went out flying?

Ive been flying the APM since Sept 2012 and have seen verification errors every time I have pushed a new firmware. I was originally concerned with checksum errors but recall a forum suggesting if I am using Mission Planner in a VM environment I will get those error and up until this point have never had a software related issue with the platform.

Ok, did not knew this. I use VMWare products all day in my job but never had troubles with mismatch on serial or usb connections.
Anyway, if you say you switched on a timer on your tx, was that the exact point the motors stopped? Sure you didnt had a mixer or something else on it?
I read something regarding escs in the drone-discuss groups, somewhere in the rc thest reports, maybe you find relevat info:!forum/drones-discuss

The RC2 ESC beeping is normal. It is because the APM was erasing logs during boot too much frequently. This has been corrected in RC3.

You must not fly with firmware verification errors. This could be the reason of the crash because of a corrupted firmware.

According to the log you have strong compass interference induced by power circuits (battery, ESC, power cabling...). Did you calibrate the compass, set the declination and apply the new compassmot calibration ?

Another bad point you have logged VCC values of 4.6 V near the limit of the AVR 2560 minimum supply voltage of 4.5V.

Witch GPS is it ? Do you have an idea why it did not had a 3D fix ? Where you flying in a clear sky area or urban area ?

First you should solve your hardware problems, perhaps you have a faulty APM. You should not have verify problems. Did it crash before ?

Second you should have a better power supply with 5.3 to 5.4V output so that the AVR is powered at its nominal voltage of 5V. Do you have the recommended power module fitted ?

Next try version 3.0 RC3, and report compassmot magnetic interference level. If it is more than something like 30 - 50 %, then you'll need to rework your power cabling, use an external magnetometer or mount the APM a couple centimeters higher.

Last, your GPS seems faulty, or you were in a very bad location or GPS outage period, or we have a bug in this firmware for GPS communication. Be warned that RC2 it is still a non release version and that you should be prepared as a tester to eventually crash because of that.

Let's wait for Randy's or someone else analysis to see if he can figure out some other causes.

Sorry to hear about your crash.

Is it possible that you forgot to charge the battery before the flight?  The sluggishness before the crash could easily be because it was running low on power.

I believe the low voltage or current battery failsafe triggered and this switched your copter into LAND mode.  The problem though is that this relies on the autopilot's alt hold working and the vibrations appear to be very high in your copter.  This led to very incorrect altitude estimates.  The copter thought it was climbing when actually it was falling.

As Olivier says, you also have fairly bad magnetic interference although this will affect your horizontal position control in loiter, rtl and missions and isn't related to this crash.

I just checked the voltage_battery field from the logs and indeed it drops as low as 9.5 volts.  The LOW_VOLTAGE parameter was set to 10.5 so it was significantly lower than the cut-off.  It doesn't trigger into LAND immediately as soon as the voltage drops it waits a bit to make sure that it's not a temporary sag perhaps as part of a fast climb.

Thanks Olivier and Randy, excellent feedback.


Thanks for the great interpretation of the tlogs! I heard you would be able to identify some problems and you did not disappoint. Ive learned a lot from what to look for already!

POWER: So first, the battery was fully charged when I went flying and I have not charged it since. I just checked it and it has a nominal voltage of 12.4v. I say it was acting sluggish but Im thinking it might have been because of the highspeed duct fan planes that were flying around at the same time. :P

I am surprised the APM wasn't getting enough power, I am using the 3drobotics standard current sensor which also powers the APM. The only other environmental factor was it was misting a bit but not enough to cause an electrical short as far as I am concerned.

GPS: I have been disappointed with the lag it takes to get GPS to lock sometimes with it sometimes not even locking for several minutes or until Ive taken it up in the air and brought it back down. I have moved the antenna to different parts of the copter but I am severely limited by the length of wire given. I also do fly mostly in wooded or urban areas which may contribute to the lock delay. I have had 3DFix, just this time I was getting impatient and decided I didnt need to fly with it on.

Are there any methods to test if the GPS is not working well? Im using the ublox lea-6 gps from 3drobotics.

Is there any recommendations on GPS placement given the future build out I am doing will be the same ESCs, similar distro board and have 915MHz Telem  and 5.8MHz Video Tx on board. I did have it sitting right on top of the APM before but a friend recommended I do not do that, but any other position on the copter put it closer to the ESCs.

LAND: Land isnt a flight mode i have configured or ever used, is that a default safety function?

VIBRATIONS: Since the crash I am now in the process of doing a full retrofit of the copter which will include much higher end blades, I am hoping that resolves the vibration issues. I will also be employing moongel to mount the APM

EMI: So my current airframe is a DJI F450 which integrates the distro board into the bottom plate and I have the APM sitting about an inch or more above on the top plate. the APM is in the dead center with the ESC as far as possible away from the APM. Now I am about to do a full rebuild on a TBS Discovery airframe, do you have any suggestions to minimize the EMI? Is there anything else like the Telemetry Tx that could be affecting the magnometer?

Again, thanks for all the insight!

Try balancing your props. This helps a lot with reducing vibrations.

Also, it's a shame log download isn't possible via telemetry adapter.

Do you have something connected on the APM that could tear down the voltage ?

Did you remove the Jumper JP1 to avoid powering the APM through the ESCs ?

Did you crash before with this board ?

See the power scheme here :

The GPS should get a fix in about 10 - 20 seconds in clear sky. If not then you have certainly RFI interferences or it is broken. Try to put a toroid on the GPS cable and see if this does help. A probable cause of RFI is the 5.8 GHz video transmitter. Try the GPS without it to see the difference.

It's better to put the GPS and magnetometer far away from radio and magnetic interferences. A mast is a good solution, lenght between 10 to 20 cm should clear any magnetic interference. The best is to use the new compassmot procedure to find the amount of compass interference you have and adjut your compass (or APM) height to get less than 30% interferences.

All tranmitters on board below the GPS frequency (1.6 Ghz) should have a low pass filter. This is the case for 3DR radios.

Land is a flight mode automatically trigged if a failsafe condition do occur and there is no GPS fix available.

Propellers are not the only source of vibrations but they should always be balanced before use. Bad or previously crashed motors, propeller fasteners rounout,  and too much flexible plastic arms can have a bad influence as well.

If you are running 3DR motors, prefer the new propeller fasteners, shorter and more precise :

With version 3.0 it is important to have a low vibration setup.

The Wiki is actually updated to better warn users about those things, low vibrations, importance of compass calibration, declination and dynamic compensation...


Reply to Discussion



Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2020   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service