Just what I was going to do right now.
So I modified rc.local to start the binary from your .deb package.
At first all looked good, a fix with 4 sats, not bad right after booting. But then it lost the fix and did not get one again. I started the u-blox server without killing APM and it could not get a fix either. It did not even show a constant stream of sats, just sometimes a spike here and there.
I will test again tomorrow.
Three other things I noticed:
- Through the short period with a sat fix, the copter was not in the right position, but it was in one place, not moving around like before.
- AHRS_EKF_USE is there in your Arducopter build.
- With AHRS_EKF_USE 1 the compass was constantly rotating, as soon as I disabled it, the compass stopped.
Does it show same behaviour with both antennas? Probably our initial guess about the faulty antenna was wrong and the issue is actually in your NEO-M8N itself.
I will test this tomorrow. Moving the old cable did change the SNR, moving the new cable does not. But I will test it with changing both antennas while running u-center.
If your tests tomorrow do not result in finding the source of the problem, we can send you a new board to test it against yours.
I think I have two faults, one was the antenna cable, the other seems to be the connector on the Navio. If I use the old antenna, I hardly see any sats in u-center, only spikes if I move the cable. With the new antenna, I get 5-8 sats indoors and if I apply light pressure to the connector (pressing it down in the socket) I suddenly get a better SNR or a fix and other sats show up. Sometimes this holds if I remove my finger, sometimes it is gone immediatly. I really do not know what could have caused the connector to fail. I did about five flights with the Navio+ and I plugged the antenna in once during installation and left it there, until the problems began. The u.fl connector on the Navio that drives my rover saw a lot more abuse and is working fine.
@schuermannsebastian We will arrange a replacement for you tomorrow. Sorry about the inconvenience, this is a really strange failure. I would like to get to the bottom of this and if replacement solves the problem would ask you to send your board to us for investigation.
I never had something like this.
My problem seems to have been trifold. First the antenna, then the Connector/Navio and third the RPi image or the APM binary.
With the new antenna and Navio+, U-Center got a fix almost immediatly, so I went to the flying field and tada! nothingā¦
APM:Copter would not get a fix, no matter what I did.
I drove home and to get a baseline, I took another SD-Card and installed everything from scratch. Now only APM is running outputting telemetry to my PC and Missionplanner shows a fix with 9 sats indoors. I did no rpi-update or apt-get upgrade this time.
I will finish the APM setup and go out again tomorrow. LetĀ“s see what happensā¦
I almost forgot about this. Nobody ever answered, if such high offsets are in order.
@schuermannsebastian I have been reviewing our copter vibration levels and we are as well in Ā±0.3G range. Navio+ is mounted with dual-sided mounting tape from Hobby King. I will further investigate methods to improve it and will write if I find a better way.
I have looked at your graphs to see whether your mount has better immunity to vibration and noticed that scaling on the two graphs is different, this is why vibration looks different between the flights.
Regarding compass offsets, I have similar values, also higher than APM limits. I suppose it might be due to our magnetometer being different, it has wider range and higher resolution. As only normalized values are used in the flight code, we might have missed scaling, which should only affect pre-arm check. I will have a deeper look into it.
Thanks for your reply, Igor. You are right about the different scaling in the two graphs. My fault :-(.
I just changed from 10mm square aluminium tubing, to 16mm carbon fibre tubes for the arms.
Now all flex in the arms should be gone. I will see what it does for the vibrations.
I did two flights yesterday, with Arducopter 3.3-dev from the emlid deb package.
The day before, I had an extensive balancing session to get rid of the vibrations, which are still messing with althold.
At the flying field I met a buddy of mine, who just got into the hobby. He built a relative cheap quadcopter to learn how to fly.
11ā¬ TBS clone frame, cheap motors, Gemfan style carbon twill props, RTFHawk (Pixhawk) flightcontroller. All put together without even thinking about vibrations or balancing and I almost started to scream after I had a look at his dataflash logs.
The vibrations looked almost like the copter never flew. Less than 0.5G on all axes. He just drilled holes in the RTFhawk case and the frame and put rubber dampeners through them.
My copter did his usual dance in Althold and to see if it would make any difference, I removed the earplugs from the rubber dampeners. On the second flight, the Althold was working without the copter jumping up and down.
Autotune went through a few roll movements, but suddenly the copter nearly tipped over, so I aborted autotune.
The next thing I tried was PosHold mode. Everything looked good for about 3 seconds, then the copter started to pitch up and fly backwards. I switched back to stabilize and landed, because I really had enough at this point.
GPS had a fix with 8 sats maximum, clear day on an open field.
how do I get a picture of vibrations like that?