This is a discussion to focus the feedback and support of ardupilot on the PixRacer board from AUAV.co.
We believe that all teething problems encountered are now resolved and so we recommend people load Copter-3.3.4 which is available for all multicopters through the Mission Planner's (and other GCS's) Install Firmware screens. Note that because regular Pixhawks are still on Copter-3.3.3 the install icon may show Copter-3.3.3 but rest assured, Copter-3.3.4 will be installed (you can check the version by looking at the very top of the Mission Planner).
Replies
Question about a Pixracer version. If were waiting on 3.4 dev for terrain following and it won't be ready for some time will the changes be ported back to 3.3.3 for Pixracer? Seems like a waste of time.
Also the conversion of PID's to 3.4 doesn't seem to be adding up.
RATE_PIT_D 0.01055579 -> ATC_RAT_PIT_D 0.0036
Thanks for the report on the PID conversions. I'll check that out.
The terrain following isn't that far off so I think it will be more efficient (in terms of developer time) to get Copter-3.4 out for beta testing than to backport everything to Copter-3.3.
Also lately I have had two fly away episodes where the copter just after lift off suddenly starts to lean left and crash. I thought it was related to an auto tune event but now I think its something to do with 3.4 and possibly the Pixracer. These both occurred in Stabilize mode so just basic controls. I looked at the log but don't understand why it wants to lean to the left even with full right input.
45.bin
Looks like I got the same issue with my Pixhawk-Clone, so it shouldn't be Pixracer related:
http://diydrones.com/forum/topics/copter3-4-dev-crash-ekf-check-2
Julian,
Thanks for the report - that adds valuable info that it's likely not pixracer specific. It does look similar to the other crashes in that the DCM and EKF roll diverge. Paul Riseborough has had a look at one of the other logs but couldn't determine the issue. What he needs is logs with the IMU data at a high rate (400hz to be precise). That can be done by finding the LOG_BITMASK param on the Standard Params page and checking "IMU_FAST" (and pressing the Write Parameters button).
By the way, we've moved the APM:Forums to a new discourse server here. We haven't started official Copter 3.4 beta testing so I'm not actively monitoring it yet.. but that's where I plan to do most of the support for the next release.
Michael,
There are two strange things I see.
1. the length of the compass vectors change a lot. It may be interference from the power wires but I'd guess it's something else (bad calibration?). In particular see the red and green lines below. Also it looks like Mag1 has failed so perhaps we're not handling the failure properly.
2. The venerable DCM and EKF attitudes don't match well at all. This can happen sometimes but it's rare so we should investigate more.
The Mag1 failed because it crashed and ripped the wire out of the GPS socket. I guess it didn't fail then just stopped talking.
Hi Randy,
I was playing with my Pixracer yesterday (after solving the driver issues on Windows) and also have some strange "problems" with the compasses.
I calibrated them several times both with auto accept and manual accept. But then the GCS periodically tells me "Compass not calibrated". The two internal compasses have relatively high field length (>500, <600) but the compass variance is in the green range. So it should be fine.
If I disable the internal compasses it seems fine.
Additionally:
I tried to arm indoors and set arming checks to -9:no GPS as well as to 0. But all GCS were still insisting on a 3D fix in the PreArm checks.
Cheers,
Thorsten
@Thorsten, how did you fix your driver issues on Windows? I have not been able to connect reliably to the PixRacer. I can connect reliably to a PixHawk.
AC3.3.3. Windows 10. Latest Mission Planner.
I made a short test and calibrated the compasses again in 3.3.4.
There is a rather large difference in the Z offsets of the second compass:
3.4dev:
3.3.4:
-
69
-
70
-
71
-
72
-
73
of 120 Next