chromvis

  • *
  • 117
Re: Revo AuxMag
« Reply #330 on: August 29, 2016, 12:14:21 am »
Hi,
I am not sure, but "Complimentary+Mag+GPS" I think does not include GPS and Mag in attitude evaluation and does not use EKF (may be developers could correct me). I do not see much reason to use this settings unless you are using CC3D FC.
Drift simply means discrepancy in in magnetometers calibration. There are some sort of a bug currently in "next" which requires work-around during calibration (https://forum.librepilot.org/index.php?topic=2187.0 ). Read recommendation by TheOtherCliff.

Re: Revo AuxMag
« Reply #331 on: August 29, 2016, 01:26:53 am »
Thank for the tip!

I'll give it a try tonight once my batteries finish recharging (and before it gets too dark out).

Re: Revo AuxMag
« Reply #332 on: August 29, 2016, 02:15:27 am »
Tried your workaround https://forum.librepilot.org/index.php?topic=2187.0 and while it didn't fix the drifting compass, it did give a perfect calibration, best setup I've had so far.

The PositionHold was rock solid, no drifting, no oscillations – perfect. VelocityRoam was still a bit skittish but I was happy with the results.

Hope to start INS13 testing this week. Thanks for the help.

Re: Revo AuxMag
« Reply #333 on: August 29, 2016, 11:19:29 am »
A question about compass drift:

I have my Revo 450 quad running better now, green across the board and a good, steady hover, but... I'm seeing a constant drift on the compass. If I calibrate everything and have the quad sitting on a level surface (not armed, no props, no interference, green board) I can watch the compass drift by about 1 degree to the East every 20 seconds or so. After a few minutes the heading can be 5 or 10 degrees off (always to the East) and counting. If I power-off and then power-up, then the compass snaps back to North but will start drifting again.

I'm using the M8N with the AuxMag connected to the I2C bus and the latest (Next) branch. M8N is mounted on a 8-inch fibreglass post and seems to be otherwise unaffected by power and motors (all cables twisted and routed away from mast).
Attitude Estimation Algorithm is set to "Complementary+Mag+GPSOutdoor" – seems to get the most stable results so far while I work my way up to INS13. To monitor flights I'm using OPLink at the ground station connected to Bluetooth which is paired either to my Mac (while at my desk) or to an Android tablet running LibrePilot2Go (while in the field).

Is this a calibration issue? Is it a hardware issue? Interference? Anyone have any advice or thoughts?

Thanks for the help!

I don't think compass drift is an issue when using INS13.  It may be that "Complementary+Mag+GPSOutdoor" only uses mag for GPS flight.

I would guess that you can at least slow the drift down by doing a thermal calibration.

Re: Revo AuxMag
« Reply #334 on: August 29, 2016, 11:29:05 am »
VelocityRoam was still a bit skittish but I was happy with the results.

VelocityRoam is ... that way ...

The reason is that you are commanding velocity.  Imaging driving your car.  When doing 99kph you command 100kph.  Not much change, so it is smooth.  Now when you are stopped, you command 100kph.  It is impossible to instantly jump to 100kph, so the best the car can do is use full throttle till it gets to 100kph.

If you suddenly tell the quad to do 5mps the best it can do is bank to max bank angle and hold that till it gets to 5mps.  The way to make is smooth is to slowly move the stick so you don't jump to 5mps, but slowly go 012345 mps.

Re: Revo AuxMag
« Reply #335 on: August 29, 2016, 04:47:00 pm »
TheOtherCliff: Thanks for the explanation about VelocityRoam, that describes it well. It makes sense when you think about it. And you could be right about the thermal calibration, I've always skipped that step so I'll go back and try it out to see if helps with the compass drift.

I tried switching to INS13 last night but the results were not encouraging. With the drone flat on the table and all calibrations done under INS13 outdoors away from potential interference, the PFD and compass were gyrating wildly all over the place. This is with a satellite lock on 13 to 15 sats. So I guess that I have some further reading to do to better understand this mode. If it were easy then it wouldn't be fun  :)

Re: Revo AuxMag
« Reply #336 on: August 30, 2016, 05:54:55 am »
Read aux mag wiki about board rotation.  For I2C aux mag (Pixhawk, two cables) you should use 180 pitch.

ggrif

  • *
  • 178
New Magnetometer tab in Attitude section of GCS in Next builds
« Reply #337 on: September 01, 2016, 02:51:49 am »
Feeling my way through all the new capabilities in Next 996, came across the new Mag tab, and managed to center all values.  Previous to Zeroing I had calibrated both on-board and aux mags.

Just curious if the wide spread in roll, pitch, yaw values means anything?


f5soh

  • *****
  • 4572
    • LibrePilot
Re: New Magnetometer tab in Attitude section of GCS in Next builds
« Reply #338 on: September 01, 2016, 07:11:30 am »
Hi,

This strange orientation values do not make sense.
See this post for the workaround about the calibration issue, maybe related:
https://forum.librepilot.org/index.php?topic=2187.msg15708#msg15708

Topic moved

ggrif

  • *
  • 178
Re: New Magnetometer tab in Attitude section of GCS in Next builds
« Reply #339 on: September 01, 2016, 03:10:52 pm »
Hi,

This strange orientation values do not make sense.
See this post for the workaround about the calibration issue, maybe related:
https://forum.librepilot.org/index.php?topic=2187.msg15708#msg15708

Topic moved

OK, will do. Revo is mounted forward and rotated 90 degrees, V9 is centered in X frame on 7" pole.

Plan to eventually place V9 directly above Revo as this has worked well for me in past.

ggrif

  • *
  • 178
Re: Revo AuxMag
« Reply #340 on: September 01, 2016, 04:09:49 pm »
Perfect now.   ;D

I think?

Re: Revo AuxMag
« Reply #341 on: September 04, 2016, 05:33:34 am »
You should be trying to get it to be zero no matter what angle you tilt the quad.

I bet you only get those zeros with those strange numbers when the quad is level.

ggrif

  • *
  • 178
Re: Revo AuxMag
« Reply #342 on: September 07, 2016, 12:33:09 am »
You should be trying to get it to be zero no matter what angle you tilt the quad.

I bet you only get those zeros with those strange numbers when the quad is level.

OK, didn't know about tilting the quad, need to read through the whole thread.

Sorry, but what "strange numbers"?

ggrif

  • *
  • 178
Re: Revo AuxMag
« Reply #343 on: September 08, 2016, 04:40:47 pm »
OK, now I got it!  Zero's no matter the orientation.

Always helps to read the whole thread, no matter how painful.


« Last Edit: September 08, 2016, 04:46:01 pm by ggrif »

ggrif

  • *
  • 178
Re: Revo AuxMag
« Reply #344 on: September 16, 2016, 10:38:01 pm »
Why would Aux Mag Orientation change from one day to the next?  :o

Twice now, following an erase and upgrade (on the same frame, same components, same layout), I've initially had to set yaw to +90 to obtain all zero's in the bargraphs. V9 GPS arrow facing front of quad.  Revo rotated -90. Both mags calibrated with no errors.  Performed auto tune (way cool btw) and continued to fly 'til battery depleted. 

Next batt checked mags, both consistently green, BUT  X=32, Y=10, Z=-2.  Changed yaw orientation from 90 to 0, all bargraphs now at zero, any orientation.

Still trying to Grok all the new stuff.