Space Goast's blog View Details
Archive for April, 2020
Posted by Space Goast | Apr 05, 2020 @ 03:07 PM | 5,931 Views
Spent some time today at the compound just testing and tuning. I have a wide range of airframes covering three flight control firmwares.
  • Strix goblin with custom vectored thrust VTOL pods, running ardupilot
  • Strix nano goblin with ardupilot
  • Volantex Ranger 2000 with ardupilot
  • Chameleon LR 6" frame with iNav
  • Lumenier QAV-R2 5" with betaflight

In the past I've never been happy with the "autotune" of ardupilot. I would typically try it, get some values, and then revert and try manual tuning because it felt sloppy. Today I changed the autotune_level parameter from default of 6 to 8. It warns that 8 is for advanced pilots. I have to say I'm extremely pleased with the tune that resulted. In my opinion this is what the tune should have always done. Worked great on my ranger 2000, but had oscillations on my nano goblin. I'd recommend everyone who has a plane that flies in FBWA to try autotune level 8 for large platforms and level 7 for smaller ones.

iNav... A couple years ago I posted about my experiences with iNav vs ardupilot. Long story short I had inav on my goblin(non vtol) and after several major crash sessions I swapped to ardupilot and haven't gone back. Since inav started as a betaflight fork I decided to build my new chameleon frame with iNav for RTH and arrow home features. There seems to be a major issue with altitude calculation in inav. In the megathread 747TestPilot did a lot of work trying to get altitude to work. I have spent 8 batteries tuning inav, and am also having serious altitude errors. It will swing hundreds of feet in an instant, usually going negative. I'll do another write up with more details but at this point I'm leaning towards flashing ardupilot to see how stable it behaves.

Quick video with the ranger after autotune level 8. It's so locked in I finally felt comfortable skimming tree tops.
FPV Compound 040520 Ranger 2000 (3 min 45 sec)