APM Copter 2.9.1.b has been released to the Mission Planner and is also posted in the downloads area.
2.9.1b is a maintenance release incorporating changes to 4 default parameters that have been optimized since 2.9.1 was released.  There are also 4 minor bug fixes.  There are no code function changes or additional features.
The changes are:
1) reduce default INS_MPU6K_FILTER to 20hz
2) reduce InertialNav Z-axis time constant to 5 (was 7)
3) increased max InertialNav accel correction to 3 m/s (was 1m/s)
4) reduce yaw_rate P default to 0.20 (was 0.25)
5) bug fix for alt_hold being passed as int16_t to get_throttle_althold_with_slew which might have caused problems if you climbed over 320m.
6) bug fix for throttle after acro flip (was being kept at min throttle if pilot switched out of ACRO mode while inverted)
7) bug fix to acro trainer to do with roll correction 
8) prevent cli from being entered more than 20seconds after reboot.
Users are recommended to:
    Update Mission Planner to 1.2.41 and
    Update their flight code to 2.9.1b from the Firmware tab on Mission Planner. There is no need to go through a new configuration process in the Mission Planner.


Views: 99886

Reply to This

Replies to This Discussion

I just updated my APM to the latest version of 2.9.1.b and then i did an auto trim i found my Quad flew more stable with the default setting's then it did after the auto learn why would it do this ? i am also going to try my first loiter and wish to know if the setting's for loiter are in inch or meters.

Thanks for all that you guy's do 


Marco, I apologize for the delay on my end. I've been busier than normal with work and family the past couple weeks, and to top it off I completed a small quad build/tune/maiden over the weekend that ate up the time I needed to prep for my test flights. Lots has changed with logging in the new code, which requires me to go through and double check my settings. Also I think even my OSD has to be verified before I go up... older minimosd-extra may not work with the changes to mavlink... might have to hack the latest code to get that working too. Anyhow, the small quad is dialed so my APM rig is again proirity #1. I am 99.9% sure I'll have something to report on this weekend. ;)

thatMerriam-Webster: that definition: the person, thing, or idea indicated, mentioned, or understood from the situation.

Hello all,

I just ran into some strange effects.

Setup: APM2.0, Arducopter 2.9.1b self compile (due to my flashing issue with APM).

Story so far: After the initial update to 2.9.1, Hexa flies quite nicely in Stabilize and Alt-Hold - better than ever before in fact. However - loiter is a BIG no-no. The moment I switch modes, the copter takes off in random directions and occasionally at scary speeds. I can stop it though, if I just steer something - anything. In that case it just flies around like in stabilize until I release the sticks. a few moments later it goes crazy again.

OK - the Mediatek GPS has not been known to be great and I had been planning to upgrade to the uBlox LEA anyways but for the record: Loiter DID work with 2.7. and 2.8 - just not as well as I would have liked.

So I bit the bullet and got a uBlox and also a pair of 433 MHz radios for telemetry. The radios are a story in themselves but I have them working now (also within legal boundaries in Germany, I hope).

Conversion today: unsolder the MTK from the APM2.0 shield (check - no damage, no shorts); I don't want to carry around stuff I am no using. Connect the uBlox - works like a charm at first attempt. Get a 3Dfix with 8 sats and HDOP1.9 indoors.

Here comes the fun now: with my electronics stack sitting on the desk, fed by a battery, I connect APM wirelessly and notice how the articficial horizon is super-nervous. In fact it does all sorts of random stuff and the GPS also wanders +-5 meters.This is with the board lying on the desk. When I move the board, things stabilize quite a bit and the horizon follows my movements. Set it down and it gets nervous again. Sounds familiar? My Loiter-Behaviour?

Disconnect the GPS (which is possible now thanks to the pluggable connection; the MTK before was always there) - artifical horizon dead stable and just does what I do with the board. Reconnect GPS - nervous; disconnect - stable...

Is it possible that there is a fundamental flaw in the code somewhere that mixes up the slightly wandering GPS coordinates with INS-information? Interpolates GPS movement and uses that as input to the INS?

I could imagine that being

  • in the Serial code - maybe mixing a buffer for the GPS and the MPU6000 (which is on I2C, I believe?)
  • In the new INS code - as mentioned, this behaviour was not there in 2.7 (I didn"t fly 2.8 a lot, but generally Loiter was stable as far as I recall)

I recall seeing a thread about Loiter performance with uBlox and MTK a while ago but I can't find right now.

Maybe someone can enlighten me.

Regards, Otto

Hi Otto

You will get lots of multipath issues and lots of jumps with the GPS position indoors.  In fact all bets are off indoors or outdoors and close to buildings.  The code is faithfully responding to those jumps in position and mixing the GPS position with the INS.

The MTK GPS w/ 1.6 firmware did A LOT of averaging so that the position was stable when you were not moving.  That was really good when your vehicle was sitting still on the ground but not such a good idea when it was flying around.

Hi Marco. They have taken into account the response of the accelerometers (inertial controller) for Large  Motors?. The PID minimum range will be enough for this system I just built:


Initial Design with "eCalc"= OK...

  APM 2.5. Firmware 2.9.1b
  U-Blox GPS
  Telemetry 3DR
  APM power module with current sensor "bridging"
  6 NTM-3548 motors. 900KV / 815 Watts
  APC 12x6 Propeller
  6 ESC F60A (HobbyKing) with "Simok Firmware" (Homemade)
  Hexacopter Aluminum Frame (Homemade)
  Lipo 2x5000 / 30C "Nanotech"

  Total weight: 3500 grams


With this Hexa Heavy, calibrating "AltHold" and "Loiter" mode that your values ​​would start in "Throttle Rate", "Altitude Hold" and "Throttle Accel". Still NOT done the first test with propellers?

Jittery HUD is a known issue, and won't affect it in flight - if you arm does it settle down?

Yes Rafael, tuning to do on a fairly heavy drone is not a problem, but the inertial x/y works worse on the waypoint functions and fine with little quad, I've to talk to Randy and Leonard about this, although I have already pointed out the problem during my test (i've only heavy quad).
Do not worry, I've already told Randy that as long as I do not see the V3 be good on my quad does not release it! :-)

Thanks Kevin, i wait your report, even though I have already ascertained with quad small AC V3 is very good.

Any idea's when a stable version will be released that sorts out the loiter and alt hold issues?

I use these features a lot and haven't been able to sort them out since 2.9.1 was released, my quad just goes nuts when in either of these two modes. Works brilliantly in stabilised or auto modes.

I went through the vibration testing when 2.9.1 was released but was told by Randy that the loiter and alt hold will be skittish until 2.9.2 is released even though my posted vib results were very good.

Might have to go back to 2.8 as that was really stable in loiter until this issue is fixed!

Is a know problem, induced by the inertial controller, not present in 2.8.1.
In V3 the upcoming Loiter is awesome as I have shown in my videos, but unfortunately those are wobbling.
I have already asked to solve the problem but for now there's no solution, and is very tied to the type of motor and esc you install.

Type of motor and esc, why is that important?

I am using DYS controllers http://www.coolcomponents.co.uk/catalog/electronic-speed-controller... and Torxpro motors http://quadcopters.co.uk/torxpower-2216-pro-short-shaft-brushless-m... with graupner eProp 10 x 5 props.

Yes, is important because the inertial creates many rpm changes to motors during the corrections, for example, we have established with SimonK reflashed ESC these wobbling are more evident.

Reply to Discussion


© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service