Hi All,

I have flown my quad 12 times now, using the stock pid param as per 3.2.1

I get ~25min per flight with small HD Tube nose-mounted FPV cam.

I now want to attempt Auto-tune to optimize the rig.

 I don't think I am experienced enough to do PID tuning manually.

However, I read somewhere around that Auto-tune can fail on bigger, heavier quads, with low kv motors.

My build is based on the Tarot 650 Sport Quad Frame (X-config)

FC - PixHawk (3.2.1)

Motors - Tarot 4114  320kv    

Props - CF 1550

ESC - HobbyWing Xrotor 40A

Batt - 8AH 6S

AUW : 3kg

1) I would like to know if Auto-tune can be done successfully on this build.

2) Do you recommend that I first upgrade to 3.3.3

Thank you for your valuable comments.



Cape Town, South Africa

Views: 3633


Reply to This

Replies to This Discussion

Hi Fnoop,

 Once I am in an open area I will do exactly that, up to now I have done

test-flying in my backyard, in the neighborhood. 

I have tried Auto Analyze but having errors loading it, loading the .bin file

See attached err refr  pics

During this procedure the Craft is not connected to MP

MP 1.3.37

Any reason for this reason.





All I can suggest is redownload the logs - you can also pull the sd card out of the pixhawk and put in your computer.  MP is a bit sensitive about bin formats but I've got the same version as you and it works for me.

..or try "APM planner 2" , never thought of ".bin formats"  using it.

I use APM planner primarily, but it doesn't have auto analysis does it?

not sure what you mean by auto analysis, in Ardupilot source Tools/Loganalyzer there is LogAnalyzer.py , a script that outputs things like :

Log size: 21.09mb (718169 lines)
Log duration: 0:21:56

Vehicle Type: ArduPlane
Firmware Version: V3.3.0 (6be0932d)
Free RAM: 0

Test Results:
            Parameters:  GOOD                                                              
                   GPS:  GOOD                                                              
          IMU Mismatch:  WARN       Check vibration or accelerometer calibration. (Mismatch: 0.86, WARN: 0.75, FAIL: 1.50)    [GRAPH]
              Brownout:  GOOD                                                              
         Dupe Log Data:  FAIL       Duplicate data chunks found in log (131714 and 194021)     [GRAPH]
                   VCC:  GOOD                                                              
               Compass:  GOOD       mag_field interference within limits (12.69%)          
                                    Max mag field length (594.43) > recommended (550.00)
        Event/Failsafe:  GOOD                                                              
                 Empty:  GOOD                           

is that what you mean ?

Hi All,

Just an update -

Friday was an excellent day here in Cape Town, Bloubergstrand, to do flying.

Good Vis, hardly a breeze.

I finally took the plunge, went to a local open lawn-area and executed Auto-tune.

It worked!!    AutoTune:Success   :)))))

I saved a copy of the param-settings prior to ATun, such to compare the ATun results, quite a difference.

ATun was configured for  R&P&Yaw axes.

My observation was that it took about 18 min to complete, then again, I had to re-position the vehicle  a few times.

After saving the ATun params, I tested in stblz, alth, posh, the vehicle 'felt'  good, compared to pre-ATun.  Unfortunately I only had  about 6 min.

Once again, thanks to everybody who provided positive comments, suggestions and pure sound advice based on

your experience. Your input built-up my confidence to take this plunge. Yes, I know it is actually small steps, in the bigger picture, but for me starting out with this hobby it is gaining quite a bit of ground getting towards my better understanding of the environment, ultimately to enjoy the hobby.

It seems to be a great day again for flying, so I am going see how the vehicle responds.



Hello, I've done an auto tune on almost the same setup, the quad became a little bit under tuned. I have even try lower KV motors too, those one became over-tuned. Good luck! 

Well done Gerhard, happy flying :)

Most people seem to get better results when doing at least 2 or three autotunes - if it has a better starting place it ends to get a better end result, and the subsequent autotunes are much quicker, at least that's what I found.

Did you have to change your throttle gains much to get alt-hold to work?

I am having similar problem with two different vehicles(110kv(4kg) and 360kv(2kg)) with pretty much identical problems. When I switch to alt-hold the vehicle drops couple meters instantly(as motors practically turn off) - I have not stayed in alt-hold mode long enough to know if it would steady out. Once the motors stop the vehicle pitches over too so its not very easy to test. I can attach logs but on one vehicle I checked Vibration and it was way within bounds. I also set the throttle mid to be correct hover level. Not sure what else to change. Any tips would be appreciated. 

The single problem I have is that the quad is loosing altitude with forward flight, even in alt hold mode. Even the slightest forward flight will result in 30-40cm altitude loss... Once it stops it will climb back to the original altitude... It is very disappointing.

Ouch that sounds pretty bad.  I've seen alt-hold do that but on a much smaller scale - a momentary dip in throttle but almost instantly recovers, but haven't seen it on smaller higher kv craft.  You might want to ask here:


As that's a more active support forum than here.

That's a completely separate known issue, it's to do with aerodynamic low pressure bubble in front of the craft in forward flight fooling the barometer to think it's higher than it is, so the FC compensates by lowering altitude.  When you stop the low pressure bubble equalises and the FC compensates by raising altitude again.  It's essentially a physical design 'flaw' that most multirotors have, there has been various discussion on how to fix it with static ports, software fudges etc, but I don't think anything concrete has been put in place yet unfortunately.

Reply to Discussion


© 2020   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service