Custom drone Stabilization, some askings

Hi,
My drone ( it’s for paraglider take off ) it’s no standard.

  1. propellers directed downwards
  2. the propellers overlap on 2 levels
  3. the center of gravity is high
  4. Thrust max is above 60kg

After lot of tests and calibrations at this moment i have no bad stabilisation but not perfect.

If AutoTune is safe for my drone ?
What i can do for better stabilisation.

Alt hold balancing up/down and sometimes loiter had a big oscilation.
Stabilize mode working no bad.

Spin_Max = 65% then stabilize is ok, else is very sensitive for RC throttle

any tip will be of great help to me.

https://youtu.be/Ne4-zsPz8YE2020_10_10_0_loiter_dobrze.param (17.2 KB)

Log File C:\Users\Prokopcio\AppData\Local\Temp\tmpC0B1.tmp.log
Size (kb) 101551.9541015625
No of lines 1373651
Duration 0:05:32
Vehicletype ArduCopter
Firmware Version V4.0.4
Firmware Hash 40502bd9
Hardware Type
Free Mem 0
Skipped Lines 0
Test: Autotune = UNKNOWN - No ATUN log data
Test: Brownout = GOOD -
Test: Compass = GOOD - mag_field interference within limits (14.96%)

Test: Dupe Log Data = GOOD -
Test: Empty = GOOD -
Test: Event/Failsafe = FAIL - ERR found: FS_THR
Test: GPS = GOOD -
Test: IMU Mismatch = GOOD - (Mismatch: 0.22, WARN: 0.75, FAIL: 1.50)
Test: Motor Balance = GOOD - Motor channel averages = [1325, 1309, 1334, 1322]
Average motor output = 1322
Difference between min and max motor averages = 25
Test: NaNs = FAIL - Found NaN in CTUN.DSAlt

Test: OpticalFlow = FAIL - FAIL: no optical flow data

Test: Parameters = FAIL - ‘MAG_ENABLE’ not found
Test: PM = GOOD -
Test: Pitch/Roll = GOOD -
Test: Thrust = GOOD -
Test: VCC = UNKNOWN - No CURR log data

Ohhh yeah it’s safe before you do this

This is given by @xfacta thanks again :kissing_heart:

Happy flying :slightly_smiling_face:

2 Likes

before going for autotune send a log file with log_bitmask = 196607 and at the least 1min of hover in althold or loiter to check that everything is ready for autotune
your video shows some problem

1 Like

Thank you,
now i see (separate topic) That my AltHold isn’t good if Acu is full.

I’m after first axis ( ROLL ) AutoTune.

Can you check if all is ok?

it was needed near 90% of my batery ( 12mins for one /roll axis only )

https://drive.google.com/file/d/1Qhhtusa22qpomIA5n9TLzKhPTBt3QULf/view?usp=sharing

Most of the time when its takes too long its meaning that you have noise on gyro
i can see asymmetry response in your autotune log

but to see that your new PIDs are good or not please send a log with 1min of hover in althold or loiter with new PIDs

Make another flight with some aggressive Roll inputs for review. Analysis of an Autotune flight is mostly useless unless you fly around after it’s complete with the new PID’s.

1 Like

After autotune my pids are better as previous, also I took a chance to rewrite the values from Roll to Pitch because my dron is very symetric, now also Pitch is better.

it’s no bad flying with no agresive, with smooth controlls.

but

If i use aggressive tilting (to max angle max) then after relase stick after return to horizontal still roll to far (can’t stop rolling in horizontal position ) The same with Pitch. Sometimes very big ( I got scared )

From this log i haven’t video but i think that my description is clear.
Log show this problem on Roll vs DesRoll and Pitch vs DesPitch

Here is my logs and parameters:
https://drive.google.com/file/d/109lYzMo3XVDqfH1T5xFLa0uScA9BP8bo/view?usp=sharing

If i’m controlling smooth ( Auto is set to smooth too )
Then all is looking no bad like on this video :

Looking good Łukasz! But you are right there is overshoot in Pitch and Roll. Pitch is more pronounced. I would say the easiest way to address this short of manual tuning is to set the AutoTune aggression to 0.1 and run another Autotune. The idea behind running the 1st one at .075 was to be cautious. Big quads are a bit challenging in this regard and there is a lot of mass to move around with big props turning relatively slowly.

I think you are right that you can run a Roll or a Pitch tune and copy one to the other but there is a difference (pitch tune is worse) so at some point you should probably do them independently

Also, now that you are done with the notch filter (unless you add more battery weight) you can set the INS_LOG_BAT_MASK back to 0 to keep the log files of reasonable size when your flights get longer…

Note: Your lower MOT_SPIN_MAX may be effecting the authority you need for aggressive maneuvers. If the next Auto Tune doesn’t help the overshoot problem you should condier increasing it.

1 Like

@prokopcio i don’t recommend autotune in this situation

Most pilots will look to move to Autotune as quickly as possible once their aircraft can hover safely in AltHold. Before Autotune is run the pilot should ensure that the current tune is good enough to recover from the repeated tests run by Autotune.

first try some manual tune to get it stable as you can before autotune

still noise on accelerometer is high


how FC mounted to frame ? could you please send some picture of it ?
also set the INS_LOG_BAT_MASK back to 0 as @dkemxr said will speed up your autotune

1 Like

I know that it’s not standard copter. But i’m fresh with drones topic. I’m learning avout it for this project only. I had no knowledge beforehand.
I’m too poor for handly set better parameters without your help.

Adding some mass to FC might help you to reduce vibration

About manual tuning this link will help
https://ardupilot.org/copter/docs/tuning-process-instructions.html#manual-tuning-of-roll-and-pitch

i tryed it year ago but working not well for Autotune.
Then some people experimented with my parameters and i was flying not bad - look video in first post in this topic, but with noises on rate.
Do you tink that in my case Autotune parameters are not good for me?

Maybe good is Dave’s idea for AutoTune again with agress 0.1

i think that vibration isn’t problem in my case.

Now It’s flying realy very nice but for this project ( for throw mode ) i think that must working perfect!

Wchich parameter is responsible for my overshoot of return from high angle only?

I think the vibration level is OK. It’s possible this is from the overlapping props so there is not much you can do about that. Hosein’s suggested of adding mass to the Flight Controller could help. But the actual mass to use may take some experimentation and there is a chance it could be worse on your craft. The best way to do that on those FC mounts is to stick a lead weight to the underside of the top mount plate. These weight are available at RC shops for Planes.

Remind me what the props size is?

Your question about which parameters.
ATC_ACCEL_P/R_MAX
ATC_ANG_P/R_P
ATC_RAT_P/R_P and I

They all work together.That’s why an Auto Tune is probbaly your best option. I manual tune but it’s very time consuming and can be frustrating if you haven’t done much of it. In the Tuning guide there is a good manual process to follow for Pitch and Roll if you want to give it a try. You really want to use In-flight tuning with a pot on your radio.

30 inches proppelers ( MF3016 model )

i using tuning by pot in my radio and i know how it works but i don’t want to crash my drone if i will change to poor value.
This overshots was very scary !

But strange that only very big one without next oscilation ?

i analise diagrams and i think that main problem can be with too hard for stop my big drone i will try to increase MOT_SPIN_MAX because now it’s very smooth to controll ( previous i didn’t know that MOT_THST_HOVER need lot of time to learn and now is below minimum = 0.15 for my big power )

How i can check if FC want to stop rotation ?
on diagram looks like not want to stop first overshot but can stop second without problems. It’s look very strange for me.

overshots2

I am after more tests. I noticed that it only overshoots one way for both roll and pitch. Always drops the same corner of the drone . I looked at the equipment and on this engine I have an older version of the T-Motors Flame 80A regulator - the other 3 are in the V2 version, but the seller assured that they were compatible with each other.

could this be the cause of this flight condition?

Also one motor (on the other output) is the older T-Motors P80, the other 3 are T-motors P80 II, the manufacturer also ensured me that is compatibility.

I have read recently that the Flame ESC’s have a thrust linearization feature which could require a non-standard MOT_THST_EXPO value. I have no direct knowledge of this but have seen suggested values of 0.2 (or lower). Are all the motors at least at the same kV?

Yes, of course all motors are 120kV ant the same proppelers.
i also read about linerization of ESC, but i don’t know what MOT_THST_EXPO value is the best for me.

Year ago i did tests (one proppeller) but i not repported stick/pwm position :frowning:
it was only for check power consumption.

  1. 11A / 550W / 5,5kg
  2. 13A / 650W / 6,7kg
  3. 18A / 900W / 8,6kg
  4. 24A / 1200W / 9,6kg
  5. 32A / 1600W / 10,5kg
  6. 40A / 2000W / 13kg
  7. 46A / 2300W / 14kg
  8. 53A / 2650W / 15kg

i think that main problem now is different firmware version in ESC on one motor because can stopped rotation guud if not using for stop this one esc.

T-Motors support not rely me…

i will change for the same version

I don’t think it’s causing this problem but you should set these parameters:
MOT_BAT_VOLT_MAX : (4.2v x No. Cells)
MOT_BAT_VOLT_MIN : (3.3v x No. Cells)

Also, why don’t you ask the ESC supplier if those ESC’s do indeed have thrust linearization built in?