Oops this make me feel , i have wasted money on bad frame. 
Yes, this is a beautiful design but with a flaw, I was thinking of reinforcing the structure with Carbon Fibre plates one day.
Yes same here, somehow have to make it work .
Well, let’s take a look at a flight log then. But, as @ppoirier says compliant frames are bad news.
1 Like
Here is one of the flight log, which i flew today morning. For your ref , pls check the behavior when it flies from WP3 to WP4.
Also when it halts at last waypoint it become unstable so have to land in Alt hold mode.
Flight Log : https://drive.google.com/file/d/1W_sLGxVSvQXFLTcFp9-PlLfEpJO--Xfw/view?usp=sharing
The 1st thing to see here is general tuning is poor. I assume you ran Auto Tune but it did not produce good results. When you have a large discrepancy between these pitch and roll parameters it didn’t tune well:
ATC_ACCEL_P_MAX,77049.09
ATC_ACCEL_R_MAX,31590.45
ATC_ANG_PIT_P,7.273218
ATC_ANG_RLL_P,1.998792
There also seems to be many parameters at default and the Dynamic Notch Filter is not configured.
I would set the Roll PID’s equal to Pitch. Run the Initial Parameter Setup in Mission Planners Mandatory Hardware screen. Set this parameter to (1) INS_LOG_BAT_MASK to log data for the notch filter config. Then make a short hover flight in AltHold for 1-2 minutes. Post that log if you like and read the Wiki entry for configuring the Notch filter.
No amount of filtering will fix a badly flexing frame but these are initial steps that should have been made long before attempting an Auto flight.
Thanks @dkemxr for your suggestions , Will make the changes as suggested and logs after test flights too.
Thanks @ppoirier for a reference parameters.
@dkemxr , Have tested today morning with the following changes
-
INS_LOG_BAT_MASK set to one
-
Changed Roll PID’s equal to Pitch
Flight Log : https://cutt.ly/DJmCnu1 -
Also i set it to default PID’s and here is the flight log
https://cutt.ly/uJmCJtn
Next i am going through Wiki to set my Notch filter configuration. Hope it may help others as well with the same frame .
Also I have checked thoroughly quad arms are not flexing on my frame.
Looking forward to your valuable inputs @dkemxr from flight logs .
Thanks to align to the right process.
You don’t want default PID’s, you want to use the Initial Parameter Setup utility in Mission Planner under Setup>Mandatory Hardware. After doing that set INS_ACCEL_FILTER to 10.
You did collect useful data to set the Notch filter. With a Cube Orange and no ESC RPM I would use FFT. You can try these:
INS_HNTCH_ATT,40
INS_HNTCH_BW,40
INS_HNTCH_ENABLE,1
INS_HNTCH_FREQ,80
INS_HNTCH_HMNCS,3
INS_HNTCH_MODE,4
INS_HNTCH_OPTS,6
INS_HNTCH_REF,1
FFT_MAXHZ,270
FFT_MINHZ,40
Then set INS_LOG_BAT_OPT to 2 to log post filter data and make another Hover flight to see how well the filter is working.
2 Likes
Hi @dkemxr Thanks for inputs, i changed the parameters as you advised.
Here is the flight log :
https://drive.google.com/file/d/10ChfXzTQtQT7ae1gbyCbvcLlGR2sxXj2/view?usp=sharing
Roll Pitch response is bit sluggish.
Odd that it didn’t log useful FFT parameters (FTN messages). Perhaps set INS_HNTCH_OPTS back to 0.
Yes, no surprise as you are at pre-tune level still.
Now INS_HNTCH_OPTS back to 0.
Flight Log: https://drive.google.com/file/d/1otXWt6LUTtNHvWuxVZVWy6FA1x2ZQtC1/view?usp=sharing
OK, pre-tune is looking good! Set INS_ACCEL_FILTER to 10Hz, set AUTOTUNE_AGGR back to default (0.1) and give Auto Tune another try when there is no wind.
1 Like
Thats good, I set the
NS_ACCEL_FILTER = 10Hz
AUTOTUNE_AGGR = 0.1
I tried only roll axis autotune, here is the log.
https://drive.google.com/file/d/101Z6W9PEIRAQlhMohDrU8Lcbcmft8OCs/view?usp=sharing
Weather was very calm, though its too hot here.
I will try pitch and yaw autotunes tomorrow .
So far so good. You cant really analyze a Tune from the Auto Tune log itself but the Roll PID’s look to be going in the right direction. The Notch Filter is working really well.
Roll Autotune :
Winds were low around 4km/hr
Flight Log: https://drive.google.com/file/d/1U6-ifR1d8sfp9by7s69ONC2WdDLgT4A_/view?usp=sharing
Pitch I suppose you meant as both Pitch and Roll are Auto Tuned in this log.
Set these now:
PSC_ACCZ_I to 2X MOT_THST_HOVER
PSC_ACCZ_P to MOT_THST_HOVER
ATC_THR_MIX_MAN, .5
Yes it was after the Pitch and Roll Autotune.
Also tried a short mission in Auto and found oscillation in Roll Axis :
Flight Log : https://drive.google.com/file/d/147B84tpG3FEBd9i1v6mT73WT1x1RtXKQ/view?usp=sharing
&
Since pitch was very stable so i experimented with same gains on roll and here is the flight log:
https://drive.google.com/file/d/1R70PddhvAjhYiGaJ72xDFTkMggOXiskE/view?usp=sharing
Seems stable from ATT DesRoll/Roll and same on pitch ,
But difference between RATE.ROut and POut is on the edge , more than 0.1.
Kindly share your opinion .
I will try with changes you suggested and post the log soon .
Hello gentlemen, coincidentally I am having the same exact problem. The exact frame and flight controller ( tdrone m690 and cube orange but px4 firmware). My issue is also the same. There is a oscillations on the roll and pitch axis. Even when testing indoor hovering it twitches on these two axis. What was your final result after tuning it? Did you solved your problem? I would be pleased to hear from you