PID tuning for 6 m spann 100 kg Drone

Dear All, we are at the end of the manual tuning of a 6 m spann drone for spraying it can carry 40 liter. The tests in Alt mode are OK in all 3 conditions empty half and full (MTOW 100kg) However in the loiter mode with full payload, the issue is that a stick roll movement results in sluggish reaction and a not nice oscilation of 2 or 3 times before it levels again, in half full and empty this is not a problem…

VID-20200416-WA0003 wr_Moment

These are the settings and logs are available as well as video’s but they can not be uploaded as too long. So the question is how can we get the drone.

Question, before we can move into waypoint we need to get the PID in loiter better! (or even in altmode. It can be seen that the PID settings are quite high already. The craft flies over the short axis, basically a flying sprayboom. some idea: 6 m width with 16 motors calibrated ESC’s PX2 cube. Lenght of the craft 1.25 meter.

So if you have ideas on how to get the loiter behaviour better let us know? We have well followed the guide of PID tuning even correcting for payload. If you like we can send you all the data by wetransfer Best, Winfried

https://ardupilot.org/copter/docs/tuning-process-instructions.html
But if you have to ask it means that you are not familiar with arducopter. PLEASE start with a reasonable sized aircraft or employ somebody who know what to do. I would rather not support a tuning of a 100Kg MTOW craft (which is obviously commercial) via forum.

Sorry, I have followed those instructions already. And already have made many small drones. But for non regular shapes ( length width ratio), we have seen this roll issue problem, while it is not there in Altmode.

Do you use a custom motor mixer ?

Dear Andras,

For this one no, it is an octoquad config with 2 motors per arm. But have done that before with other V shapes, and that went OK! It was given by a very good guy who gave all sorts of configs, and from that I learned how to make the V shape. The uploading of the new motor mix in this library, was OK, but for this one I stopped that, as unpractical when you move to newer versions. So I thought we could do the octocopter and do different Roll and Pitch PID settings and other parameters as in pretuning guide. When we were experimenting I found the yaw also slow, so the tip rotors were tilted to give more moment and that works fine.
Best, Winfried

Did you try autotune with no payload and low aggressiveness ?

No, I feel uncomfortable in autotune with such size craft… However have logs of the flights, as well as videos. That should allow an expert to analyse…

Well, then do not hold it back, post the .bin files…

Herewith

tried to upload but does not…

If it does not work send me an email: winfriedrijssenbeek at gmail.com

Upload the log to a file hosting service like google drive, dropbox or onedrive, for example, then post us the link to it

This should work? Best and have a look? You see that the craft full load is oscillating in Loiter over the roll axis,

https://drive.google.com/drive/folders/10MtoqBx935sNThiMCXoFidy2hN6Y16CQ?usp=sharing

You’ll need to set permission so anyone can download it

Sorry, but then it asks for an email account… so I do not have that…

It’s now public? Hope that should do it, Best Winfried

There’s no battery voltage logging - I’d set that up first.
This will help with the MOT_BAT_VOLT_* settings having their desired effect.

A quick look at your PIDs and other tuning params, I suspect you’ll have to continue with manual tuning as per the guide. Unless you’re prepared to try Autotunes one axis at a time.
Try lowering the pitch and roll PIDs in 10% steps to see if you can get that pitch/roll and desired-pitch/roll under control.
https://ardupilot.org/copter/docs/tuning-process-instructions.html#manual-tuning-of-roll-and-pitch

Also check the definition of these params and see if they help.
ATC_INPUT_TC -> 0.2 to 0.3, something like that for more controlled RC input
ATC_ANG_LIM_TC

Thanks will do, Do you see loiter behavior is quite different from Alt mode. Somehow I feel that the oscillations that are there (due to my stick movement) in roll in loiter but not in altmode (there it corrects OK) either need to be adressed in Alt mode PID’s and that Loiter PID tuning is last resort? What’s you r best guess?

Cheers Winfried

You need to make PID settings “in flight tuning” using the RC-control. Assign a pitch-roll stab P to the RC-6 channel, and select the desired behavior with the twist (rotary resistor:) See more - https://ardupilot.org/copter/docs/ac_rollpitchtuning.html#ac-rollpitchtuning

Thanks Juri, and yes that would be a good way of doing it, but I need full concentration on the craft to overcome the oscillations that appear in loiter mode… So yes may be I should be doing that!

Thanks, Winfried

Most likely your device is unique, no one in the world has done this yet, so each parameter will have to be searched for manually for a long time and painstakingly. Great luck for you!

Dear Juri, correct it seems that we have to do it all, as there is no people with similar experience. The beauty is that we already made a flight of 200 meter in altmode using 1/2 payload. In loiter we got it right for 1/2 load and empty. But in the full load it is horifying: it starts auto oscillation when too high a P dist 1 or 1.5 PID1, .5. When doing it at P dist .5 it is sluggish and does not nicely stick to its position Sluggish response! So I think that there is some clue: it seems that the power variation needed is just not there? We use 18650 tech with good C value but ca C15 to C20, but this is not like lipo… Today tested the lipo cells we prepared. And although not tested yet in loiter full load, we already tested these in altmode full load and the craft did respond quite neat! So what I want to do as a next step is trying in loiter full load with P dist 0.75 and PID 1, 0.5 and 0. using the lipo’s. On the flight we can see behavior, but can we see on the logs motor power Amps or voltage drop? or motor response to change of roll axis? Are you familiar with this reading of logs.

So first is visual behavior with lipo cells and then compare logs with the ordinary 18650 cells we use normally.

cheers Winfried