Copter 3.5.0-RC6 autotune problem

So I updated from RC4 to RC6. In RC4 auto tune seemed to work ok.

In RC6 the auto tune process ran for about 35 mins before I got some sounds from the drone. (I assume this means its finished?)

I then switched out of autotune flight mode and back again. At this point the drone was flying amazingly well so I figured that the auto tune was a great success.

I landed and disarmed with autotune engaged to save tune settings.

I then re-armed and took off. The craft instantly got into uncontrollable oscillations suggesting the PIDs were way way out. I cut power and ā€˜crash landedā€™ to prevent the oscillations getting any worse.

I reset the gains back to their pre-autotune values and the issue went away. To check the issue I then autotuned only one axis and again I got a good tune until I saved settings and tried to fly with themā€¦ Back to crazy oscillations.

Anyone else having similar issues?

To use the new gains the vehicle needs to be landed, disarmed and then re-armed. So when it was flying amazingly well it would have been on the original PIDs. Maybe just stick with those if theyā€™re ok.

There are a number of reasons why the autotune can fail including:

  • things on the vehicle moving during flight (i.e. gimbals)
  • vibration dampening foam on the vehicle is too soft
  • wind is too strong

Iā€™m sure youā€™ve seen the autotune wiki page but just in case: http://ardupilot.org/copter/docs/autotune.html

Thereā€™s some advice there on trying different levels for the AUTOTUNE_AGGR parameter.

Beyond that, Iā€™d say attach a .bin log file and perhaps Leonard can have a look when he gets a moment.

1 Like

Hi Duncan,

It is unusual to get oscillations after autotune. Most issues result in a very docile tune when there is a problem. Post a log of the autotune flight and I will take a look at what happened.

1 Like

Hi @Leonardthall

Apologies for the slow reply. I have been away working. Your help would be enormously appreciated.

Autotune worked ok on RC5 and gave ā€œokā€ results but was by no means perfect hence me retrying it with rc6.

I would really appreciate your input as to what is going on. I do wonder if the frame itself is too flexible and auto tune isnā€™t working properly.

I dont seem to be able to share the log files on here, they just seem to timeout on the uploadā€¦

I can share them from my dropbox if that is ok?
1st flight is the autotune flight https://www.dropbox.com/s/g98ufzl8d45wsna/2017-05-23%2016-41-11.log?dl=0
2nd flight is me testing the auto tune with bad oscillations and heavy landing. https://www.dropbox.com/s/wv6ilgtgpnelqpy/2017-05-23%2017-14-57.log?dl=0

Aircraft is a quad with 17" blades running PH2.1.

Hi Duncan,

Are you using esc breaking? This makes a big difference on large diameter propellers.

Are you tuning the copter with a dummy weight replacing the payload or with the payload attached?

I see what you mean. This is a very low frequency oscillation. There are a couple of things you should change. With a high spin when armed you need to drop the expo value down a little.
Try setting
MOT_THST_EXPO to 0.5
INS_ACCEL_FILTER to 10

You do have a significant 90 Hz wobble in your response that is interesting. You may like trying a tune with
INS_GYRO_FILTER set to 10
ATC_RAT_PIT_FILT to 5
ATC_RAT_RLL_FILT to 5

You donā€™t need to tune your Yaw any more so turn that off.

Chat soon

Hi Leonard,
You make an interesting point about ESC breaking. I actually donā€™t know. The ESCs are an obscure Tarot part. I assume ESC breaking would be a good thing?

Yes I am running a dummy load on the front of the AC. I am trying to get long duration flights so the weight is very light compared to the power available which may be an issue in its self!

I will try those settings and re-run an autotune. Its currently blowing 24mph winds so annoyingly not today!

Am I best to update to RC7 before I try again?

Appreciate your input, its a nice looking machine but would be even better if she flew nicely too

Yes, I understand your frustration.

I have done commercial tunes on a number of very lightly loaded aircraft and without ESC breaking it is one of the most challenging tunes to do.

If you find your first tune takes both D terms down to 0.001 then you can try increasing your autotune aggressiveness up to 0.15.

You can also do a manual tune on your D term and set that as your minimum D term and let autotune set the rate p, I and stab P.

If you donā€™t get any luck with these approachs I have another suggestion but it is a little dangerous so lets see how you go first.

Brilliant thank you.

Looks like ESCs with breaking will be required for V2 of this machine!

The winds look like they should drop on Tuesday so I will get out into the field then and try this out!

Will report back.
Thanks again :slight_smile:

Leonard: Itā€™s better to use esc breaking on large props for flying too? I have (perhaps mistake) the idea, readed in some forum that itĀ“s better that with big props no have esc break so the copter can autorotate if something fails? or something like that?
My only one experience crashing over 30 mts with 17*5.5 props was flipping; any autorotation Uuu.
Impressive Pos hold autotune, many thankĀ“s for that function

Hi Cala,

Yes, ESC breaking is definatly better for control of large multirotors. There can be a loss of flight time but I have not noticed it when the tune is done well. Extreem duration record breaking aircraft that will only ever hover in the calmest of conditions are the only case where I would say to leave breaking turned off.

Leonard,

This is interesting - could you share why breaking would help improve performance?

I am assuming it is because with large propellers (i fly 18x5.5 ones) there is significant inertia that prevents the props from slowing down fast in a case where the FC decides that a given prop needs to do so to stabilize etc.

Right?

Christian

Yes, when the propeller needs to go faster there is the full power of the motor to speed it up. When it needs to slow again there is only the drag of a lightly loaded and efficient propeller. While the increase in speed is very quick the decrease is very slow and prevents a good tune.

1 Like

ThankĀ“s Leonard, always something new to learn here :slight_smile: , I use esc break in the racer and works nice, IĀ“m going to recive new esc for a bigger one next week, I give a try, y fly in windy conditions.

Hey Leonard, interesting point about escā€™s and braking. What do you think of the Castle DMR 30/40ā€™s? Iā€™ve got them on a 2.9kg quad and they work pretty good. Iā€™m in the midst of swapping old motors on my 6kg hex as theyā€™re getting pretty old. Maybe so braking escā€™s would be a good addition. Theses escā€™s seems so small though lol! Maybe something youā€™d put on a quad racer : )

They look good but I have not used them. But you are right, they might work well on the F500ā€¦

I would trust Castle to spec their product accurately. Just keep in mind that those specs assume the ESC is in 40 mph airflow (under propeller wash). They are one of the most responsive companies I have dealt with!!

Just to get some things right here for meā€¦ I have some problems with a big 960 quad and the vibrations from this frame. not really sure how to deal with it because i tried some different things. Now i wanā€™t to have a look into the chance of esc braking the props because i fly some heavy 18ā€™ā€™ props which carry a lot of momentum.

Currentyl iā€™m using some cheap dys 40a escs which i canā€™t program for braking.
Could someone point be in the right direction of some better and available escs with the braking option?