strange behavior in level mode

This forum is dedicated to software development related to MultiWii.
It is not the right place to submit a setup problem.
Software download
Post Reply
doduveloce
Posts: 5
Joined: Fri May 13, 2011 9:43 pm

strange behavior in level mode

Post by doduveloce »

Hi all,

I'm experimenting the last dev (MultiWii_dev20110607), I'm using a Paris aerial V2.0 board and an allinone LLC 5v for sensors.
I'm in quad + config.

So in normal flight everything is correct and my quad flight very well. But in level mode I'm expericing some trouble.
I'm starting flight in "normal mode" then switch to level (wihtout baro and mag), my quad flight well since I give only some small order to stick.
If you try to apply a large amount of roll and release stick quad stay level at approx 20 degres.
So I've done another test, if you keep the quad in your hand start motor and try to apply an angle pertubation it works well till you reach aprox 20 degree.
Before 20 degres quads try to recover a level flight after 20 degres quad stop to level back the quad and use 20 degres has new level till you apply a 20 degres deviation in opposite direction.
So you have to flight carrefully in level mode, and keep a finger on the switch just in case.
I've try to unerstood what's wrong, but without success.

Did someone has same behavior?

Olivier

Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Re: strange behavior in level mode

Post by Alexinparis »

Hi,

I discovered there is an overflow variable error here:

Code: Select all

      PTerm      = errorAngle*P8[PIDLEVEL]/100 ;                                  //680*10 = 6800: 16 bits is ok here


For the moment, this should be changed to:

Code: Select all

      PTerm      = errorAngle*(P8[PIDLEVEL]/10)/10 ;                                  //680*10 = 6800: 16 bits is ok here

mahowik
Posts: 332
Joined: Sun Apr 10, 2011 6:26 pm

Re: strange behavior in level mode

Post by mahowik »

Yes, I agree in acro mode copter very stable (in diff with 1.7version), BUT in stable mode I got unpredictable flip and damage my copterX.
Probably this issue was the reason of damage?

http://www.youtube.com/watch?v=Rie47p45 ... r_embedded

Alexinparis
Posts: 1630
Joined: Wed Jan 19, 2011 9:07 pm

Re: strange behavior in level mode

Post by Alexinparis »

Sorry about his. this might be tied to this problem.

In the next dev, no violent flip could happen.
I found a way to limit in all occasion the attitude correction sent to the motor without affecting the stability = no more huge wobble ;)

mahowik
Posts: 332
Joined: Sun Apr 10, 2011 6:26 pm

Re: strange behavior in level mode

Post by mahowik »

thanks :)

doduveloce
Posts: 5
Joined: Fri May 13, 2011 9:43 pm

Re: strange behavior in level mode

Post by doduveloce »

Thanks Alex,

I will test tommorow. To late now to make noise at home ;)

I'm not sure about Mahowik issue, I've take a look on the video and in my case I did not get wobble or unattended flip.
It was just like an "offset" on the angle.

I will give a feed back here.

Olivier

doduveloce
Posts: 5
Joined: Fri May 13, 2011 9:43 pm

Re: strange behavior in level mode

Post by doduveloce »

Hi all,

I've done some tests this morning, so fix provide by Alex works for me, many thanks Alex.
I've not already done flight test outside :oops: (It's raining a lot here) :| , but multi reaction seems ok at bench.

I've test by holding multi and apply a roll or pitch deviation by hand til 90°, in this case motor still continue to try to come back to level position.
Same if you apply a large stick deviation, when you release stick it comes back at level position.
It seems also if you keep stick deviation multi stabilize to a flight angle (instead of doing a flip).

I will validate multi comportment in flight, when weather will be ok.
I'll keep you inform.

Olivier

Centurian
Posts: 44
Joined: Sat Jan 22, 2011 10:55 am

Re: strange behavior in level mode

Post by Centurian »

I have a Y6 that I just got flying(seeduino mega, WMP & NK). Flew a few packs thru it with the 6-7dev w/o issue. This morning as I was heading out the door, I uploaded alex's change above. Trim in autolevel changed significantly(did not recalibrate the NK) but was flyable. Trimmed it with the full throttle & full pitch or roll inputs. After about the 5th trim change, when I switched into autolevel it immediatly went into a horrible roll oscillation that did not stop when I turned autolevel off. Ended up with a minor crash and just wanted to share my experience. I havn't checked the gui yet but I'm thinking the acc calibration values may be rather off scale due to the significant trim changes.

I was having similar issues on another machine in an earlier dev version with flips in autolevel.

Anyway, just a heads up if this means anything important.

Stalk
Posts: 49
Joined: Tue Apr 05, 2011 12:39 pm

Re: strange behavior in level mode

Post by Stalk »

Alexinparis wrote:Hi,

I discovered there is an overflow variable error here:

Code: Select all

      PTerm      = errorAngle*P8[PIDLEVEL]/100 ;                                  //680*10 = 6800: 16 bits is ok here


For the moment, this should be changed to:

Code: Select all

      PTerm      = errorAngle*(P8[PIDLEVEL]/10)/10 ;                                  //680*10 = 6800: 16 bits is ok here


Something similar happens to me with version 1_7. It could be the this problem?

Post Reply