Drift problem in auto level mode

[3Dfresh]

I have a problem with the auto-level function. My hexacopter is well calibrated. Auto-level and wind is not going well. My hexacopter does not stay horizontal in the air.
He is blown sideways by the wind. He does not stay level. This he drives too fast.

What can I do about it.

Is this the drift problem that you're talking about?

[chris ables]

Autolevel just keeps aircraft upright and doesnt hold it in one spot ! You have to have gps for it to stay in one spot !!

[3Dfresh]

Thats not what i mean. I know that you need GPS to stay on one spot. But my problem
is with windy weather my hexcopter is not stay level. It takes some angle and fly fast
with the wind. Normaly it shoot stay level olso if there is some wind. The drift is to big
with this small angel. If i fly in acro mode and keep it manualy level the drift with the wind
is much smaller if you know what i mean.

I hope somebody can help me.

[cliffkhew]

Do you mean your hexa performs "position hold" in Normal mode better than in Auto Level mode?

[3Dfresh]

I make a movie of the problem so you can see what i mean.

[dramida]

I can confirm that (on two separate copters) level mode is not working as it should. When you fly in LEVEL mode stationary or at low angles, everything is fine BUT,
I'v flown my tricopter and a quad of my buddy with MWC2.0 and both have the same behaviour: When you lean hard for some time, let's say 10 seconds, and gain some momentum, then if yo leave the sticks centered the copter will hardly come back to level, drifting in random direction for a while, even in good calm weather. That's why GPS POS HOLD dosen't work as it should, some time RTH and PosHold banks hard to a side (45 degrees) even it has limit D= 10.
Both copters have BMA180 and ITG3205 on MWC2.0 from 25 March.
After that I compared the flight with an ArduCopter in level mode and the feeling was different as i let go the right stick.

[urkadurka]

I have a quad with BMA180 ITG3200 (the all in one).
With the quad levelled in hovering I switch on the level mode trought the TX, so two different issues
1. If I change the quad's pitch, trought stick on Tx, then I put the stick on the center the level is regained but in a very very fast way (I used the PID values for Level suggested by shickra.
2. The quad at this time is levelled and it seems to work fine but afer some seconds (5 to 20) the quad loses the level, changing autonomously for example the pitch, and starts to accelerate, without correction in order to reach the level again.

Obviously the quad is trimmed both for acro and level mode.

I do another test, I changed via C-Code the BMA180 sensitivity from 8G to 4G, after that I didn't any acc calibration or ACC trimming, surprise when I change the pitch or the roll the level is regained smoothly but the level drift isn't solved.
After this test I checked trought the GUI the ACC values X=-12, Y=-23, Z=600, so I done the ACC Cal and ACC trimming in this way the ACC X,,Y,Z values was 0,0,255 but the I lost the smooth behaviour.

Please Alex could you give us your opinion on that ?

[kos]

1. If I change the quad's pitch, trought stick on Tx, then I put the stick on the center the level is regained but in a very very fast way (I used the PID values for Level suggested by shickra.

the level is regain as fast as the internal spring of your tx get the stick back in center
-> graduate the force you applies as you release the stick

2. The quad at this time is leveled and it seems to work fine but after some seconds (5 to 20) the quad loses the level, changing autonomously for example the pitch, and starts to accelerate, without correction in order to reach the level again.

5 or 20 seconds hand of the tx while hovering is fine , you wont get much more without either :
- an optical flow control
- angled motor to improve static stability (at the expense of decrease maneuverability)

[urkadurka]

So, I think there is a misunderstandig, I tray to explain in a better way.

The drift level issue is : the quad keeps the level for some seconds but it changes autonomously the pitch or the roll with roughly by 10° - 15° without any stick operation, the sticks are in center position (acc calibrated and trimmed).

My expectation is that the level functionality corrects the changed pitch or roll in order to restore the level but this doesn't happen, the quad keeps the changed level.
I guess that this behaviour isn't correct.

Finaly, if there is wind, I expect that the quad drifts in same direction of the wind and also the pitch and roll could be change sligthly

But when I done the tests the air was calm, wind was very low, I didn't do the tests in windy days.

please fell free to ask me more details, I'm wondered of this behaviour, on youtube there are lot on clips that show a good behaviour

[Glambert]

I have 3 multiwiicopter: a big quad, a small quad and a Y6

On the small quad, I cannot use the level mode because I experience somthing similar to what is described here.

When I move fast forward, it tend to lean right, so during a fast forward, I have to compensate and I end up with a significant Ail stick input towards left.  Then when I release ail stick, i have to compensate him trying move to right strongly, releasing again then he finally flies horizontal.  It is just like if his reference to what is level is changing during the flight between hower and FF.    I tried to recalibrate ACC in the GUI a few times, it changes nothing.

Gregory

[chris ables]

Have you tried trimming the acc with transmitter sticks, not in the GUI ? This must be done at least once after installing firmware !!!

[fr3d]

Have you tried trimming the acc with transmitter sticks, not in the GUI ? This must be done at least once after installing firmware !!!

yes, of course, we did

[chris ables]

ok how about commenting trusted acc in firmware ? Turning trusted acc off !!! If you tried this i dont know what else to tell you !! Sorry .

[fr3d]

ok how about commenting trusted acc in firmware ? Turning trusted acc off !!! If you tried this i dont know what else to tell you !! Sorry .

can you fly with your mw 2.0 in stable mode and acro mode without pb ?
"flying" is not hovering because with hovering, without wind, the copter mw 2.0 fly very good.
If you fly like this : http://www.youtube.com/watch?v=9CsZBu88lEM in this video stable mode is used only to re-ajust my sunglasses.... no trouble
if you do hovering no troubles too...just a little drift.

but if you fly a lot (2-3 min) in stable mode and after you trying to do hovering the will see your quad at about 10° tilting with stick centered. it will be impossible for it to reach his plane stable attitude.

I will try as soon as possible to uncomment trusted acc we will see. of course this issue doesn't appear in 1.9...

[Hamburger]

Wrong. at least for me.
I had this drift problem in stable mode with one of my two copters with wmp and bma020 since pre v1.9.

[chris ables]

I fly with just itg3200 gyro's and bma130 acc with no problems in either mode ! And i fly aggresive aerobatics ! And also have flown acc trusted and not trusted with no problems ! It does drift in wind but stays level even at long periods !

[fr3d]

I fly with just itg3200 gyro's and bma130 acc with no problems in either mode ! And i fly aggresive aerobatics ! And also have flown acc trusted and not trusted with no problems ! It does drift in wind but stays level even at long periods !

can you send me your config.h file ? I ve surrely missed something.

[chris ables]

ok fred i sent it but the board i selected was for the quadrino and not the zoom ! I flashed my friends zoom with it last !!

[chris ables]

im also running ronco's output file but dont know if that would help you ! Ive heard diffrent results with it !! it's the 400hz output file and im running skywalker 20 amp esc's with it !! Warthox said it made no difference on his but it smoothed mine out a good bit . Might be my board but i really don't know .

[fr3d]

Wrong. at least for me.
I had this drift problem in stable mode with one of my two copters with wmp and bma020 since pre v1.9.

hum ???
I've never seen this issue in my big quadX but, I apologize, I can't remember if this summer it was 1.8 or 1.9 ,
*what is your i2c speed 400 or 100 khz ?
* on the other, the working one ? what is your hardware ?

[fr3d]

im also running ronco's output file but dont know if that would help you ! Ive heard diffrent results with it !! it's the 400hz output file and im running skywalker 20 amp esc's with it !! Warthox said it made no difference on his but it smoothed mine out a good bit . Might be my board but i really don't know .

many thank I will investigate futher with your files

[Hamburger]

Wrong. at least for me.
I had this drift problem in stable mode with one of my two copters with wmp and bma020 since pre v1.9.

hum ???
I've never seen this issue in my big quadX but, I apologize, I can't remember if this summer it was 1.8 or 1.9 ,
*what is your i2c speed 400 or 100 khz ?
* on the other, the working one ? what is your hardware ?

My notes have the 'drift problem' first mentioned and posted here around 12/2011.
Both copters have same hardware; one has the symptoms but nott the other.
both copters with wmp-clone and bma020.
i2c speed is same 400k for both.
The symptom seems to have lessened with 2.0+ svn releases for me - but that observation could also be due to less windy conditions and I have since switched to re-flashed ESCs. I am looking forward to what explanation and fix you may come up with.
Happy hunting.

[Hamburger]

Update on the drift problem: on the copter with the drift problem I replaced the wmp+bma020 with a freeIMU v0.1 (adxl345+itg3200) and the drift problem seems to be gone.

Drift problem was: copter trimmed and level trimmed, flying in level mode tilted against wind for long time -> then without wind level mode without stick input does not level the copter anymore but keeps a tilt angle of 10 to 20 degrees (my guess).

[Alexinparis]

Hi Hamburger,

Note that in the last dev, the IMU code changed a little. (other ACC-GYRO correlation factor, and a float LFP)
I don't know if you mixed the changes, but it could also explain the improvement.

[Hamburger]

Hi Alex,
you are right. My observation included both changes (sensor replacement and software update) in one set. So it does not provide any usable info, sorry. And I do not want to revert the copter back neither in hardware nor software. Let us hope the symptoms (drift) will not show any more.

[Bledi]

for me with the last dev, the stable mode is just perfect. No drift

[Hamburger]

ok. So the drft problem is solved, case closed.
The thread subject could be appended with '- solved; closed' ?
very good.

[carter]

Hi everyone

I still say its not closed.. I have the same exact problems.. here is my hardware setup:
-Crius multiwii se http://www.goodluckbuy.com/mwc-multiwii ... tml?&sl=en
-1000kv motors and 10x4.5 props (flexable)
-detrum 30A esc
-2200mA 11.1V 20C LiPo
-frame http://www.goodluckbuy.com/kk-mk-multic ... 400mm.html
-16.5" motor to motor , quad-X configuration

I would really like to FIX this problem but I'm just clueless as to what to do from here

now I have read from -many- forum pages that version 1.9 (and lower) have stable "STABLE" mode.Planning to flash this back to 1.9 but the site (crius mwc board) says ITG3205 for gyro but I can't find it in config.h file..

suggestions pls?

thanks


EDIT:
ok wait, there is a new code dev20120522.. I will try this out first..be back soon

[Bledi]

yes you have to try with the last dev version. Before this version I am ok with you there is drift.

[urkadurka]

I Agree, with latest version the autolevel does work well.
I'm going to test deeper in the next week end.

Bye

[carter]

ok, timecheck... im using v2 dev0522

I don't know if this STILL is a drift problem.

just this week, I tried flying my quad in the field, after some seconds of continuous "rolling" maybe more than 6 seconds, it would FIX itself in that position, and crash!! no amount of counter-roll control works.. this has happened twice this week

is this problem common in STABLE mode?

could someone else give me a good answer here?

or multiwii FC -error free- only in tricopter mode?

im just pissed(sorry)..maybe ill just reconfigure this quad into a tri

[shikra]

That ain't right for sure!

If your just hovering around, shouldn't see any drift type issues. Certainly not rolling around intermittently!
Flying in level mode isn't perfect as sensors become fooled.

[Sebbi]

I also noticed this problem and explained it in the r964 thead. When it is completly windstill the auto level mode seems to work for some time as expected. But it will eventually tilt in some direction (trim not perfect, small wind ... i don't know) and accelerate in that direction if the pilot does not compensate. However, this acceleration should be seen by the algorithm ... but it's not.

The problem gets worse when flying in moderate to strong winds. The copter seems to have a hard time to compensate and stay level and sometimes tilts in the wrong direction but considerably stronger than with no wind :/

However, this could all be just a PID problem. I don't know yet ...

[carter]

hi shikra

if only hovering, I don't see any problems.. well, there is *very* small drift(even indoors a few feet up) , but that's just normal.unbalanced motors, balanced props,some vibes..

the only worst part to this is when your just way UP there, coupled with less accurate vision, and the copter starts speeding up to the drift direction. and there you have it, forced landing(Crash) and some broken props. sometimes those counter-drift moves does work, sometimes NOT. you can tell its not "countering" the drift because no matter what you do, it wont budge. this even happened at about 15ft above ground. and I thought it was just *maybe* and altitude problem(note:my baro IS dead so im not using it)or motor RPM thats causing havoc to the FC (wild guess:at certain RPM, it "ruins" sensors(?), but FC and rx has its own reg(uBEC from LDO)...luckily I "panicked" , quickly throttled down, throttled UP again, and countered the drift, and it DID save it by colliding-crashing at a nearby bush, saving my cheap props.

quick guess again, maybe quickly throttling down and up again, it "resets" something, and the autolevel works again?

regarding PID, settings, mine is 4.0 0.035 35 and rcrate 0.90 (w/c I think is "loosy")... upping the P to 7 and above makes the copter more stable(?).. but it feels like a bit harder to control(maybe ok for FPV)

well, thats all for now

[Chaosdestro0]

Just a quick update, I upgraded my sensors to the GY_80 IMU, I was previously using WMP+ AND NK. I get much less drift now, at least it's usable now on level mode .

[carter]

ok.. got another update to make

using the autolevel ONLY, is OK

but adding magnetometer ruins it