Gimbal drift (stable mode)

[UndCon]

I attached a gimal to my Quadcopter for the 1st time and did some testings.

When I fly in acro-mode it works great with acceptable drift
When I fly in stable-mode the camera drifts away and so does the acro level mode

Is this sensor related? if so it must be the original NK... I don't fly so much in stable mode so I didn't notice until today when It was very obvious

(bot acc and gyro were calibrated)

If the camera is tilted I can land and wait for camera to regain preset mode (and I assume so does level-mode) and when I take off the camera is OK again - for a while...then its all bad again...

Stable mode must work longer than ~1 minute right?

I have no spare NK's or similar sensors so I cannot rule that one out
//UndCon

[copterrichie]

How do you define Drift? and what ACC are you using?

[UndCon]

I calibrate Acro gyro and Stable ACC so there is no difference between them 2 when I take off.

Camera is level as supposed to.

I flip stable -mode ~1 meter above ground and the quad is very stable - no need to add trim.

Then I fly around in stable mode for a while and suddenly I have to compensate more and more and more until I cannot compensate more with my stick.
I have to turn off stable mode and take manual control to save the situation.

If I fly close in front of me I see the camera is tilted towards 1 side as much as ~20 degrees and so is Stable mode - that is why I have to compensate with my stick input.

I have a recording from last night - dark but clearly shows the issue. I can upload it after work.


Now if I only had the BT module connected so I could record in GUI live from flying...or add SD card logging.
It would be very interesting to see how much drift there really is and most important...why
//UndCon

[copterrichie]

Well, I am not sure which Accelerometer you are using but I had a similar problem with the BMA020.

[UndCon]

I use Stock Nintendo sensors...and 1.8

Will change to 1.8_2 and see if it hels fight the drift issue

[KeesvR]

I should try dev20111017 for me the drift is completely gone in this version.

[UndCon]

1.8_2 did no change at all...

testing dev20111017 indicates something interesting...

When I hit stable mode i get lots of noise in GUI graphs


http://www.youtube.com/watch?v=aPs0dUqtnTM

//UndCon

[copterrichie]

1.8_2 did no change at all...

testing dev20111017 indicates something interesting...

When I hit stable mode i get lots of noice in GUI graphs


http://www.youtube.com/watch?v=aPs0dUqtnTM

//UndCon

I have seen this behavior before but with a BMA020, in my situation, the Accelerometer was bad.

http://www.youtube.com/watch?v=UIkPb-qTfT0

[UndCon]

Exactly my thoughts...Ill buy a replacement tomorrow

Thanks for your assistance
//UndCon

[cardboard]

i have those same gliches in my gui, not as frequently though.ive had a few nasty crashes before it started happening. might need to buy a new one as well

[UndCon]

My Cycle time is very high...broken acc?

[PatrikE]

I Have around 6000 with Wmp & NB.

[UndCon]

Ill replace it today and see if the crazy lines disappears.

//UndCon

[Noctaro]

inspired from your captures i did test my acc over time.
what would be critical values that may cause a drift?
What i did:
- Pluged in the copter
- 10 Minutes later
- Calibrate ACC

To simulate my inflight time i left it for 15 minutes, but did not move.
ACC Z Values did climb up from 507 to 511
Pitch from -1 to +1
Roll from 5 to 4

cycle time is between 4000 and 5000

reconnecting 10 minutes after cutoff
i got accZ 508 again going up to 511
roll and pitch have constant 0

Should i matter about this?

[UndCon]

Compared to mine that is nothing - I would say totally acceptable

//UndCon

[Noctaro]

ok, thanks
greetz noc

[UndCon]

I tested yesterday with a new NK but the same patterns showed up. I messed with my gear for a while and then my FTDI adapter gave up so I ordered 3 new ones right away.

I have a 2nd Parisboard (just like this) so I can compare graphs with them 2 - and I have a flyduino shield with WM+ only where I can add the removed NK for more tests.

[UndCon]

More confusing...I got my new FTDI adapters and fired up 3 boards with the same software and I see the same ticks on all of them when boards are stable on desk.

I have not tested the boards for drift as I do not have RX for all of them.

Ill give it a try later in the dark

//UndCon

[UndCon]

More tests shows same glitches on every board I have - so I'm gonna test other things like running GUI on a different computer and see if it is still there.

//UndCon

[UndCon]

I have found a reason that causes the wicked lines.

When I have the RX connected (Genuine Spektrum AR6100) I get noisy lines
When I disconnect it the lines gets level

When I reconnect the RX same noise appears.
This is tested on my Jussi 1.3 shield...

[UndCon]

I have solved parts of my problems

I recalibrated all ESC's endpoints - I had 2 crashes yesterday due to 1 motor was very sensitive on throttle - all in
Now I can fly in acro mode for a long time.

I cleared EEPROM and downloaded all code again.
I calibrate both sensors on level ground.

I fly acro and when I hit stable mode the Quad is way off... (I thought that is what the calibration is for - to reset level setting)

Now I have 1 motor that has no throttle at all and 1 motor that has all throttle in stable mode (verified by meters in GUI)

I suppose I will follow copterrichies guide to calibrate stable mode...

http://www.youtube.com/watch?v=knagzCgXGEg

[UndCon]

My issues above seems to be solved with the release of v1.9

I want to test ASAP!