forum.thequadshot.com
http://thequadshot.com/forum/

Expresso: drifting in mode 1
http://thequadshot.com/forum/viewtopic.php?f=5&t=834
Page 1 of 2

Author:  david.erzen [ Sun Sep 02, 2012 4:44 pm ]
Post subject:  Expresso: drifting in mode 1

Hey guys

Does anybody have a problem with drifting in hover mode? Soon as the QS takes off, it starts do drift to the right and I can't correct this movement. Using trimmer doesn't help. :(
I did calibrat the accelerometer part of the IMU.
Regards,

David

Author:  psinha [ Tue Sep 04, 2012 5:33 am ]
Post subject:  Re: Expresso: drifting in mode 1

Is this a yaw drift or a roll drift?
Also, could you please tell me what values you used for the accel calibration?
Thanks.

Author:  david.erzen [ Wed Sep 05, 2012 3:52 pm ]
Post subject:  Re: Expresso: drifting in mode 1

Hey,

the drift was in roll direction and the values i have are

<define name="ACCEL_X_NEUTRAL" value="33"/>
<define name="ACCEL_Y_NEUTRAL" value="33"/>
<define name="ACCEL_Z_NEUTRAL" value="28"/>
<define name="ACCEL_X_SENS" value="4.90893710325" integer="16"/>
<define name="ACCEL_Y_SENS" value="4.89174059236" integer="16"/>
<define name="ACCEL_Z_SENS" value="4.85787113112" integer="16"/>

I comapred them to IClausius's values and are similar.

Regards,
David

Author:  psinha [ Wed Sep 05, 2012 7:51 pm ]
Post subject:  Re: Expresso: drifting in mode 1

Try 0's for the ACCEL_X/Y/Z_NEUTRAL values. That is what we use and it seems to work quite well.

Author:  david.erzen [ Mon Sep 10, 2012 4:24 pm ]
Post subject:  Re: Expresso: drifting in mode 1

Tried 0's.
Motors didn't start spinning at (almost) the same time.
I don't want to go to the field because I am sure it will crash.
Regards,

David

Author:  david.erzen [ Sat Oct 20, 2012 2:56 pm ]
Post subject:  Re: Expresso: drifting in mode 1

I have done countless sensor calibrations untill now and there is always a drift in hocering mode, whether in roll direction or pich direction.
Any suggestions?
regards,
David

Author:  david.erzen [ Sun Oct 21, 2012 4:20 pm ]
Post subject:  Re: Expresso: drifting in mode 1

Today I noticed also yaw drifting as well.
Anybody has the same problems?
Regards,

David

Author:  psinha [ Mon Oct 22, 2012 3:12 am ]
Post subject:  Re: Expresso: drifting in mode 1

David,
Please ensure the following:
1) Your elevons are centered correctly. To do this, power up the quadshot but disconnect the battery before the delay finishes. At this point, the servos will be centered. Loosen the screw on the control horns and ensure that the elevons are in the centered position before tightening them up again. If the elevons are not centered, you will get drift in (helicopter) yaw and pitch.

2) Your elevons are moving in the correct direction. If your elevons are plugged into the Lisa/M in the wrong order (ie, if the left one is plugged into the right output and vice-versa) then you will get a downward pitch behavior when transitioning into mode 2. If the elevons are not movingin the correct direction, you will get drift and instabilities in yaw and pitch.

3) Your motors are not crooked. This is especially important if you have had a crash. If the motors are crooked, the controller will not be able to correct or stabilize effectively. This can lead to drifts in roll, pitch and yaw. Also make sure all your motors appear to be spinning up the same amount.

4) You mentioned in another post that you had calibrated your sensors multiple times. Please try it with the default sensor settings available on github, just to make sure the calibration does not have anything to do with it. Another good check is to hold the quadshot in forward flight orientation and check the python attitude visualizer to see if it indicates the attitude correctly.

5) Finally, note that your quadshot will not fight the wind. If there is wind, you will need to correct for it, although I am assuming you are reporting drift without any wind?

Author:  david.erzen [ Mon Oct 22, 2012 7:09 pm ]
Post subject:  Re: Expresso: drifting in mode 1

Hey Pranay,

Ad1) The elevons are centered.

Ad2) Today I tried folowing test: i switched on the QS and waited for the sensor self aligning. Then I held the QS in my hand and I raised the thorttle. Then I switched to mode 2 and held the QS horizontally. I rolled it to the right, the right elevon lowered and the left eleron elevated. When rolling QS to the left, the right elevon raised and the left lowered. When I put the QS at negative angle of attack, the motors and elevons seem to try to correct the QS pitch but the effect at too much positive angles of attack seems bigger then at the same negative angles of attack.

Ad3) The motors move nicely. When checked for axial movement, three of them had around 0.5mm of free movement, the fourth had about 0.1mm. Nevertheless the fourth did not seem to turn differently then others.

Ad4) Did check the python attitude visualizer. Seems ok to me. The estimate QS was in the same position as the real QS. From time to time the reference QS would drift in yaw direction and therefore the real QS would like to move (two motors slowly stop and the other two slowly accelerate) but as I was holding it in my hand, it didn't move. :) I guess this drift has something to do with qyro drift?
The calibration parameters are:

<define name="ACCEL_X_NEUTRAL" value="35"/>
<define name="ACCEL_Y_NEUTRAL" value="34"/>
<define name="ACCEL_Z_NEUTRAL" value="34"/>
<define name="ACCEL_X_SENS" value="4.90723256941" integer="16"/>
<define name="ACCEL_Y_SENS" value="4.89101204714" integer="16"/>
<define name="ACCEL_Z_SENS" value="4.83638039912" integer="16"/>

They seem ok too. i will try with default values nevertheless.

Ad5) no wind present. i really don't want to push my luck flying the QS in wind. Not yet.

Any other idea?
Thnx,

David

p.s. I would really like to increse mode1 angle of attack. If nothing else I would see wether the numbers have a correct sign.



psinha wrote:
David,
Please ensure the following:
1) Your elevons are centered correctly. To do this, power up the quadshot but disconnect the battery before the delay finishes. At this point, the servos will be centered. Loosen the screw on the control horns and ensure that the elevons are in the centered position before tightening them up again. If the elevons are not centered, you will get drift in (helicopter) yaw and pitch.

2) Your elevons are moving in the correct direction. If your elevons are plugged into the Lisa/M in the wrong order (ie, if the left one is plugged into the right output and vice-versa) then you will get a downward pitch behavior when transitioning into mode 2. If the elevons are not movingin the correct direction, you will get drift and instabilities in yaw and pitch.

3) Your motors are not crooked. This is especially important if you have had a crash. If the motors are crooked, the controller will not be able to correct or stabilize effectively. This can lead to drifts in roll, pitch and yaw. Also make sure all your motors appear to be spinning up the same amount.

4) You mentioned in another post that you had calibrated your sensors multiple times. Please try it with the default sensor settings available on github, just to make sure the calibration does not have anything to do with it. Another good check is to hold the quadshot in forward flight orientation and check the python attitude visualizer to see if it indicates the attitude correctly.

5) Finally, note that your quadshot will not fight the wind. If there is wind, you will need to correct for it, although I am assuming you are reporting drift without any wind?

Author:  psinha [ Mon Oct 22, 2012 8:59 pm ]
Post subject:  Re: Expresso: drifting in mode 1

Hey,
Thanks for the detailed reply!
Seems like both the hardware and software are working correctly, which is weird if the a/c is still drifting.
You can definitely try to increase Mode 2 pitch angle.
To do this, please go to paparazzi/conf/airframes/esden/toytronics/setpoint/QS4_setpoint.xml
Look for the line <define name="ABSOLUTE_FORWARD_PITCH_TRIM_DEG" value="8.0"/>
Change the value to the angle you want.
Note, this is not actual angle of attack, just the pitch angle.
Hope that helps.

Page 1 of 2 All times are UTC
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/