[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Paparazzi-devel] CHIMU Drifting Issues

From: Christophe De Wagter
Subject: Re: [Paparazzi-devel] CHIMU Drifting Issues
Date: Fri, 21 Sep 2012 12:57:16 +0200

I meant:  The PFD coming back to zero at 45 deg is absolutely not normal and not acceptable.   (remove the NOT) 

PS: we had something very similar with an aspirin2.0 when invensense (also used on CHIMU-J) suddenly changed the gain of the accelerometer by a factor of 2, yielding that the AHRS filter did not use it anymore since it is programmed to "neglect too dynamics situations"... 


On Fri, Sep 21, 2012 at 12:52 PM, Christophe De Wagter <address@hidden> wrote:
Dear Cameron,

From the seconds problem (drifting also when GPS is not connected) it sounds like your accelerometers are broken/not-well-calibrated or something like that. The PFD not coming back to zero at 45 deg is absolutely not normal and not acceptable. 

And if the accelerometers are not correct (e.g. all measure zero), it can make sense that GPS speed feedback would make the attitude turn upside down after a while, even when held still. 

Did you connect the CHIMU to their test/debug/calibration tool (called CHIMU control panel)? 


On Wed, Sep 19, 2012 at 5:02 AM, Cameron Lee <address@hidden> wrote:
Hello guys,

I'm having troubles with drifting of the attitude values. It's very noticeable in the PFD. My setup is TWOG, Ryan Mechatronics CHIMU, XTend, UBlox (Leah 5H I think). The CHIMU is connected using SPI. I've put the airframe file and three flight logs here: . The drifting occurs when still on the ground. It's difficult to describe, but I'll try: the PFD responds to movements in the IMU, but there is a constant, gradual drifting. It makes the plane appear upside down at times when it's right side up. See the logs for more details. The ReadMe.txt file at the link includes more details.

I noticed somehow that the drifting would only take place when the GPS was connected to the autopilot and had a lock. Based on this, Stephen Dwyer pointed out that the problem is probably an error in how the velocity is accounted for by using the GPS velocity when there is a GPS lock. This is done in lines 97-117 in this file:
So I did a test where I changed condition of the if statement from gps.fix == GPS_FIX_3D to 0 == 1 so that the update would not be made. And in the test, the PFD behaved as if the gps was not connected. So it seems as if Stephen was correct.

Anybody have an idea what this issue is and how to fix it?

Also, when the GPS was not connected, the IMU would constantly drift such that the PFD was level. So if the IMU starts level, then you tilt it 45 degrees, the PFD would show that. But if you hold the IMU at 45 degrees, within 10-20 seconds the PFD would show level again. This behavior is also very noticeable in the real time plotter - you can see the angles always decay exponentially to zero. Does anybody have an idea why this is and if it's a feature, or something I need to fix? When I connect the CHIMU to it's my computer and use the CHIMU control panel, I don't see this behaviour.


Cameron Lee

Paparazzi-devel mailing list

reply via email to

[Prev in Thread] Current Thread [Next in Thread]