[Top][All Lists]

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

Re: [Paparazzi-devel] re: razor imu

From: Felix Ruess
Subject: Re: [Paparazzi-devel] re: razor imu
Date: Wed, 19 Jan 2011 15:36:41 +0100

Hi Oliver,

the SENS values in hb_tiny_razor.xml airframe file (4096 = 12 Bit) are only set to reverse the BFP
formating what ist done for rates in RATES_FLOAT_OF_BFP(gyro_float, imu.gyro); in ahrs_propagate();


/* convert bfp imu data to floating point */
RATES_FLOAT_OF_BFP(gyro_float, imu.gyro); // olri div by (1<<12)

/* unbias rate measurement */
RATES_DIFF(ahrs_float.imu_rate, gyro_float, ahrs_impl.gyro_bias);

The trick to get it work is to scale the values from deg/s to RAD/s,
see the lines below from my ahrs_propagate();.

// scale gyro adc raw to [rad/s] FIXME // olri
ahrs_float.imu_rate.p /= 61.35f;
ahrs_float.imu_rate.q /= 57.96f;
ahrs_float.imu_rate.r /= 60.10f;

Putting scale factors in the source files is a really bad solution in my opinion. That is what the SENS parameters are for.

The result of the that *must* give closest-by 0.0... rate on silence!

It's at the moment *impossible* to combine these 2 factors,
becourse the bias is substracted between these to steps.

Why should this be impossible? That is how it is done for all other IMUs...
The aligner runs on the already scaled sensor values, then the averaged gyro values are converted to float in ahrs_align and used to unbias the (scaled) gyro measurements that were just converted to float as well.
The neutrals are basically the raw values (imu.gyro_unscaled) you read when not moving, the sense parameters do the proper scaling to imu.gyro. The BFP values imu.gyro then get converted to float with RATES_FLOAT_OF_BFP.

I worked hard for finding out that ;-), but i didn't see the
jumping point to solve it better, i am not fimiliar with BFP.

BFP (Binary Floating/Fixed Point) is just how we store the values in fixed-point. We just set the decimal point to a fixed position so to speak. (INT32_RATE_FRAC=12, INT32_ACCEL_FRAC=10)

So the easiest way to get this right is to just log the raw values and then run the calibration script as explained on
Of course feel free to update that page.

Hope this clarifies things a little.
As far as I can see it should work with the correct neutral, sens and sign parameters, although I might be missing something...

Cheers, Felix

reply via email to

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