Gyro sampling rate?

[skogsvargen]

I am experimenting with my gyros and calculate the roll and pitch angles. What sampling rates of the gyros are necessary to integrate the angular velocity do an angle?

/Magnus

[hugo_vincent]

I use 50 Hz, many people seem to get away with 25-30 Hz. It depends what sort of airframe you use, how much CPU power you have available, and what sort of accuracy you need.

[skogsvargen]

Well I will use a quite stable airframe, but I really don't know what accuracy i will need. So you are sampling the gyro-rate once for every Kalmanfilter iteration? I thought I needed to sample the gyro at a higher rate to get good integrated angle value.

I have also 50 Hz as my target sampling rate for the whole IMU.

/Magnus

[Unterhausen]

50Hz is more than adequate for a stable airframe.

[hugo_vincent]

Skogsvargen, yes, I am sampling the gyros once for every KF iteration. The angle integration happens inside the Kalman filter in my case, although other ways of integrating are possible. I used to use a complementary filter, where the gyro data was integrated using a 5th (or was it 6th?) order Simpsons rule numerical integration method.

If you don't have good analog filtering of the gyro outputs before the ADCs (or if you are using digital-output gyros), then you will need to sample them much faster (e.g. 400 Hz) and low pass filter the measurements in softare (e.g. with an FIR filter), then downsample them to 50 Hz. It is much easier to build good analog filtering into the hardware. (We used 3rd order Bessel filters. )

Hope this helps.

[skogsvargen]

Thank's Hugo
I am using a gyro with analog outputs and i have a low pass filter before tha AD. Looking att the sampled data it looks quite noisefree so i thing I will continue with one sample per iteration to begin with.

What resolution are you sampling the gyros with?

/Magnus