2
\$\begingroup\$

Short question: how to select the correct bandwidth for a sensor?

Some further details. I'm working with an ADIS16260 (rate gyro) placed upon an industrial vehicle. This sensor has both an analog filter on the front-end and a digital filter. I need to set their parameters.

I'm asking a rule-of-thumb to set them. What empirical measures I might take? I don't have a mathematical model of the equipment. It's quite heavy, so movements are slow, but I still need to determine the order of magnitude of the bandwidth.

\$\endgroup\$
2
\$\begingroup\$

As always it depends on what your are doing with the data,

If you are using the data in a closed loop system, the rule of thumb is that the sensor bandwidth should be at least 5-10 times the loop bandwidth

If you are just recording data for later analysis or whatever, its better to choose the highest bandwidth possible and filter in software later if needed, (since this is possible but it is not possible unfilter! and increase the bandwidth in post processing), the important thing to keep in mind is that your sampling frequency should be at least 2times the sensor bandwidth, so if you cant achieve high sampling, lower the bandwidth to prevent aliasing.

\$\endgroup\$
1
\$\begingroup\$

By the context I take it by 'sensor bandwidth' you are referring to the frequency range you are going to measure using the sensor, and not the frequency response of the sensor itself.

The cutoff frequency depends on the maximum frequency components of the measured process. What is is exactly that you are trying to observe with the sensor? Get some data sets of this process and determine where the frequency components are. From this you can set the analog filter to remove all noise components higher than this value.

Note also that due to Nyquist rate, the cutoff frequency should be less than half the sampling frequency.

\$\endgroup\$
  • \$\begingroup\$ I'm trying to observe the angular rate to improve the PID algorithm to maintain the heading (measured with a compass). The gyroscope is more fast than a compass and this help to avoid overshoots. To mix both information I use a complementary filter. \$\endgroup\$ – Mark Nov 2 '17 at 9:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.