Hello, I'm working with a Maxon quadrature encoder and an E-series PCI card. Due to the limited capabilities of my hardware for motion control, I am counting encoder pulses and estimating velocity from the derivative of the pulse count. I would like to deal with the significant quantization error of this calculated velocity (see rawVel.jpg). Rather than lowpass (or other) filtering, I would like to downsample or decimate the array of calculated velocities, perhaps by as much as a factor of 100. Simply decimating the array leaves zero values, however, as you can see in the attached image (Decdzeros.jpg). I think I would like to fill these zero values with an average of the leading and trailing velocity values. In other words, if the decimated array looks like:
3000000004000000005
I would like
3 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5 4 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 5
or something like this, maybe even a more sophisticated averaging or weighting scheme. Am I describing some other filtering scheme that is easily implemented?
Please see the attached VI. Eventually, this velocity is going to be the measured variable in a PID control loop, so I would like to clean up the signal as much as possible. I would also like the loop to run as quickly as possible because of the PID control. Problem is, I just don't need to sample my velocity at such a high frequency.
I know that I could use the measure period VI to detect velocity from my encoder with this e-series board, but I would also like position, and I haven't been able to simultaneously count pulses and measure pulse period, I believe because of the STC counter on this E-series board.
Any help is appreciated!
