The 28069 is a floating point part, so you don't need to do fixed point.
Thetransfer functionblock has an S->Z bilinear transform button so you can convert from S domain to Z domain easily.
Thetransfer functionblock also has ause scaled fixed-pointoption so you can do fixed point sim and codegen if you want.
The simplest derivative is simply (u(k) - u(k-1))/dT which you can do with our 1/Z blocks, but due to standard ADC sampling noise, this method would tend amplify the noise and obscure your derivative. You can filter this by inserting a low or high pass filter before the derivative stage, or you can use a single pole filter that will give derivatives of low frequency components:s / (Tc s + 1)where Tc is the time constant of the signal you want a derivative of. How you go depends a lot on the frequency and s/n ratio of your signal and what use you want to make of the derivative.
Submitted by Anders89 on Tue, 01/31/2012 - 03:16.
A couple points Ben:
How you go depends a lot on the frequency and s/n ratio of your signal and what use you want to make of the derivative.
Unable to find an attachment - read this blog