Hi Dean, I can see how that statement is confusing... It is correct about the group delay: The group delay of the linear phase filter is larger. It is incorrect about the settling time: Both filters have a settling time of 62 conversion periods. Keep in mind that the group delay and settling time are different... Settling time is related to the number of filter taps. A fully settled filter is one where you've collected sufficient data to fill up all of the filter taps (after a step input, for example). Both the minimum and linear filters have the same number of filter taps, so the settling time is the same. Group delay is a measure of how many conversion periods are required until an input signal of a given frequency begins to have an effect on the filter's output. The group delays of 5 and 31 conversion periods are for a DC step input. However, if you have an AC signal composed of multiple frequency components you may see that the ringing of the filter response (as seen in figure 31) lasts much longer than what is shown for the DC step input. If you know that you input signal is mostly bandwidth limited (going by the group delay plot in Figure 32, lets say no frequency components above 60 Hz), then the group delay will fairly linear. In this case if you are continuously converting, and you switch the MUX channel without synchronizing or restarting the ADC's conversions, then the filter output will be mostly settled after 31 conversion periods as you can see in FIR step response (figure 31) Does that make sense? Best regards, Chris
↧