Explicit dynamics analyses typically involve a large number of time history samples, sometimes in the order of hundreds of thousands, and the results tend to include high frequency noise that can obscure slow rate phenomena. A low-pass filtering option is available that allows you to separate slow-rate trends from high frequency noise in signals. This feature can be controlled from the Details view of a Result Tracker object.
The filtered results are displayed by default in the Graph window after the solve. By setting Display Filter During Solve to Yes in the Details view of the Solution Information object, the filtered results can also be displayed in the Worksheet at each refresh interval of the Result Tracker.
To configure the low-pass filter for the sampled data:
Under Filter, set the following controls:
Type: Set to one of the following:
None: (Default) No filtering is applied to the data.
Butterworth: Applies a four-channel low-pass Butterworth filter to the data. Two channels are passed twice, once in the forward direction and once in the reverse direction, to prevent phase shifts.
Cut Frequency (displayed if Type is set to Butterworth): Set to the desired cut frequency in Hz or MHz depending on the current unit system. The default is 0, which implies no filtering.
Notes
A time history data is composed of a limited number of frequency signals that bound the range of meaningful cut frequencies to use for filtering. If the cut frequency is too low, most signals will be lost. On the other hand, if the cut frequency is too high, the signal may remain unaltered.
In determining a good cut frequency, sampling frequency plays a role. The sampling frequency can be obtained by dividing the number of samples by the sampling duration. The cut frequency should not exceed a quarter of this value. For example, if 15,000 samples occur in 0.015 seconds, the sampling frequency will be 15,000/(0.015 s) = 1,000,000 Hz = 1 MHz. Consequently, the cut frequency should not exceed 0.25 MHz.
The process of filtering pads the original signal with extrapolated data. This may produce unexpected shapes in the filtered signal near the margins. The data away from the margins should reflect, however, the proper trends and slow rate phenomena.
The signal is not filtered at all if it has less than 11 samples.
Under Filter, if Type is set to Butterworth, there are also read only indications for the Minimum and Maximum values of the filtered data.