The Filter command provides several methods for digitally filtering time histories. All of the filters employed by DPlot use the following linear filter:
x = input amplitudes
y = output amplitudes
Dialog box options for the various filters are described below:
User defined. Values for the C and D arrays are set by the user. This option is not recommended for users unfamiliar with signal processing.
Frequency rejection (Carleton). Removes the frequency (specified in the “Reject frequency” box) from a time series using the methods described in WES Miscellaneous Paper N-70-1, “Digital Filters for Routine Data Reduction”, by H.D. Carleton.
Frequency rejection (Press). Removes a narrow frequency band from a time series using the procedures described in “Numerical Recipes”, Cambridge University Press, by W.H. Press, B.P. Flannery, S.A. Teukolsky, and W.T. Vetterling. The frequency to remove is specified in the “Reject frequency” box. The width of the frequency band to remove is specified in the Bandwidth box.
Bandpass. Preserves the frequencies between the lower and upper cutoff frequencies, following procedures described in “Numerical Recipes” (see above). For a low-pass filter, set the lower cutoff frequency to 0 and the upper cutoff value to the desired limit; for a high-pass filter do the opposite: set the upper cutoff frequency to a very high frequency and the lower cutoff to the desired value. Note that this filter does not have a very sharp cutoff. To obtain a sharper cutoff you can apply this filter multiple times. NOTE: For an alternative, effective low-pass filter see the Help topic for the Smooth command.
Zero frequency rejection. Removes an offset from an oscillating time function.
Option x low-pass. The four low-pass filters are described in the Low Pass Filter Help Topic. These filters each use the linear filter equation described above, but make multiple passes using different filter coefficients.
CFCx. Uses the above digital filter for a CFC (Channel Frequency Class) value of 60, 180, 600, or 1000 according to SAE J211. The minimum sampling frequency for each of these filters is 10 times the CFC number, in Hertz. For example the minimum sampling frequency for a CFC60 filter is 600 Hz (timestep no greater than 0.016667 seconds), while for a CFC1000 filter is 10 kHz (timestep no greater than 0.1 milliseconds). Filter coefficients are determined by:
where C = CFC number, T = timestep in seconds.
The time scale is important for each of the Frequency rejection and Bandpass filters and the CFC filters, since the coefficients C and D are dependent on the sampling frequency of the input.
There are two considerations with this entry: 1) Each of the filtering methods used by DPlot introduce a small phase shift to the input record. This situation may be corrected by entering a value of 2 in the Passes box. This will repeat the filtering operation in time-reversed order. 2) In some cases, particularly for noisy records, it may be desirable to repeat the operation more than twice. Odd values for passes always end in a forward pass; even values end in a reverse pass.
Check this box to replace the existing record with the filtered data. Leave this box unchecked to produce a new curve. This option is disabled in data processing mode (the existing record is always replaced).
These controls are disabled for all but the User defined filter. The coefficients used by DPlot for each of the other filters other than Option x low pass (frequency rejection, zero frequency rejection, CFCx) are shown but cannot be edited directly.
Related macro commands