03-10-2011 05:05 AM
I have a series of point which I have as an array, and a separate array which contains the timestamps for the points (and a version where I have converted the time to double). I want to perform several operations on this data (filter,downsample, plot etc) so the easiest option appears to be to convert the array of points into a signal. The problem I have is the "build signal" function appears to require a constant delta t between the points and my data has a variable delta t.
Does anyone have any suggestions on how I could create a signal type from this data?
Thanks.
03-10-2011 07:21 AM
First, most signal processing algorithms (filters, FFT, ...) require uniform sampling to work correctly. So the VIs which take either arrays or signal datatypes as inputs are assuming that the data is uniformly sampled in time.
Look at the interpolation and the resampling VIs. I have not used these, so I do not know the details. I think they include methods to convert non-uniformly sampled data sets to uniformly sampled ones. Look carefully at the help files for details. In work like this it is very easy to generate meaningless results from perfectly good original data by not following the rules.
Lynn
03-10-2011 09:53 AM
Thanks. I feared as much. I don't want to resample as it is important to keep my data with the current timestamps. Looks like I will have to process the data myself as it stands.