ivan00 Posted October 22, 2008 Report Share Posted October 22, 2008 Hi everyone ! Once again congrats for this forum. For someone like me (coming from a totally different background) it is a great opportunity to learn more about LabVIEW. :thumbup: I sorted my shift register issue out. On the pic below you can see two waveform probes. I would need an appropriate filter to get rid of the "noise" in the right one to get it to similar shape as the left one. I am using "7.1 full" version and put in the filter express. However rectangular smoothing seems to be insufficient and curve fit is too hectic on the original data. :thumbdown: I red about point to point filter and array filter maybe an array filter would be better Could someone help out Regards, Ivan Quote Link to comment
Anders Björk Posted October 22, 2008 Report Share Posted October 22, 2008 How about using medianfilter instead of averaging? Quote Link to comment
Antoine Chalons Posted October 22, 2008 Report Share Posted October 22, 2008 QUOTE (Anders Björk @ Oct 21 2008, 01:56 PM) How about using medianfilter instead of averaging? I agree with Anders, I almost always use median filter instead of average (and not only for signal acquisition). Quote Link to comment
Francois Normandin Posted October 22, 2008 Report Share Posted October 22, 2008 QUOTE (Antoine Châlons @ Oct 21 2008, 08:49 AM) I agree with Anders, I almost always use median filter instead of average (and not only for signal acquisition). I agree. Moving averages can be tricky when doing signal processing. It "can" smooth a trace efficiently in some circumstances but it has the bad habit of shifting your signal temporally. Frequency filtering is much better for time-critical traces. Quote Link to comment
mje Posted October 22, 2008 Report Share Posted October 22, 2008 The "best" filter to use often involves a solid understanding of both your signal and noise source(s), it's fairly large topic in analytical science. Median filters are a very powerful swiss army knife type of filter, and produce excellent results in many situations. A properly set low-pass filter can often do wonders to data where noise is of much higher frequency than your signal, as can a properly constructed convolution filter. I believe all of these options are available to "Full" versions of LabVIEW, though I can't say for sure in your case, I skipped the 7.X generation. Quote Link to comment
ivan00 Posted October 22, 2008 Author Report Share Posted October 22, 2008 Thanks for the input ! Below the result with a medianfilter (right) compared to the curve hoe it should look like ... the curve doesn't look too bad, but I can't apply low pass somehow, as the data was already filtered through low pass 5Hz, 2 oder Butterworth (tried with express vi "filter") I just need a god da... curve fit or whatever too smooth it ... :headbang: P.S. Am from Cape Town / South Africa = someone from the same city Regards, Gunnar Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.