It seems to me that you are obscessing with how while not worrying enough about what. What is it that you want to do? From your last post, it sounds like you are sampling at a fairly high rate (>1KHz, right?) and maybe displaying at a much lower rate (perhaps 50Hz or slower). I already demonstrated to you that a logical explanation for your Spider Web was uncorrelated "noise" in your signal. So what you should think about is diminishing the noise level of your signal.
Suppose you sample at 1KHz, and want to display 20 points/second. You could display every 50th point, in which case you'd (probably) see a Spider Web, or you could display the mean of the 50 points, which would reduce the variance of the noise by a factor of 50! This, all by itself, might be all that you need to do!
Here's a Funny Story to tell on myself. I had exactly this situation -- I'm sampling data (as it turns out, of eye movements) at 1 KHz, streaming it to disk, and displaying it on a scrolling Chart display at 20 Hz. I "follow my own advice" and display the mean of every 50 points being saved, which gives me a nice, smooth trace. However, I'm using an old-fashioned technology to get eye position, inherently noisy, and the quality of the noise is a useful indicator of whether my electrodes are working or not, but I've "accidently" averaged the noise right out of my signal! The solution was simple -- instead of calculating (and displaying) the mean of the 50 points, I just showed the first one, instantly "restoring" the Noise in my Signal.
Bob Schor
