Assuming the vertical scale doesn't change, there should be no change to a particular trace as it scrolls across the screen, even if there is more than one sample per pixel. That's where a peak detect plotting algorithm helps out. Imagine an ink pen plot of some signals then stand back about 10 feet -instead of the individual points you would just see bands almost like bar graphs. So when plotting to a screen with a finite resolution, for each horizontal pixel location, there are two pixels plotted in the vertical axis with the pixels in between filled in.
-Bob On May 2, 2011, at 11:38 AM, Ed Palmer <[email protected]> wrote: > I suspected that the graph issues might be involved with some display > resolution issues. So then if a spike appears and disappears at some point > in time, does that mean that in the original data there was a combination of > spike and no-spike data and the sub-sampling first picks one and then the > other? But does that mean that the standard deviation values are calculated > from the sub-sampled values rather than the original data? > > Thanks, > Ed > > Mark Sims wrote: >> When you are viewing a time interval that is longer than one second per >> screen pixel, the program must sub-sample the data. Every second the >> screen is redrawn with a new set of samples. Also the plots are rescaled >> according to the data being plotted (if auto-scaling is turned on). With >> very noisy data, like the raw oscillator plot, you see the effects of what >> amounts to a new data set every second. Turning on display filtering (F D >> command) helps smooth out the changes (but can hide very short >> disturbances). >> _______________________________________________ >> time-nuts mailing list -- [email protected] >> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts >> and follow the instructions there. >> > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. > _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
