PWL: Data Reduction

Pondering... can you do an A-to-D in Excel? ...Jim Thompson

The data is already digital...discrete samples to be more accurate. However, if you "merge" the three "incoming" channels (columns), and make/create/calculate sample times for each point, then the result will be a jittery sampling of the overall incoming waveform (the D part of the A-D) if i interpret your initial query correctly. X-Y scatter plot that and let Excel "draw" its "approximation" or "average" curve. You might even be able to get Excel to figure and show the error bands for grins. Do not know if one then can "sample" that "average" curve at regular intervals or not. But, all data points can drop into a Fourier transform (courtesy of VB) and that should show the main components with the "fuzz" or error bands being "in the mud" so to speak. Do a dB snip and an inverse Fourier on the results for a cleaned waveform?

Your point-by-point (aka connect-the-dots :-) works pretty good.

Digging thru all the various stuff I've saved from here, I came across a 2-year-old post by "The Phantom" suggesting...

(1) Take derivative (easy enough to do with almost any Spice) (2) Make a step approximation to that (3) The step break-points are the break-points from which to collect data.

General question, can (2) be even steps, or is there a more subtle way to fit?

Is the "The Phantom" still in residence? ...Jim Thompson

The derivative method seems to work pretty good.

One failing in my PSpice "tool" is that I'm watching only the LSB of my ideal A-to-D.

Any ideas on how to watch 6-bits (broadside) for _any_ change. ...Jim Thompson

Jim Thompson schrieb:

Hello,

I would use a software which can handle the 1000 data points in each data column.

Not every data set can be reduced from 1000 to 100 data points with loosing significant information.

Bye

Aha! A Non-solution. I can't choose the end-use software. It was developed by an IEEE committee... need I say more :-) ...Jim Thompson

Idealized Expectations of Electrical Engineers? ;-)

Ignorant Engineers "Engineering" Emetically ...Jim Thompson

That explains Sloman's IEEE E-mail address. :)