What kind of analysis should I use?

[jimalex]

I have been using an oscilloscope more at work, and I would like to learn more about analysis and curve fitting, rather than just be at the mercy of analysis software on the computer or the tablet. I'm an electrical tech.

But there are dozens of flavors of analysis, and I don't know which one I should study. most of the time simple trig and algebra are good enough, but some of the measurements are pretty out there in terms of finding the function to fit the data.

Which branch of analysis would be the best place for me to start? I've had math up to calculus.

Thanks for the help,
Jim

 

[Stephen Tashi]

If you say what you are measuring with the oscilloscope perhaps you'll get better advice.

 

[jimalex]

I will be measuring electronic components, communication signals and industrial transmitters, usually low voltages and currents, as well as higher power devices like motor drives, power supplies and transformers.

often when the loads are simply RC, RL and RLC (combinations of resistive, inductive and capacitive) the functions are exponential or trigonometric or a superposition of both. sometimes, though, they are a lot more complicated, such as when i look at the waveform of a power circuit feeding a motor drive, or the stray signals on the shield of a motor cable (noise). one of the specific things i want to work on is making noise filters on signal cables that are specific to the particular noise sources in the immediate environment of one of my systems.

I looked around on the 'net for a free pdf book on curve fitting and basic analysis, but came across so many different types of analysis I got lost trying to figure out which particular branch of analysis i need.

i apologize if i am vague on the question, i just don't know enough about the analysis subject to put the question well, maybe.

thanks,
Jim

 

[Stephen Tashi]

I'm not an electrical engineer, but my guess is that the subject of "fourier analysis" would be useful. It amounts to expressing data as a sum of waves, but even irregular shapes can be represented this way. I'm sure you can get software to do Fourier analysis, but you said your interest is in understanding the methods. There are various levels of understanding. The typical electrical engineering student would have taken some calculus before tackling it.

Are you going to design filters with all passive components? Or will you use active components or even digital signal processors?

 

[jimalex]

I think I'll be using passive components wherever possible, but I am still looking into that also.

Thanks for the Fourier analysis tip; I've seen that but haven't messed with it much. I found a pdf on the 'net from a college math class called numerical methods which goes over some simple curve fitting methods. That is getting close to what i was originally looking for, but I can tell there is a lot to the subject I don't know and still appreciate good advice.

Thanks,
Jim

 

[chiro]

Hey jimalex and welcome to the forums.

If you want to do this kind of signal analysis, I would look into the area of integral transforms which include Fourier decomposition and the generalization of the Fourier series method known as wavelets. One reason why wavelets are so useful is because with wavelets you can get bases for signals/functions across the whole real line (and the same kind of thing for higher dimensions) vs over only a finite interval.

In terms of the curve fitting, get a good book on interpolation. There are many techniques for this but the basic one is the Lagrange polynomial:

http://en.wikipedia.org/wiki/Lagrange_polynomial

 

[jimalex]

Thanks Chiro-

Interested in the wavelets. found what i think are a couple of good books including numerical methods that work, which seems highly recommended.

also, found smath studio and started playing with that. It's a mathcad clone.

i appreciate the info, very much.

Jim