Help me with some electrical engineering stuff!

Status
Not open for further replies.

alphatarget1

Diamond Member
This is probably going to sound retarded to many EEs... but here it goes.

Let's say I'm trying to measure a DC signal. The oscilloscope picks up some noise in the signal and I get discrete points for data. I can apply a digital filter with my software to filter out some of the noise, but I want to know which frequency to apply it. I'm assuming this digital filter is some algorithm written to behave like a bandpass filter, but I'm not 100&#37; sure.

I know for a continuous function, I can apply FT to convert everything to frequency domain to see where the noise is. If I do DFT to a discrete set of collected points, I'll turn the points into the frequency domain...

That's fine and all but how do I pick out which frequency is which from only discrete points? Obviously the points will have a time associated with it, but this is kind of beyond my field of expertise.

Gibson486

Lifer
This is probably going to sound retarded to many EEs... but here it goes.

Let's say I'm trying to measure a DC signal. The oscilloscope picks up some noise in the signal and I get discrete points for data. I can apply a digital filter with my software to filter out some of the noise, but I want to know which frequency to apply it. I'm assuming this digital filter is some algorithm written to behave like a bandpass filter, but I'm not 100&#37; sure.

I know for a continuous function, I can apply FT to convert everything to frequency domain to see where the noise is. If I do DFT to a discrete set of collected points, I'll turn the points into the frequency domain...

That's fine and all but how do I pick out which frequency is which from only discrete points? Obviously the points will have a time associated with it, but this is kind of beyond my field of expertise.

problem is that you are on a DC signal. Normally, you would subtract the frequency of the signal you are measuring and that leaves the noise. Noise is spurious and random, so you need to just guess, and that's if the noise has a frequency at all. If you do not know the frequency of your own signal, it is kind of tough.

Last edited:

Born2bwire

Diamond Member
What I would do is use a very low pass filter and then take the mean.

Cogman

Lifer
problem is that you are on a DC signal. Normally, you would subtract the frequency of the signal you are measuring and that leaves the noise. Noise is spurious and random, so you need to just guess, and that's if the noise has a frequency at all. If you do not know the frequency of your own signal, it is kind of tough.

This. Since your signal is supposed to be DC (correct?) Any noise that is occurring should be eliminated. The easiest way to do that is throw in some capacitor from line to ground. Using some sort of frequency filter only makes sense when your signal has some sort of AC component to it (IE a sound signal)

drinkmorejava

Diamond Member
What Cogman said.

If you do the FFT, there are a lot of things you need to be aware of first if you want good data. You'll need to estimate a frequency of interest, then determine the Nyquist frequency and the sample rate. You'll also need a low pass filter at the Nyquist and know what it's roll-off is. You're dataset is also going to need to be of a relatively short interval, maybe 1/10 of a second, to not violate the stationary principle for FFTs. You also need to keep in mind that there will always be 1/f noise around 10Hz and below

Basically, the point is that you're probably over your head and should just use a capacitor to filter out the noise.

Status
Not open for further replies.