Hi,

I would like to do some spectral analysis of some square waves and step responses. I would like to see what frequency the noise that is sat in the signal is, in a hope to either find the parasitics that lay within my circuit and nerf them or be able to design a filter them out. I would also like to see the bandwidth within the signal. Eg. I know that the sharper a rising/falling edge is, the higher the bandwidth of the signal. As i want to verify that the bandwidth is as desired.

I assume that i would just do an FFT on the captured waveforms to obtain what i'm looking for?

I've been looking at windowing to improve the results. Am i correct to believe that for a square wave i should use a uniform/rectangular/no window since it has leading and following zeros?
I'm not sure what i should use for the step response.....my gut instinct says i should also use a rectangular window because i imagine that all other windows will distort the waveform before being processed by the FFT. But a step response won't have both leading and following zeros, they'll just have one of the other. So what would i do here?

Thanks

 

Hi,

I would like to do some spectral analysis of some square waves and step responses. I would like to see what frequency the noise that is sat in the signal is, in a hope to either find the parasitics that lay within my circuit and nerf them or be able to design a filter them out. I would also like to see the bandwidth within the signal. Eg. I know that the sharper a rising/falling edge is, the higher the bandwidth of the signal. As i want to verify that the bandwidth is as desired.

I assume that i would just do an FFT on the captured waveforms to obtain what i'm looking for?

I've been looking at windowing to improve the results. Am i correct to believe that for a square wave i should use a uniform/rectangular/no window since it has leading and following zeros?
I'm not sure what i should use for the step response.....my gut instinct says i should also use a rectangular window because i imagine that all other windows will distort the waveform before being processed by the FFT. But a step response won't have both leading and following zeros, they'll just have one of the other. So what would i do here?

Thanks

Are you talking about a data stream or just a pulse?

 

I'm talking about just a single pulse.

 

But i didn't want to say pulse on the chance that people think i mean an impulse. As the pulse length is ten to twenty times longer than the leading edge.

 

But i didn't want to say pulse on the chance that people think i mean an impulse. As the pulse length is ten to twenty times longer than the leading edge.

Maybe an example would help.
Do you have a circuit with an example of the step?