10-21-2020 03:40 AM
I am acquiring two signals in time domain and I am sending the data through a queue to another loop where I apply FFT and PSD analysis to those signals to get them in the frequency domain and have an idea how the noise in the signal looks like.
But as you see in the attached pictures the noise behaviour is non existent and as far as I know every sensor no matter how amazing it is, it doesn't acquire the signal directly without any noise so I am quite confused now of how the signal looks so clean. You can find my FFT SubVI attached as well.
10-21-2020 03:48 AM
Hi Amr,
@Amr95 wrote:
But as you see in the attached pictures the noise behaviour is non existent and as far as I know every sensor no matter how amazing it is, it doesn't acquire the signal directly without any noise so I am quite confused now of how the signal looks so clean.
You may switch the graph to a logarithmic Y scale…
@Amr95 wrote:
You can find my FFT SubVI attached as well.
Well, it misses many subVIs/dependencies. And it doesn't include any example data…
10-21-2020 04:29 AM
@GerdW wrote:
Well, it misses many subVIs/dependencies. And it doesn't include any example data…
That's right. The SubVI is a part of a massive program/project that I am not allowed to upload but I thought that maybe uploading the subvi would give a better understanding of what I'm trying to do or achieve.
@GerdW wrote:
You may switch the graph to a logarithmic Y scale…
Well that looks better. I thought the graph is in decibels automatically when I apply a FFT or PSD, thanks for the tip.
That's how my signal looks now. Could I conclude from it that the noise has no big effect on my signal in the time domain as the power of the noise is so low?
10-21-2020 04:56 AM
Hi Amr,
@Amr95 wrote:
That's how my signal looks now. Could I conclude from it that the noise has no big effect on my signal in the time domain as the power of the noise is so low?
We don't know anything about your signal.
Atleast there seems to be some offset in your signal as the amplitude for f=0 is rather large…
10-21-2020 05:13 AM - edited 10-21-2020 05:16 AM
Hi GerdW,
thanks for your quick reply. I am have a molar conductivity sensor which is measuring the conductivity of two fluids passing through it. I'll take the BIC signal as a reference, the first fluid is water which is flowing continuously as it is being pumped by continuous flow pump and has a conductivity of almost zero and the other fluid is some kind of acid, it is pumped to the fluid by rotary vane pump so it only a small volume of the acid is being fed to the continuous flow of the water after every pump revolution and it has a conductivity of around 1,7. Due to the behaviour of the rotary vane pump the value of the conductivity keeps getting high as the conducitvity of the acid is measured and then gets low again as the conductivity of water is measured which explains that representation of the signal in the time domain. The frequency of the Pump is around 0,75 revolution per second so that's why it looks like the amplitude at 0 is so large.
My goal is to evaluate if the noise has an impact on my signal or not and if it has then I have to calculate the periodic signal without that noise and my idea was to do the PSD and see what is the power of the noise and then only take the values above that power and do a reverse FT to get the periodic signal. I am not sure if it's worth the effort though as I see that noise has a quite small power but I don't have much experience with signals so I would be happy to hear what you think
Best regards,
Amr