Extracting high dynamic DC offset helps needed

Hi everyone,

I have a problem with extracting high DC dynamic offset from a signal to get a relatively small AC perturbation.
I have been struggling to find out the proper method but still no way out. I post this thread here hoping that someone would help me out of this

Here is my case:

- A voltage combined of: slowly decreasing from 12V to below and a small AC component with amplitude is around 100mV to 300mV (enclosed picture)

- I want to measure AC perturbation by extracting increasing DC offset and then amplify it before going to measurement devices.

- I have found some techniques but those are so complex, high accuracy and high cost as well; in my case, I need a very reasonable accuracy and low cost solution...



Please drop your comments, I am grateful.

Max.

first of all you need to find out the method to store the waveform displayed by your CRO on PC then you will get this as *.txt tile or some kind of file.....I think LeCroy support that....then you need to read the Values of this from PC as and array apply FFT to it to get the require analysis like DC component....frequency component contribution from the magnitude spectrum.....or some statical method like average value of array and range of variation as maximum value of array - Minimum Value of array the you can get the percentage variation as 100*(range/average value).....if you have more than 10 significant contribution from different frequency component then you need bit complicated analysis method....

Good Luck

Max_Soo said:

- A voltage combined of: slowly decreasing from 12V to below and a small AC component with amplitude is around 100mV to 300mV (enclosed picture)

- I want to measure AC perturbation by extracting increasing DC offset and then amplify it before going to measurement devices.

Click to expand...

You could use a high pass filter. A simple RCRC filter may be enough, but if you are going to amplify the signal anyway, you may as well configure the amp as a VCVS filter.

milind.a.kulkarni said:

first of all you need to find out the method to store the waveform displayed by your CRO on PC then you will get this as *.txt tile or some kind of file.....I think LeCroy support that....then you need to read the Values of this from PC as and array apply FFT to it to get the require analysis like DC component....frequency component contribution from the magnitude spectrum.....or some statical method like average value of array and range of variation as maximum value of array - Minimum Value of array the you can get the percentage variation as 100*(range/average value).....if you have more than 10 significant contribution from different frequency component then you need bit complicated analysis method....

Good Luck

Click to expand...

Thank you very much for feedback.

I got your idea but I am gonna measure this with a microprocessor in portable application then I don't use PC, Oscillator.

Cheers.

Max.

- - - Updated - - -

godfreyl said:

You could use a high pass filter. A simple RCRC filter may be enough, but if you are going to amplify the signal anyway, you may as well configure the amp as a VCVS filter.

Click to expand...

Hi godfreyl,

My frequency range of AC perturbation is from 1Hz to 1kHz. Is it okay if I use a simple High pass filter RCRC.
Btw, I am too much consider the moving DC offset, how come a simple filter can remove the dynamic DC offset??

Hereinafter some materials I found before:
Figure 4.9: http://etd.ohiolink.edu/view.cgi/Park,%20Seok-Bae%20.pdf?osu1149024229
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1277630&tag=1

Thanks so much!

Max

I suppose whether it can work depends how fast the "DC" is shifting. The filter cut-off frequency needs to be below the lowest frequency you want to measure. 1Hz is very low.

Max_Soo said:

how come a simple filter can remove the dynamic DC offset??

Click to expand...

The slowly moving signal looks like part of a very low frequency signal.

godfreyl said:

I suppose whether it can work depends how fast the "DC" is shifting. The filter cut-off frequency needs to be below the lowest frequency you want to measure. 1Hz is very low.

The slowly moving signal looks like part of a very low frequency signal.

Click to expand...

So we can look DC offset like a very low-frequency signal (mHz) then applying High-pass filter to extract it. I am confused that 1Hz is very low, then my cut-off frequency has to be under 1Hz, it seems not okay for a High-pass filter. I might to find out other way

Posting up, hoping someone has a solution for this

All time-continuous analog methods to separate the signals need to refer to active filters with rather low (< 1 Hz) characteristic frequency. You seem to assume, that they can't be implemented, why?

On the other hand, methods based on digital signal processing are most likely more versatile and promise higher resolution.