Hi Anush

Please bear in mind that I am not a specialist in signal analysis.

For what it is worth

• For a linear system, you can reduce the influence from noise by averaging. https://en.wikipedia.org/wiki/Signal_averaging,https://en.wikipedia.org/wiki/Signal-to-noise_ratio
• If your task is not time critical, the system is time invariant (or reasonably so), perhaps you can borrow some ideas from undersampling (https://en.wikipedia.org/wiki/Undersampling) where the signal is reconstructed in any order? You can then perhaps vary the sampling rate to squeeze more effective bits out of your hardware. I doubt this would improve your amplitude resolution much but you might then be able to improve the temporal resolution.

Sincerely

Claes

Dear Anush,

I aw your wave forms after sampling and after ideal D/A converter.I noticed the following:

- That your signal is not pure sine wave, It i a sine wave superimposed on a high DC signal level of about 1 V superimposed on  a sine wave of about .5 volt peak to peak.

Using 16 bit resolution is not not sufficient to resolve the amplitudes since one bit is resolved for the sign and 3 bits are reserved for the amplitudes. If VFS=2V, then the step size or the resolution will be 2/8= 0.25 volt, which is not sufficient to resolve the sine wave with high precision. To reconstruct the waveform with smaller quanttization 

noise or error, one has to use a higher resolution data converters. I would propose to study the effect of the A/D resolution in your system. You will get appreciable enhancements with increasing the resolution of the A/D inverter. A an example for the resolution required for voice is 13 bit /ample without compression.

What i see i the effect of too smaller resolution of the A/D converters.

wish you success

As Abdelhalim Zekry indicates, the signal has a large dc offset (1 V). Is this of interest? If not, could you apply an analogue dc notch filter prior to digitization? That will allow you to decrease the voltage range to say +/- 0.5, instead of the larger range that you currently need to use to accommodate the dc offset. That way you will be able to better resolve your higher frequency components (in amplitude). I'm assuming here that the voltage range of the A2D is adjustable. If not, could you use an analogue pre-amp to boost the input signal so that it fills the V range of the A2D?    

It seems to me that the pink plot (input signal for ADC) is clipped. If you look at the grid of your sampling resolution of 10ns then you see that there is one missing value  in the maximum of your half-wave. As long as you apply a distorted signal to your DAC input you won`t get better results on SNR or ENOB.

Hi 

I advice you to see these documents. You will find what you need.

Best regards

https://www.google.fr/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&cad=rja&uact=8&ved=0CC0QFjABahUKEwiUt7eLy8TIAhVJWRoKHWyBBM8&url=http%3A%2F%2Fwww.cse.psu.edu%2F~chip%2Fcourse%2Fanalog%2Flecture%2FSFDR1.pdf&usg=AFQjCNFc3j0RG1gr6ezO3CsQorqe2YHoOg&sig2=bBIvMzqcAys9TUc7NrvWBg&bvm=bv.105039540,d.d2s

https://www.maximintegrated.com/en/app-notes/index.mvp/id/728

As I mentioned on Oct 1, 2015: in the meantime did you try to get signals without clipping?