The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> Unusal FFT of the ADC Output
https://designers-guide.org/forum/YaBB.pl?num=1604129646

Message started by niloun on Oct 31^st, 2020, 12:34am

Title: Unusal FFT of the ADC Output
Post by niloun on Oct 31^st, 2020, 12:34am

Hi Everyone,

I'm getting an unusual FFT from my ADC output. I have attached an image of the FFT here. Could you please tell me what is happening here?

Thanks

Title: Re: Unusal FFT of the ADC Output
Post by polyam on Oct 31^st, 2020, 6:19am

What did you apply to the ADC? Is it a single tone? Can you DAC the output of the ADC to see what you get?

Title: Re: Unusal FFT of the ADC Output
Post by niloun on Oct 31^st, 2020, 11:25pm

polyam wrote on Oct 31^st, 2020, 6:19am:

What did you apply to the ADC? Is it a single tone? Can you DAC the output of the ADC to see what you get?

Thanks for your response polyAM.
The input is a single tone sinusoid wave. The input frequency is 387.5M Hz, and the sampling frequency is 800M S/s (Nyquist frequency). The type of ADC is Time-Interleaved ADC.

I use Matlab for signal processing. Here is the reconstructed output:

Title: Re: Unusal FFT of the ADC Output
Post by polyam on Nov 1^st, 2020, 7:45pm

Are you sure you are correctly DACing the output signal ? Clearly the amplitude of your output signal is being modulated. Maybe you need to check your demux. I am guessing there is something wrong with the demux. I did a quick simulation for a TI-ADC. Please see the input and output of the ADC. You should be able to see your input signal (of course with a delay) when you DAC the output of the ADC.

Title: Re: Unusal FFT of the ADC Output
Post by niloun on Nov 1^st, 2020, 8:13pm

polyam wrote on Nov 1^st, 2020, 7:45pm:

Since I'm doing the sampling in Nyquist frequency, I think I should see the modulation. If you decrease the sampling frequency to less than half you should see the same thing. Let's say you have Nfft=64 Cycles=31 then if your sampling frequency is 800M S/s your input frequency must be 800*31/64 which is 387.5M Hz. If you run the simulation for at least Nfft*(1/Fs) seconds then you will see the same thing.

I have also done the simulation for less than Nyquist frequency (1/10), and my DAC output looks more like yours.

Title: Re: Unusal FFT of the ADC Output
Post by polyam on Nov 2^nd, 2020, 2:26pm

Likewise, I am using the same approach. My sampling frequency is fs=2GHz (BW=1GHz) and I am running 2048 FFT points for 191 cycles. That results in an input frequency of (191/2048)*fs. So, I am not indeed down-sampling and I can do FFT. My suggestion is to trace the signal from the input, output of the T&H, check each sub-ADC individually and the output of the demux. Run FFT at each node and see if your single-tone is there.

Also my simulation time is NFFT/fs. Also I considered a couple of cycles for transient.