Dear Parminder Kaur ,

In order to get signal with noise like the bottom figure, you have to input the same signal to the ADC containing a sinewave signal in addition to noise assuming noiseless circuit. In order to get noise at the output you have to input a noise signal in addition to the pure sine wave input.

The other solution is to introduce the noise sources in the circuit itself.

Best wishes

Hello, Ms parminder kaur

The best and easiest way to get FFT is to use version 6.1.7 of Cadence. First, you apply a sine input to the SH circuit, the frequency of which is proportional to the following relation:

f_in=(f_sample* M)/N

In this regard, f_sample, M and N indicate the sampling frequency, the number of sampled cycles, and the number of samples, respectively. Note that M must be a prime number. Also, N must be a multiple of 2. Then simulate the circuit and draw the output waveform (transient time = N/f_sample). On the waveform page, go to the measurement tab and select the spectrum option. Finally, set the number of samples in the window that opens, select the desired sampling window, and press the S button in the Start/ End freq section. Select the desired number of harmonics and finally click on the plot. Dynamic specifications and FFT shape are shown to you.

Good luck

Dear Mehdi Sotoudeh ,

I simulated the spectrum of bootstrapped sampling circuit according to what you said. But I find an interesting phenomenon. When I increase the number of samples, i.e. M, the ENOB of this circuit increases. when the input frequency is (503/1024)*1MHz, the ENOB is 4.19bit. when the input frequency is (2039/4096)*1MHz, the ENOB is 5.17bit. when the input frequency is (4091/8192)*1MHz, the ENOB is 5.65bit. It's very confusing! Can you help me?

Dear Jiexi Wang,

Increasing the number of samples at low or medium resolutions can improve ENOB. But in general, too much increase can affect the FFT process, so the result is unacceptable. For example, increasing the number of samples to 16,384 is no longer acceptable and ridiculous in your simulation (you can test it). In fact, as the number of samples increases, it is likely that the samples will be similar or close to each other and will ruin the result of the FFT.