Hello all, Analog simulation question here: I'm calculating the FFT of the time-domain output ngspice simulation of a first order sigmadelta VCO-based ADC to obtain its PSD to evaluate its SNR. I'm using simple ring oscillators, but I wouldn't expect noise sources to be active (thermal + flicker) in the simulation. Nevertheless, i'm seeing some low-freq noise floor in the PSD that I was not expecting. I wonder if it does have noise sources activated in the simple transient simulation (.tran)? can anyone here help me with this? if so, how can I shut those noise sources down? Or maybe there's some other effect I'm not considering? Sorry if the question is too specific, but there's a lot of knowledgeable people in the analog domain here, so I was hoping someone could give a hand.

How are you computing the FFT? You need to use a window to suppress quantization noise leakage from out of band to in band.

Hello @Boris Murmann, thanks for the reply. OK, no noise, as i was expecting. I tried both Hanning and Blackman-Harris windows, but I probably need to come back to analyze the computation better.

Increase the number of samples you are using for your FFT calculations.

People tend to use Hann windows for delta sigma modulators, see literature.

I hope it helps. I'm also encountering the same issues with my Delta Sigma modulator design. I can't go past 55 dB SNDR

Thanks for the reference, Luis!