Hello, I'm making a SAR ADC. Is there a way to analyse the 1/f noise? Or to model it to have an estimate of from which frequency it will give problems (maybe define problem as it will become equal to my quatization noise)? As a starting point I found the very general K/(Cox*W*L*f) formula but I am not sure which paramters to use for these constant from the spice files. But I still hope there is some way to verify it directly based on my schematic. (let me know if it is case specific and it would help if I dropped my schematic/spice netlist here) Have a nice day

A key spec to think about is what you expect to be the maximum record length over which the noise will be observed. You need integrate the 1/f noise PSD from the reciprocal of that time to the bandwidth of your system. You can get the parameters for the equation you have above by running a quick ac noise simulation. You really just need to find the constant in PSD = constant/f.

@Boris Murmann, Nice and clear explanation!💪