@User this is a quick adaptation of a design i did on a totally different process (and on a different commercial simulator) , so it seems to work but some optimizations (or even fixes / improvements) are still to be done. I have individually set gaussian variability on the

sky130_fd_pr__***__vth0_slope_spectre

parameters and on the

sky130_fd_pr__res_high_po__var_mult

that is used for poly resistors, doing some simulations varying one at a time these parameters until a reasonable variation was observed on individual transistors/resistors. Sad i do not have actual variation parameters for the sky130 devices, the good thing is that variations are scaled by sqrt(w*l*mult) (so big devices show less variability).