How well-modeled is NMOS with -10mV <= V_DS <= 10mV? In the raw data, it seems the smallest V_DS step they do is 50mV, but maybe the theory is good enough to make up for this?