The FX-120i has a ~1 second stabilization time, and SD of repeatability of 0.015 gn, and a linearity error of 0.03 gn. I don't see it specifically listed but looking at the specs I would think the minimum detectible change would around 0.015 to 0.03 gn.
Creedmoor doesn't list that information (at least not anywhere I can find), but if you called them I assume they'd be able to tell you.
This is a very interesting article apparently from the guy at Creedmoor who came up with the specs for their scale, as a direct counter for the A&D FX-120i. The software side uses grains natively instead of grams and the calibration weight is better for the range of weights loaders use.
www.theballisticassistant.com
If the McD team could get software compatible with the TRX-925 it would look to be a good match.