I think you are looking at this the wrong way: You are comparing the reading of one device to another, both with different input impedance, at an extremely low voltage level (100uV). At this voltage level, small things like current leakage, bias/input currents, noise, input offsets, etc. play a significant role. Reviewing your schematics, you do not seem to have any buffering/gain stage for the signal coming from the load-cell. Furthermore, there is no filtering. Instead of approaching it this way, you should probably look at it from what is important in your design, linearity. You should probably "zero" the readings at your circuit and then place a known weight(s) to find out how linear your circuit is. All other problems can be resolved with calibration.
↧