Kei, I thought about the same, but there's WAY too many variables to learn and too sophisticated dependencies to calculate to be able to do it all accurately by means of calculation. V. must forget the idea of measuring precise time of the pulse. This is unachievable. But what IS achievable - a certain input (given weight) generates certain unknown but unique pulse. Extract one characteristic unique measurable property of the signal you can (just like complete static load value in coulombs, carried by the signal, modified by unknown but fixed distortions, leaking unknown but fixed amount out of the wires, getting mangled by some unknown but fixed way.) Then measure the property and write the input-output pair into a table. You know that 1kg 387g and 12mg gives readout of 10.777.216 on the ADC and you don't really care what strangely mangled signal in electrical terms does the ADC value correspond to. You just experimentally measured and you know that it means that precisely given weight, that within specified range of environment conditions the value won't differ by more than 3 bits (or differs in some predictable and correctable way) and that no other weight will produce the same value. And that's all you care about. You don't care that the pulses differ by 0.12ns. You just care that they differ.