Hi Thipan,
If Zeno is right about the Xbar and R charts, and autocorrelation is a problem, you may need the autocorrelation coefficient applied to the sigma estimate (1/(1-r^2)^0.5) to inflate the control limits a little bit.
Sigma(x) = MRaverage/1.128*(1/(1-r^2)^0.5)
X/MR charts are the best tool for continuos process (even as a replacement for attribute charts), that are very common on the chemical industry.
About the sensitivity of the measurement, you know your variables and the capabilities of the equipment used to weight, you could also try X/MR charts as weight or as a percentage out of specs per unit of time.
Good luck
