For two weightings it's quite simple:
1. Take 1 ball from every machine, `ActualWeight - 10gr*10` is the difference for the faulty machine
2. Using information from #1 you have the original task which is solvable with 1 weighting
I believe 1 weighting is not sufficient and here is why: if you take n1,n2,n3,...,n10 balls from corresponding machines, it's equally possible that 1st machine is faulty with weight difference 1/n1 and that 2nd machine is faulty with weight difference 1/n2 - these cases don't seem to be distinguishable...