This study involved the development and evaluation of a new filtering technique, as based on adaptive free mass load cell. The technique's application in reduction of vibration and impulse noise (commonly added to the sugar cane harvester yield monitoring signals) was tested. This technique is based upon assimilate spring- mass vibration system of weighing scale with a free mass load cell in its natural frequency and phase. This was achieved through a calculation of free mass. Finally, the scale signals were corrected online through subtracting the free mass load cell signals. Experiments were conducted in a 3 factor randomized complete block design. In a first test of 3 replications, the effects of 6 frequency levels (3, 4, 5, 6, 6.5 and 8 Hz) on vibration table and 3 speed levels (2.2, 2.4 and 2.7 m/s) on elevator simulator were tested on standard division of scale output signals before and after corrections at 9 levels (the first with no correction and 8 correction levels of 2, 2.1, 2.2, 2.3, 2.4, 2.5, 2.6, and 2.7 coefficients). In a second test of 4 replications, the effects of 5 different crop weights of 0, 1.5, 3, 4.5 and 6 Kg were tested. The results indicated that in a wide range of vibration frequencies, different elevator chain speeds and different excess weights on the scale as well as for different impulse strokes, the differences were significant at 1% after and prior to correction and while after correction being made, the standard deviation of scale signals did not exceed 0.4 Kg. either the best correction coefficient was obtained 2.1 for the case of before loading and 2.4 for the case of when crops passing. An important advantage of this technique is its good performance at low frequencies where analog or digital filters can not be employed due to their possibility of eliminating main signals.