One of the basic problems in highly automated industrial production line is to assess the weight of the produced items in motion. The devices that do this work are called conveyor belt type checkweighers and their speed is an essential factor for production lines. The challenging technical difficulty is to measure mass with required accuracy shorter than the settling time of the weighing system. Furthermore, production speed is a key issue in modern production lines therefore advanced signal processing techniques were needed to handle these challenging difficulties. In this work, a time-variant infinite impulse response (IIR) filter was designed and implemented on an embedded system integrated into the industrial checkweigher system. The parameters of the designed filter were variable over time such that the bandwidth of the low pass filter changes from wide to narrow passhand. Optimum filter parameters which give the weighing result in desired error range in the fastest way were found through brute-force search. The time-variant filter was compared to different types of classical and time-invariant filters. It is concluded that the time-variant filter outperformed to other filters.