The paper reveals a novel model (structure) for realizing a tunable digital filter (TDF) having tunable frequency bandwidths. This model is comprised of cascaded low-order sections that contain 2nd-order denominators, while their numerators can be higher orders. Since this cascade model consists of only low-order sections, the low-order sections can be utilized as fundamental building elements for implementing the entire cascade model. The cascade structure also features low implementation noise induced from coefficient quantization. Since the low-order sections have 2nd-order denominators, it is easy to guarantee the stability by exploiting an explicit stability condition. Based on the explicit stability condition, the paper also reveals a stability-ensuring scheme, which uses a uniquely defined function to represent the denominator parameters of the low-order sections. Representing the denominator parameters as the functions enables the stability condition to be definitely satisfied, and thus keeps the whole cascade model stable during bandwidth tuning. This paper includes the simulation results of designing a highpass bandwidth-tunable filter for exemplifying the achieved high accuracy as well as the ensured stability. The simulations verify that exploiting the novel cascade model can obtain a highly accurate TDF with tunable highpass responses. The design results also demonstrate that the highpass TDF remains stable during frequency-bandwidth tuning.