Abstract: Conservation tillage has been one of the most modern agricultural technologies in recent years. Stalk returning and no-tillage seeding have been combined to improve the soil's physical and chemical properties. The soil fertility can also be protected to reduce the soil disturbance for the stalk coverage and crop yield. At the same time, the long-term high-intensity tillage and stalk burning have caused serious erosion with the limited content of soil organic matter in Northeast China. It is highly required for the conservation tillage with the high level of agricultural mechanization. However, the existing anti-blocking device for no-tillage seeders cannot fully meet the high requirement of high-speed no-tillage seeding. In this study, a stalk cleaning device was designed to combine the seeding belts, particularly for high-speed no-tillage seeding. The double-row planting mode of large ridges was also selected in Northeast China. A pair of drive discs and stalk cleaning teeth were composed in the combined device. Among them, the profile of the disc blade was utilized as the normalized curve of variable parameter logarithmic spiral. The dynamic sliding cutting was then realized under the curve operation. The stable angle of the interception curve was achieved at 22.5° during dynamic sliding cutting. The curve number of the blade disc was further determined to be 18 using kinematic analysis. The two sides of each disc were equipped with the stalk cleaning teeth. The stalks were fully cleaned inside or outside the ridge. A systematic analysis was implemented to optimize the structural, spatial, and operation parameters of the stalk cleaning teeth. A series of simulation tests were also carried out to clarify the interaction between the device and the stalk under high-speed operation using the discrete element method. The ridge soil model was established using the Hertz-Mindlin and Bonding models. The flexible body of the maize stalk was designed with lengths of 130, 150, and 170 mm, respectively. The positioning and laying of the stalk were set at different angles. The stalk cleaning performance was evaluated under different declination angles of the device. The range of deflection angles was determined to be 30°-75°. An optimal model of ridge soil was constructed with a stalk covering amount of 1.66 kg/m². The quadratic rotation-regression-orthogonal experiment was carried out using Design-Expert software. The target variables were taken as the operation speed and inner and outer declination angle. The optimal parameters of the device were obtained: the inner and outer deflection angles of the stalk cleaning teeth were 45.3° and 75.0°, respectively. The average working power was 7.653 kW, and the stalk cleaning rate was 95.00%, where the relative errors were 3.06% and 1.35%, respectively. The performance of the device was also verified under different amounts of stalk covering. The stalk covering amounts of 2.06, 1.59, and 0.87 kg/m² were selected in the field experiments, corresponding to the small, medium, and large stalk covering. The field experiments showed that the ideal performance of the device was achieved in the stalk covering amount of 0.87 kg/m² at the operating speed of 3.0 m/s. Specifically, the stalk cleaning rate of the anti-blocking device was 96.1%, and the average operating power was 8.219 kW. Once the stalk covering amounts were 2.06 and 1.59 kg/m², respectively, the stalk cleaning rates of the anti-blocking device were 86.7% and 92.2%, while the average values of operating power were 10.460 and 9.024 kW, respectively. The optimal device has fully met the needs of no-tillage seeding under different stalk covering rates in Northeast China. The finding can also provide a theoretical and technical reference for the anti-blocking devices under high-speed seeding.