Abstract: This study addressed the trade-off between feeding speed and endpoint accuracy in screw-fed batch weighing of rabbit pellet feed. Discharged pellets cannot be retrieved, making endpoint overshoot difficult to correct. A segmented proportional-integral-derivative (PID) controller with integral separation was developed to regulate feeding throughout the weighing cycle. The controller switched parameter sets according to the difference between the target feed mass and the real-time weighing value, and enabled integral action only when the error entered a prescribed small-error range. The study evaluated both the control scheme and the conversion from controller output to pulse frequency, because their combined effects determined the feeding response. Two selection stages were conducted at a target feed mass of 200 g. First, six schemes representing conventional PID, segmented PID, integral-separation PID, and segmented integral-separation PID were compared using auger pitches of 80, 90, and 100 mm. Mean absolute error, coefficient of variation, maximum deviation, and mean weighing time were included in entropy-weighted scoring to identify a preferred scheme for each pitch. The three selected pitch–scheme combinations were then tested with five frequency division coefficients, producing 15 system configurations. These configurations were evaluated using mean absolute error, coefficient of variation, and mean weighing time. The highest-scoring configuration was subsequently tested with three rabbit pellet feeds at five target feed masses from 180 to 390 g. A separate static test assessed the weighing subsystem at tilt angles from 0° to 5°. The preferred scheme varied with auger pitch. Segmented integral-separation PID with a threshold coefficient of 0.10 ranked first for the 80 mm auger, whereas integral-separation PID with the same coefficient ranked first for the 90 mm auger. Segmented integral-separation PID with a coefficient of 0.05 ranked first for the 100 mm auger. Compared with conventional PID at the same pitch, the three selected schemes reduced mean absolute error by 60.3%-87.4%, coefficient of variation by 44.5%-72.0%, and maximum deviation by 53.2%-77.8%. Mean weighing time increased by 44.7%-76.5%, indicating an accuracy–time trade-off rather than an improvement in all performance indices. All three selected schemes used integral separation, while two also used parameter segmentation. In the second stage, the 100 mm auger with segmented integral-separation PID at a threshold coefficient of 0.05 and a frequency division coefficient of 1 received the highest score. Its mean absolute error, coefficient of variation, and mean weighing time at 200 g were 0.92 g, 0.49%, and 8.59 s, respectively. The preferred frequency division coefficient was 1 for the 80 mm combination and 0.75 for the 90 mm combination. Reducing the coefficient from 1 to 0.75 for the selected 100 mm combination shortened mean weighing time from 8.59 to 6.18 s but increased mean absolute error from 0.92 to 8.05 g. With controller parameters and the coefficient fixed, validation across the three feeds and five target masses yielded an overall mean relative error of 0.94%, a mean coefficient of variation of 1.09%, and a mean weighing time of 9.08 s. Mean relative errors across individual conditions ranged from 0.38% to 1.87%. The maximum relative deviation from the horizontal baseline in the static tilt test was 3.10%, below the prescribed 5% engineering limit. These results supported joint selection of the control scheme and frequency division coefficient for a given auger pitch. The selected configuration provided a practical balance between weighing error and cycle time within the tested feeds and target masses. Further work could combine discharge-dynamics measurements with threshold-decoupling tests and extend validation to additional feeds and target masses.