With a growing interest in outer space, space robots have become a focus of exploration. To coordinate them for unmanned space exploration, we propose to use the "mother-daughter structure". In this setup, the mother spacecraft orbits the planet, while daughter probes are distributed across the surface. The mother spacecraft senses the environment, computes control commands and distributes them to daughter probes to take actions. They synergistically form sensing-communication-computing-control ($\mathbf{SC^3}$) loops, which are indivisible. We thereby optimize the spacecraft-probe downlink within $\mathbf{SC^3}$ loops to minimize the sum linear quadratic regulator (LQR) cost. The optimization variables are block length and transmit power. On account of the cycle time constraint, the spacecraft-probe downlink operates in the finite block length (FBL) regime. To solve the nonlinear mixed-integer problem, we first identify the optimal block length and then transform the power allocation problem into a tractable convex one. Additionally, we derive the approximate closed-form solutions for the proposed scheme and also for the max-sum rate scheme and max-min rate scheme. On this basis, we reveal their different power allocation principles. Moreover, we find that for time-insensitive control tasks, the proposed scheme demonstrates equivalence to the max-min rate scheme. These findings are verified through simulations.