S4 did not reward release authority above 4.569852 m/s in its best tested high-authority campaigns. For the 4 kg reference that is 41.77 J, so the next machine is being treated as a compact programmable release cell rather than an inherited eight-metre launcher.
A slow actuator charges stored mechanical energy before release. Preload sets the commanded energy. An independent latch releases a short guided pusher, and a local catcher keeps that pusher with the host. Shared power, command and navigation remain common-mode services.
The functional picture is intentionally one level shallower than CAD. Spring form, latch geometry, guide, catcher, motor and structure have not earned detailed geometry yet.
Freeze coupon bands for accumulator force-displacement, preload repeatability, latch shock, pusher friction, exit velocity/tip-off and catcher load before detailed CAD.
Compare independent cells, banks and the shared path using structure, actuators, controls, energy, envelope and failure consequences per successfully delivered payload.
Add release/navigation uncertainty, clearing/settling time, host attitude recovery and complete-manifest effects where S4 deliberately stopped.
The reference survives only if installed burden and release-cell evidence survive their declared bands. A failed band changes the architecture instead of widening the band.
Start from the 4 kg 3U reference, then test whether the architecture survives adjacent payload classes instead of enlarging CAD.
Drawings, BOM, assembly/inspection, instrumentation, calibration, uncertainty and frozen acceptance criteria before physical testing.