RUBISCO2 V2 · pre-fab gate · 2026-05-18 · build v7 (inline current CAD)
Two counter-rotating eccentric motors on coil springs is a valid industrial pattern — that's Blekhman self-synchronization, the principle behind thousands of vibratory feeders since the 1950s. The motors phase-lock through the body's own vibration, and the resultant force is linear along the tray axis. The architecture isn't broken. What's missing is an explicit motion contract under the transient slug load case: when 1.5 t lands at one end of the 2.5 m tray, the center-of-mass shifts 200-300 mm, the resultant force line stops crossing CoM, and the four near-isotropic coil springs don't strongly oppose the resulting yaw mode. The real failure shape isn't "it won't work" — it's "ugly transient response that can drift the trompa, walk the feeder, and eat the spring fatigue budget."
Hardened plates on the sub-frame ride against UHMW shoes mounted to the static support frame at the 4 corners, with a 1-2 mm cold gap. They do not touch during clean steady-state operation — they engage only when transients try to grow yaw or drift. Smallest possible change that preserves the locked spring stack, motor cradle, sub-frame tubing, trompa, and height envelope. Pedro fab unaffected.
Why the architecture works in principle, and exactly where it stops working in practice.
Two counter-rotating eccentric motors mounted on a common rigid body lock into anti-phase on their own through the body's vibration — they don't need a mechanical coupling. Discovered by I.I. Blekhman in the 1950s and exploited in every twin-motor industrial vibratory feeder since (Cleveland Vibrator EMF, Hindon, OLI's own installations).
Means: the "two motors will fight each other randomly" failure mode doesn't happen. They co-operate by themselves.
The resultant force is linear if its line of action passes through (or very near) the body's center of mass, if the four corner spring stiffnesses are roughly symmetric (±10% typical), and if the body is stiff enough to act rigid. Under steady-state operation with material spread along the bin, all three hold.
Means: when material is distributed, the stroke is clean. The principle is checkable, not aspirational.
When the loader dumps 1.5 t at one end of the 2.5 m tray, CoM shifts by 200-300 mm along X. The resultant force line no longer passes through CoM → yaw moment on the body. Coil springs are nearly isotropic laterally, so nothing strongly opposes the yaw mode. The body twists on its springs until the slug spreads out.
Means: the worry isn't paranoia. It's a real transient response.
A pure linear stroke has zero net displacement per cycle. Any yaw oscillation superimposed on the stroke creates asymmetric contact at the leveling feet → ratcheting → the feeder slowly walks across the deck. Bounded over time by the yaw amplitude itself. Without explicit lateral guides, walking on the order of mm/hour is plausible.
Means: even without "chaos," long-running drift is real and worth bounding.
Scoped for the V2 schedule: Pedro fab next week, demo 2026-06-09. Anything that pushes fab past this week is flagged for V2.1 or later.
Ship as currently CAD-locked and hope steady-state behavior dominates.
4 brackets on the static support frame; each holds a UHMW shoe with a 1-2 mm cold gap to a hardened plate on the sub-frame.
Replace the 4 coil packs with 4 inclined leaf-spring stacks. Direction comes from the suspension itself.
One fixed motor → V-belt → horizontal shaft with eccentric cams → pitman to bin. Eriez / Brunette / Carrier pattern.
| Option | Time to fab-ready CAD | Controls slug-load yaw / drift? | Disturbs locked V2 package? | Verdict |
|---|---|---|---|---|
| A · Current alone | 0 h | No | No | Not a release state. |
| B · Bolt-on guide cartridges | 3-6 h | Yes, with clearance tuning | Minimal — brackets only | V2 release fix. |
| C · Flexure suspension | 1-2 hard passes | Yes, inherently | Spring stack + height + supplier | Best long-term. V2.1 retrofit. |
| D · External eccentric drive | Full drive redesign | Yes, by construction | Reopens drive + guards + height | V3. |