2026-06-28
Compact V1 corner print
The active corner design moved to a 45 mm sphere with 14.5 mm strut sockets, 13.8 mm wire throats, and an 18 mm rear relief. The DX06 magnet pocket was deferred to avoid enlarging V1 or adding magnetic ambiguity.
Log
The build, dated - assumptions still attached.
The public progress stream for K4: device state, solver visibility, and the pieces ready to be shown without overstating what the hardware has proven.
2026-06-22
Full-edge drive clarified
The hardware target is all six edges active for the full model. Four H-bridges remain a useful interim for steering tests, but all-edge operation requires six independent drive channels.
2026-06-19
First-edge bench expectations
The N=6 first-edge predictor set a gaussmeter target for bring-up: for a 100 mm edge with six straight AWG22 strands, ~13.86 uT at the centroid-distance probe at 0.50 A/strand, excluding return-lead contamination. The sidekick captured the lead-cancellation trap that can make a correct edge look broken.
2026-06-19
Live field engine in the browser
The cockpit became a live field instrument rather than static diagrams: slice planes, streamlines, and build parameters can be checked against the same engine used for the bench predictions.
2026-06-14
Levitation synthesis
K4 was reframed as a compact graph-topological electromagnetic control manifold, not a new levitation force. The practical demo path is permanent magnet plus active feedback first.
2026-06-13
Wire crystal
Structured conductor fibers became the second layer: common-mode multi-wire edges preserve the finite-wire solver, while non-common routing, shells, and chirality become future control handles that must be measured before they are claimed.
2026-06
Inverse solver and browser cockpit
The inverse-solver path replaced the old atlas-first interface. The browser cockpit exposes the field model directly and keeps claim class, model scope, and build assumptions visible.