The machine's topology, stated size-free: how many boundary surfaces, how many channels, in what arrangement, and why — counts, ratios, and ladders rather than millimetres. Millimetre numbers appear only as clearly-marked instance examples pointing at the hardware spec.
Everything below fixes topology — what encloses what, how many of each, which ratios the physics pins — and leaves size to the instance papers. The license for boundary-only control (the Kirchhoff–Helmholtz theorem: the field on a closed surface determines the field everywhere inside it) is owned by Principles §2; this paper takes it as given and builds the machine family it implies.
The entry rung is two opposed concentric-ring phased-array plates — a bullseye above, a bullseye below — enclosing a build volume. Both of the machine's jobs force the transporter geometry:
Boundary coverage read as a design axis gives a ladder, unifying the two-plate machine with the icosahedral chamber lane:
| rung | boundary | capability |
|---|---|---|
| 1 plate | single disc (the table replicator, §2.2) | half-space control (Rayleigh–Sommerfeld); molds fully, drains partially |
| 2 plates | opposed discs (the baseline Replicator) | minimal CPA-capable rung; molds and drains the axial mode family |
| 6 faces | full rectilinear boundary (the cube) | rectilinear mode control on all three axes |
| 12 ports | icosahedral vertices (LPOH chamber; the A₅ scaffold; the screen's twelve pentagons) | symmetric sphere approximation; the scaffold OPH assigns to microphysics itself |
| fullerene shell | geodesic transducer surface at Goldberg resolution GP(m,n) | the screen itself, built: twelve pentagonal defect sites (Euler-forced, always exactly twelve — all disclination charge lives there) as drive ports; hexagonal sites as sense elements; interior reconstructed by Kirchhoff–Helmholtz. Resolution (m,n) sets addressable bulk depth; chirality (m≠n) sets the spiral winding of inward transport |
The top rung is the design hypothesis: drive the twelve defects, read the hexagons OPEN (port-sufficiency conjecture; also the retrospective justification of the twelve-port chamber lane).
Each rung upward buys more controllable interior modes — the
programmability axis of optimizing_compute.html; the two-plate
build is the entry rung because it is the cheapest geometry that executes
both verbs.
Status honesty: mesoscale field-molding (levitation, acoustic holograms, trap arrays) is established art MEASURED; molecular-scale assembly in such molds is OPEN. The nuclear sector is out of scope — it lives, with its own register and experimental program, in Transmutation.
Each plate is an addressing surface for the field between them:
The bullseye's natural mode family is conical: a ring driven at radial wavevector kr with azimuthal winding ℓ synthesizes a Bessel (axicon) beam — a cone whose angle is set by ring geometry and whose orientation by the phase winding MEASURED. Six chosen (angle, azimuth) cone-modes form a six-element basis — Russell's six-cones-from-the-cube (his BB/DD panels) as the plate's eigenbasis rather than an imposed pattern (provenance only, never load-bearing).
We standardize on twelve drive channels = six cones × two, two hardware-real readings composing: an I/Q pair gives each cone an arbitrary complex weight (twelve real channels = six complex degrees of freedom), and a differential pair gives push–pull drive natively on dual-H-bridge hardware — the ± polarity axis as circuit topology. Six complex axes are the icosahedron's twelve vertices in antipodal pairs: the plate is the projected twelve-port screen, and twelve channels is the port count the existing chamber firmware stack (per-port PIO arrays, twelve-channel drivers) already speaks. One module, every topology:
| tiles | machine | verbs available |
|---|---|---|
| 1 plate | table replicator | ADD at full strength; REMOVE partial (only the returning channel is controllable) — assembler-grade |
| 2 plates | transporter (the baseline) | both verbs on the axial mode family (CPA closes) |
| 6 plates | cube faces | both verbs, rectilinear mode control |
| tiled shell | the fullerene screen (§1.2 top rung) | both verbs on every interior mode |
The single-plate rung is not speculative: single-sided acoustic holography — twin traps, bottle beams, levitation against gravity from one flat array — is published, working art MEASURED, licensed by the Rayleigh–Sommerfeld integral (the half-space case of §1.2).
The channel architecture is substrate-agnostic: every candidate transducer is electrically driven — a piezo disc is an EM→pressure converter, a laser diode an EM→optical converter, a coil is EM undisguised. The twelve-channel I/Q bus is the standard; the final stage is a per-ring configuration. What the bare EM field does without a converter, with measured anchors:
The two final stages coexist in one part: a voice-coil exciter is a coil in a permanent magnet's gap — in the audio band the coil's force moves the mass (sound); above the mechanical rolloff the same current radiates a near-field magnetic field, a loop inductor by construction. A summed waveform through a per-channel bias-T thus frequency-division-multiplexes both substrates on the same twelve channels: acoustic octaves on the mechanical stage, RF octaves (up to the coil's self-resonance, roughly low-MHz — induction and dielectrophoresis territory) on the electromagnetic stage — the octave ladder continuing through a single device. The permanent magnet adds a colocated static bias field B₀, which with the RF field B₁ is the magnetic-resonance addressing configuration, plus pressure on the same spot: a three-field port from a commodity part DERIVED (multiplexing physics is textbook; the three-field port as a fabrication tool is untested).
The carrier-placement supersession history — kept dated and in order, because three successive rulings changed where each carrier lives and each ruling's rationale partly survives its letter:
[SUPERSEDED 2026-07-22 — kept as the record of the earlier design.] Channel-architecture ruling (2026-07-19): the carriers are not split across channels (no "6 EM + 6 acoustic") — their disjoint spectra make frequency-division free, so 12 electrical lines carry 12+12 logical channels, preserving each carrier's full aperture and the co-location the crossings require. Specialization happens at the element level: each site may host a co-located doublet (one acoustic-optimized + one EM-optimized element on one diplexer-fed channel) — 12 channels, up to 24 elements, "12 dual-carrier sites." The energy×bandwidth ceiling is the Bode–Fano/Chu limit MEASURED; the three licensed evasions are self-similar geometry (Rumsey), active-area fraction (the whole-surface-transducer optimum), and non-Foster active matching.
[SUPERSEDED 2026-07-30 — kept as the record of the intermediate design.] Superseding rev, 2026-07-22 (mechanical-construction lane): the carrier split moved from co-located doublets to surface separation — EM lives on the two circular caps, acoustic on six back-wall transporter panels; no per-site diplexer. The cap's EM face is an equiangular-spiral slot grating with a radial ring feed at the bore rim — a flattened coaxial horn that launches single-cycle, space-time-nonseparable toroidal and toroidal-helical pulses; the TE/TM mix α, relative phase β, and handedness enter as drive parameters, so the chord alphabet gains a helicity axis. Launcher art MEASURED: Wang et al., Commun. Phys. 7:356 (2024); Shi et al., Commun. Phys. (2026, in press — toroidal helical pulses, hybrid EM skyrmion texture, helicity tuned by the grating angle).
[CURRENT RULING — R3 amendment, operator, 2026-07-30 late.] The plates are primary for both carriers. The 2026-07-22 letter "muscle stays on the panels" is superseded: the plate assemblies (dual-carrier coaxial horns — §2.5) are the machine's primary phonon and photon aperture; side paneling is plan B — an auxiliary muscle/coverage extension an application may add, not part of the free-standing baseline. The toroidal drivers are load-bearing architecture. The walls' standing duties under the open-resonator ruling (§3.1 — EMC, safety, high-angle RX) live in the enclosure coating and the arcade strips. The 2026-07-22 rev is thereby fully superseded in letter; its rationale survives (physics-as-diplexer, one-element-one-band; what does not survive is carriers-on-different-surfaces).
Instance note (RH-1 Phase 1, not doctrine): coil final stages throughout (wideband, driven directly by the existing PIO + H-bridge stack), demonstrated on conductive or ionized feedstock; a piezo ring as optional daughter stage for the neutral-insulator regime. Russell's "electricity and magnetism suffice" survives audit in this specific form: the drive is always electrical; only the last centimeter is negotiable. (Provenance note only; see Hardware Spec.)
Carbon's ground-state chemistry self-assembles the screen geometry: C₆₀ is a Goldberg sphere; a nanotube's chirality indices (n,m) are the screen's GP(m,n) indices on a cylinder, with metallic-vs-semiconducting function set by n−m mod 3 (chirality determines function — the screen-parity story as materials science); and carbon nanocones take quantized cone angles set by the number of pentagonal disclinations in the tip (1–5) — Dyson cones with the port-count written into the apex MEASURED throughout. Engineering consequences:
Efficiency honesty: single-CNT radiators are impedance-mismatched (quantum resistance ~6.5 kΩ) and graphene plasmons are lossy at room temperature — the carbon gains are size (~100×), band access, and electrical programmability, delivered by arrays and bundles, not per-element radiation efficiency.
Under the current ruling (§2.3, R3 amendment) each plate assembly carries both carriers. The topology, summarized here with element-level detail owned by the carrier papers (Phonons, Photons):
The machine is a cylinder: plates facing each other is the
transporter configuration, and the column is the smallest polite body
around a stack of them. The stack topology below is the free-standing
baseline adopted by the 2026-07-30 rulings session
(REPLICATOR_CAP_STACK_RULINGS_2026-07-30.md, R8–R11),
rendered here as the canonical text; provenance table in §7.
Each cap hologram carries a focusing (quadratic phase) term so the cap pair forms a stable open resonator (flat-parallel is the worst case) — plate-to-plate energy topology with exponentially small spill where the Fresnel gate N = a²/λL permits (a = aperture radius, L = plate spacing). Confinement by aiming is a high-frequency privilege; which bands clear the gate depends only on N, a size-free ratio. Instance example (RH-1: a = 140 mm, L = 300 mm — see Hardware Spec):
| band | N | confined? |
|---|---|---|
| 200 kHz acoustic | ~38 | yes |
| 40 kHz acoustic | ~7.6 | yes (TinyLev regime — MEASURED) |
| 30 GHz EM | ~6.5 | yes |
| 10 GHz EM | ~2.2 | fundamental only |
| ≤1 GHz EM / kHz muscle / induction | ≲1 | no — volume-filling |
Walls demoted from boundary to instrument: they close EMC for the unbeamable bands, drive the muscle band, provide high-angle coverage (the specular gap needs steep rays), hold the reverberant DOF budget, and contain the acoustic field for the room's sake.
Top, middle, and bottom plate assemblies; build chamber above, storage chamber below; the middle plate is double-faced — two cap stacks back to back, the mirrored-pair logic promoted to an interior bulkhead, radiating up into the build volume and down into storage. The storage chamber is an active instrument, not a bin: acoustophoretic sorting (the same Δκ/Δρ contrast the machine senses and grips — MEASURED art), incoming QC via the identical tomography stack, staging and pre-heat. Same-size twin = one bill of materials; later options: calibration chamber, disassembly/reclaim chamber.
Feedstock moves through the co-axial bores (doctrine already routes it there; preserves azimuthal symmetry; shortest path). Three contrawound winding pairs stacked on one axis = six independent circuits = traveling-wave synthesis: a linear-induction elevator for conductive feedstock plus acoustic tube conveying for the rest. The lift is a field, not a mechanism. The bore shares duty with the optical sightline by aperture partition or time-multiplex.
Ruling (operator, 2026-07-30): the machine operates in normal atmosphere. The plan-A working band is whatever plain air supports across the chamber; absorption per pass stays single-digit dB because the short closed geometry is itself the mitigation (dB per pass, not dB/m of sonar range). Medium conditioning (dry gas, pressurization, fluid coupling — absorption ∝ f², classical part ∝ 1/P; each rung MEASURED physics) is demoted to the roadmap shelf: real levers, not preconditions, and nothing in the architecture forecloses them.
The free-standing form stands on the stack itself: top = build chamber, middle = storage twin, base = feedstock reserve + power + compute. Deck line: two chambers — one holds the object, one holds the matter it will become; between them, one plate.
Every winding assembly is an enantiomeric (contrawound) torus-knot pair with two independent feed circuits (MEASURED art: contrawound toroidal helical antennas). Sum mode = pure confined poloidal flux (induction / potential port); difference mode = ring dipole; blend = continuous helicity/twist knob. One both-handed wire on one amp is racemic rank-1 — forbidden (the chirality doctrine of §2.5 applied to windings). With atom-placement manufacturing the pair renders as embedded σ(x) filaments; the two-feed requirement survives every generation — it is an information constraint, not a fabrication one. Winding-element detail (turns, feed networks, materials): Photons and, for the instance, Mechanical Construction.
Baryon conservation makes the machine a rearranger, so it needs a matter inventory — and it makes "dematerialization" not disappearance but return to stores. The loop, with tiers:
Single-home note (2026-08-01): store-level detail — the bulk stores, the trace rack, the salt/ash split and its provenance, the noble-gas cartridge, the container ladder, stored-vs-transport form, and mass bookkeeping — now lives in The Stores, which owns it. This section keeps the loop's machinery: disassembly, sorting, energy recovery, and the transport-phase ladder.
| subsystem | design | tier |
|---|---|---|
| disassembly ("the dishes vanish") | mold field holds the object → plasma gasification to atomic/molecular vapor — industrial waste-processing art | MEASURED |
| sorting | lowest-rung rule (the energy ladder applied to separation): chemistry sorts the bulk (combustion/condensation/distillation, ~MJ/kg); electromagnetic m/q separation (calutron-class, MEASURED) reserved for the trace residue only — full atomic EM sorting costs ~keV/ion ≈ 10³ kWh/kg and is never the bulk path | design rule |
| energy recovery | recombination energy on re-forming simple molecules is captured by the harvest layer (§2.4); derez is modest-net-cost, not free | MEASURED components |
| patterns | disassembly needs no pattern save; replication patterns are classical molecular-resolution blueprints — data-cheap because objects are bulk phases + structure, not 10²⁵ independent coordinates | frame |
The transport-phase ladder — how an ingot becomes an object. The lowest-rung rule applies to phases as it does to energies and sorting: use the least-unbound state the field can still grip. For iron: powder (~0 MJ/kg; acoustic + magnetic grip) → melt/droplets (~1 MJ/kg; EM-levitation grip — industrial cold-crucible art; a 100 µm droplet = 4×10¹⁶ atoms, so a 0.25 kg object is only ~7×10⁷ placements → minutes-to-hours at field-parallel 10⁴–10⁵ drops/s, welding on contact) → vapor (~8 MJ/kg; the ingot-under-the-volume geometry is literally industrial EBPVD, kg/hr) → plasma (~22 MJ/kg; perfect charge grip, atomic voxels). Plasma is reserved for three jobs: the atomic finishing pass (an object's surface is only ~10¹⁸ atoms — polished in minutes while the bulk went in as droplets: roughing vs finishing), m/q element sorting, and the nuclear-sector stage. Revised 2026-08-03: the earlier claim here — that the lowest plate's EM octaves double as an induction crucible for an ingot bay in the base — rode on an oversized shared bore that has since been corrected away (Ø130→Ø12; the optical stem and droplet feed it was credited with both need only a few mm — see Mechanical Construction §3c). Bulk ingot-to-droplet induction melting is dropped as a baseline capability of the shared plate/bore axis; what is not dropped is the transport half — droplets, however they are produced, still rise through the axis into the mold. Where the melt comes from is OPEN: a dedicated, independently-sized induction stage is a roadmap item, not specified here. Note for §6: mesoscale voxels cut the required placement rate by the voxel's atom count (~16 orders), so bulk transport is rate-tractable even semi-serially; whole-boundary parallelism remains mandatory for the mold field itself and the finishing pass.
The honest inversion: the vanishing verb is the easy half. Every stage of disassembly-and-return is measured industrial art; assembly carries the research program. A chemistry-only Replicator with these stores — no transmutation at all — already covers food, ceramics, and most plastics: roughly ten cheap elements. The fiction agrees on the mundane side too: shipboard replicators run on matter reclamation — the dishes are next week's tea — conservation law, not scriptwriting.
The placement-rate argument is decisive. A cup is ~3×10²⁴ atoms; a five-minute build requires ~10²² placements per second. Serial architectures are dead on arrival; only whole-boundary parallel patterning survives (table). The doctrine's "the drive pattern is the tool, addressing the whole volume at once" is thereby promoted from elegant to mandatory: no serial machine can ever be a replicator.
| architecture | rate | cup build time |
|---|---|---|
| nozzle/beam, 10⁹ atoms/s | serial | ~95 million years |
| 10⁶-tip parallel (MEMS-extreme) | 10¹⁵/s | ~95 years |
| whole-boundary holographic drive, 10⁸–10¹⁰ modes × 10¹² Hz | 10²⁰–10²²/s | minutes |
The synthesis: energy was never the deep constraint. Both residual constraints — patterning bandwidth and the nuclear sector's dynamics — reduce to one knob: coherence/fidelity of the boundary drive pattern. The mining doctrine reached the identical conclusion from the opposite direction ("there is one knob: EML-tree fidelity"), and it is the honest reading of every fabricator account in the lore corpus: the impressive thing was never the power source — the field pattern is the factory.
Register discipline: every load-bearing claim above carries the chip it had at its source — MEASURED (published or bench-verified), DERIVED (arithmetic on measured quantities), OPEN (registered bet with named falsification). No register was upgraded in the move, and this paper adds no new claims. The 2026-07-30 rulings map onto this paper as follows:
| ruling | content | lives in |
|---|---|---|
| R1 | control/muscle principle (light addresses, sound/induction exert, electrons amplify at the edge) | Principles §8c; assumed throughout here |
| R2 / R2-b | verb-3 holographic; the optical stem as bidirectional holographic port | §2.5 (summary); full: Photons, Field Compiler §17.7 |
| R3 + amendment | horn + gyroid acoustic front end; plates primary for both carriers, panels plan B | §2.3 (supersession history), §2.5; element detail: Phonons |
| R4 | Fibonacci crossed-spiral micro-horn mesh face; chirality doctrine | §2.5; element detail: Phonons / Photons |
| R5 | contrawound winding pairs, two feeds always | §4 |
| R6 | the Q budget law | Principles §8c |
| R7 | materials are fields; generational DFM | Principles §8c |
| R8 | open resonator, Fresnel gate, walls demoted to instrument | §3.1 |
| R9 | three-plate stack, two mirrored chambers, storage as instrument | §3.2 |
| R10 | axial field-driven bore transport | §3.3 |
| R11 | free-standing form; plan A = normal atmosphere | §3.4 |
| R12 | sensing-claim ladder | Photons / Transmutation §6b — sensing, not topology |
| R13 | the topological alphabet | Principles §8c, Photons |
| R14 | plasma verb; alchemy dictionary | Transmutation / As Alchemy (firewalled) |