The Franklin Construct

Developed by: Christopher J. Perger, Volunteer Steward & Founder of Liberty Shore

(Mark II Revisioning) Released on: February 23rd, 2026

THE FRANKLIN CONSTRUCT MARK II

~ COMPLETE ENGINEERING DOCUMENT ~

Ver 2.1
February 21st, 2026

0. Document Spine (How to Read This Document / How Claims Work)

0.1 Measured-First Doctrine

All performance claims in this document are categorized using three labels:

Target (T):
Design intent prior to validation. These are engineering goals based on analysis, simulation, or informed estimation. They carry no implication of achievement.

Measured (M):
Verified under logged test conditions. A value only earns “Measured” status when it has been recorded with complete metadata (surface type, mist state, ambient conditions, thermal state, battery state), is repeatable across a minimum number of runs (defined per gate), and was collected without any stop-rule triggers during the measurement interval.

Derived (D):
Calculated from measured inputs using stated assumptions. Derived values inherit the confidence level of their weakest input. For example, range derived from measured usable kWh and measured Wh/km is labeled D, and the supporting measured values and their Run IDs must be cited.

The governing rule: No Target value is presented as achieved until verified under logged test conditions. No Derived value is presented without explicit citation of its measured inputs and the assumptions connecting them.

This doctrine applies to every number in Sections 2–6, every row in the Performance Metrics table, every claim in the Claims Register (Appendix H), and every gate artifact in the Evidence Index (Appendix M).

0.2 Non-Negotiable Early Reality Checks

Before any scaling, range projection, or rider-scale discussion is meaningful, Phase 1 must be falsifiable. The following pass/fail criteria are defined as non-negotiable prerequisites for advancing beyond earliest-stage prototype work:

Phase 1 Success Criteria (Non-Negotiable) — Rev A

• Achieve 50 N lift at 10 mm gap over deposited layer for 60 seconds (static hold, control off)
• Keep coil/driver temps under 15 degrees C rise over the hold window
• Maintain lateral stability within +/- 2 mm without runaway oscillation (closed-loop, Gate 2)
• Demonstrate a repeatable force-vs-gap curve over at least 3 representative surfaces with variance <= 10%
• Maintain closed-loop stability at a fixed gap for >= 60 s with estimator confidence Conf >= 0.70
• Characterize mist track properties: sheet resistance (ohm/sq), uniformity (<= 25% variance), persistence window (>= 60 s), and variability vs surface type

These criteria convert the timeline into a pass/fail gate. Without them, the document is a manifesto. With them, it is a testable engineering plan.

“All performance claims are categorized as Target (T), Measured (M), or Derived (D). No Target value is presented as achieved until verified under logged test conditions.”

0.3 Rev A Targets Rationale

These thresholds are conservative initial targets intended to validate measurable force, stable gap control, and safe fault behavior at module scale. Values will be revised after the first 10 logged runs.

Rev A numbers are chosen to be large enough to prove real physics (not noise or magnetic “tingle”) but not so large that they implicitly claim rider-scale performance. They exist to make the earliest gates falsifiable and to establish the measurement infrastructure that all future claims depend on. Treat Rev A as a starting calibration, not as a specification.

1. Executive Summary (Credible, Bounded Claims)

1.1 What It Is

The Franklin Construct is a modular personal mobility platform that targets controlled, low-altitude hovering and assisted propulsion without fixed infrastructure. It is not a maglev train derivative, not a ducted-fan aircraft, and not a novelty toy. It is an electromagnetic suspension and propulsion system designed to operate over unprepared surfaces using a combination of passive and active magnetic strategies, with an optional surface-conditioning layer to improve coupling repeatability.

1.2 Core Architecture Statement

The current design is a hybrid electromagnetic architecture:

• Passive permanent-magnet arrays provide a strong baseline magnetic field (B-bias) without continuous power draw. This reduces the burden on active systems.
• Actively driven resonant coils provide stabilization, thrust vectoring, braking, and dynamic gap control. These coils are driven by a high-efficiency inverter using a PLL-tracked, ZVS-capable topology.

The passive field establishes a foundation. The active system rides on top of it — modulating, correcting, vectoring, and braking. Neither component alone claims to provide full rider-scale lift; together, they form a force budget that is validated incrementally.

1.3 “Mist Track” Framing (No Magic)

A temporary surface-conditioning layer (“mist track”) is used to improve repeatability of magnetic/electromagnetic coupling on arbitrary terrain. This layer is not assumed to behave as an ideal conductor or ideal magnetic track. Instead, it is treated as a controllable, time-limited surface modifier whose properties are explicitly measured and used by the control system to adapt lift and stability in real time.

Key properties that must be characterized:

• Sheet resistance (Ohm/sq): measured via 4-point probe or calibrated proxy
• Uniformity: measured across a grid (>= 9 points over defined area), reported as mean, standard deviation, and percent variance
• Persistence window: time interval after deposition during which film properties remain within defined thresholds, measured under stated ambient conditions and surface type

The mist track is not assumed to “biodegrade into nothing.” The carrier fluid may be biodegradable; particulate fate and exposure controls are addressed in the safety plan (Section 7) and the Ethics & Environmental Handling Plan (Appendix G). Deposition is engineered to minimize airborne risk.

1.4 Program Structure

The project is structured around staged validation. The first technical milestone is not “full rider hover.” It is:

• A measured and repeatable force-vs-gap curve over representative surfaces
• Defined thermal limits observed and logged
• Stable closed-loop gap control demonstrated on a rig

Once these are validated, the design scales by increasing active area, optimizing magnetic circuit geometry, and improving surface conditioning uniformity. Each scaling step is itself a gated milestone with pass/fail criteria and required artifacts.

This structure means the document reads as a validation program, not a product announcement. Every section that follows is designed to support this approach: claims are conditional, numbers are tied to test conditions, and progress is earned through logged evidence.

2. Hybrid Lift and Propulsion Architecture

2.1 Passive Magnetic Bias (Halbach / Permanent Magnets)

Permanent magnets provide a strong baseline field (B-bias) without continuous power draw. This bias field reduces the burden on active coils, enabling the active system to focus on modulation — stability, vectoring, braking — rather than generating the entire magnetic field from scratch.

The passive array is designed using Halbach-style geometry to concentrate field strength on the working side and reduce stray field on the rider side. The specific contribution of the passive array to net lift force is not stated as a fixed universal number. Instead, it is expressed as baseline field (B-bias) vs distance, measured on the static rig, and net force contribution vs gap, measured under defined surface and mist conditions.

Baseline magnetic bias is intended to reduce active power demand and improve stability margin. Its net force contribution is measured as part of the force-vs-gap validation program and is expected to increase with array area and reduced gap.

This framing eliminates the force inconsistency flagged in earlier reviews. The passive contribution is real and important, but its numeric value is only meaningful when paired with gap distance, interacting substrate properties, and active area.

2.2 Active Resonant Coil Modulation (High-Efficiency Power Electronics)

Resonant coils driven by a high-efficiency inverter (PLL-tracked, ZVS-capable topology) provide time-varying magnetic fields for:

• Gap stabilization (active suspension behavior): The coils maintain commanded gap by modulating field strength in response to estimator feedback.
• Thrust and lateral vectoring: By introducing controlled phase and amplitude gradients across the coil array, the system generates lateral forces for translation and directional control.
• Magnetic braking and damping: The coils can operate in a damping mode that converts kinetic energy into electrical energy where feasible. Braking authority depends on coupling quality, speed, and surface/mist state.

Mode transitions are governed by the supervisory state machine (detailed in Appendix J) and are subject to safety constraints including thermal limits, estimator confidence, and coupling quality.

2.3 Force Budget and Scaling Statement (Critical)

Hovering a rider-scale mass requires roughly 1,000–1,500 N of upward force depending on payload. For a 150 kg total system mass, the gravitational force is approximately 1,470 N. The Construct therefore treats lift as a force budget problem, not a single “magic component” problem.

Important: Early prototypes will not claim “150 kg hover” until the force budget is validated at smaller scale. All lift contributions are categorized as either validated (measured) or target (to be measured).

Force Budget Template (Targets vs Measured):

2.4 Mist Track as a Controlled Surface Modifier

The mist track is treated as a time-limited surface conditioner that may improve uniformity of electromagnetic coupling, repeatability of force-vs-gap behavior, and braking authority and lateral stability. It is not assumed to be a permanent installation, an ideal conductor, or a frictionless track.

Mist characterization is a required part of Gate 1 testing (if mist is enabled) and includes sheet resistance measurement, uniformity mapping, and persistence window determination. Results feed directly into the estimator’s Cq calculation and into the force budget’s condition metadata.

2.5 Phase 1 Non-Negotiable Validation Criteria

• Demonstrate a repeatable force-vs-gap curve over at least 3 representative surfaces with variance <= 10% (Rev A)
• Achieve >= 50 N lift at g = 10 mm for >= 60 s (static, control off) — Rev A
• Maintain closed-loop stability at a fixed gap for >= 60 s with Conf >= 0.70 and no runaway oscillation
• Keep coil + driver temp rise within <= 15 degrees C over the hold window — Rev A
• Characterize mist track: sheet resistance, uniformity (<= 25% variance — Rev A), persistence window (>= 60 s — Rev A)

These criteria are the “truth gate.” If the force is not there, the architecture is revised before scaling. No amount of software or enthusiasm substitutes for measured newtons at measured gap.

3. Power System and Energy Budget

3.1 Definitions

Power is tracked in four buckets:

• Active field power (P_field) — real electrical power dissipated in coils/drivers
• Control and compute (P_ctrl) — MCU/FPGA, sensors, comms
• Mist conditioning (P_mist) — pumps, valves, ultrasonic conditioning
• Overhead (P_aux) — cooling fans/pumps, BMS, contactors, precharge

Total:
P_total = P_field + P_ctrl + P_mist + P_aux

3.2 Power Numbers Are Conditional

Instead of stating a single hover power, the Construct reports power as a function:

P_total(g, F, surface, mist_state, thermal_state)

“Hover = X watts” is never used as a standalone statement. Every power number carries its conditions or it carries a TBD with a note explaining what test will produce it.

3.3 Targets vs Measured Values

Design Targets (Pre-Validation):
Idle: TBD W;
Stabilized gap-hold at small scale: TBD W at defined F and g;
Translation: TBD W above hover-hold;
Peak transient: TBD W for < ___ s.
All promoted to Measured only after bench/rig validation with logged conditions.

3.4 Thermal Budget and Derating

• Level 0 (Nominal): Full control authority. All modes permitted.
• Level 1 (Warm — Rev A: 70C): Limit peak current and aggressive maneuvers. System logs transition and continues with reduced authority.
• Level 2 (Hot — Rev A: 85C): Reduce gap target, reduce translation mode, prioritize stability. Operator alerted.
• Level 3 (Critical — Rev A: 95C): Controlled descent to skids/wheels. Coils disabled after touchdown.

3.5 Energy Recovery Claims

Regenerative braking is treated as an efficiency improvement, not a primary power source. Any regen claim must be stated as recovered energy per braking event (J or Wh) under test conditions, net benefit after conversion losses, and measured vs target.

Regen is reported as Wh recovered per event and net benefit after losses. No “self-sustaining” claims, no perpetual-motion implications, and no vague efficiency percentages without defined test conditions.

3.6 Power Budget Table (Targets vs Measured)

4. Gap Sensing and Height Estimation

4.1 Requirements

Gap sensing is designed to meet these requirements (values are targets to be validated):

• Update rate: >= 200 Hz fused estimate (higher for inner loops)
• Resolution: <= 1 mm at operational gap range
• Latency: <= 10 ms end-to-end for stability loop use
• Robustness: Tolerate mist aerosols, dust, textured asphalt, and vibration

4.2 Candidate Sensing Modalities (Use at Least Two)

• 1. ToF Optical Ranging (IR/LiDAR): Direct distance measurement, high resolution, fast update rate. Degraded by mist/dust/specular surfaces/sunlight.
• 2. Ultrasonic Ranging: Works in darkness; unaffected by optical interference. Affected by wind, temperature gradients, rough surfaces. Address ultrasound cross-talk with frequency separation.
• 3. Capacitive Proximity Sensing: Fast, compact, strong for short range. Depends on substrate dielectric properties. Must be calibrated per surface type.
• 4. Field-Based Inference (Auxiliary): Uses coil current/voltage, phase, and Hall readings. No additional sensors required, but ambiguous under changing conditions. Cannot serve as primary gap measurement.

Minimum credible configuration: One direct ranging method (ToF or ultrasonic) plus one secondary method (capacitive or field-inference) to maintain partial functionality when a sensor is blinded.

4.3 Sensor Fusion and State Estimation

The Construct fuses IMU attitude, ranging, and electromagnetic state into a single estimate of: gap (g), vertical velocity (dg/dt), roll/pitch/yaw, and coupling quality metric (Cq). Fusion via complementary filter or EKF — choice driven by Gate 2 testing results.

The estimator outputs both an estimate and a confidence score (Conf). When Conf drops below threshold, the system transitions to conservative stability mode and prepares controlled descent. This is a hard rule, not a suggestion.

4.4 Degraded Modes and Fail-Safe Behavior

• Mode A (Single-Sensor Degraded): Fused estimate remains within confidence bounds. Reduce acceleration and translation; widen stability margins. Log and continue with reduced authority.
• Mode B (Distance Estimate Unreliable): Conf below operating threshold. Transition to controlled descent on skids/wheels. Disable mist output. This is a “get down now” mode.
• Mode C (Power Fault): Immediate descent to skids/wheels. Coils de-energized after touchdown. Mechanical stop is primary safety.

4.5 Validation Tests for Gap Sensing

• Static height accuracy over 5 surface types: concrete, asphalt, sealed wood, bare metal reference, plus one additional surface.
• Dynamic response test: Step change in commanded gap. Measure sensor response time, estimator tracking lag, and overshoot.
• Mist-on vs mist-off comparison: Quantify effect on each modality. Record whether any sensor becomes unreliable.
• Sunlight, dust, and aerosol interference tests: Verify sensor performance or graceful degradation.
• Cross-talk test if both ultrasound ranging and ultrasonic conditioning are present.

4.6 What We Will Measure Before We Claim

• Force vs gap: The fundamental relationship.
• Power vs force: Electrical power required to produce a given lift force.
• Thermal rise vs time: How fast components heat up.
• Gap accuracy vs environment: Sensor suite performance across surfaces, mist states, environments.
• Stability margin under disturbances: How much the system can tolerate before it must land.

5. Power and Range (Measured-First)

5.1 Reporting Standard

All power and range values are reported with: supported force (N) and equivalent mass (kg); gap target (mm) and stability band (+/- mm); surface type; mist state; ambient conditions; thermal state; battery state (voltage, SOC, pack temperature).

A power figure without its conditions is not a power figure — it is a guess wearing a lab coat.

5.2 Mission Profiles

• Profile A — Demonstration Hover Hold: Hold gap for ___ minutes with minimal translation. Report: Wh/min at F=___N and g=___mm.
• Profile B — Low-Speed Mobility: Controlled translation at ___ km/h average. Report: Wh/km at specified speed and surface/mist state.
• Profile C — Urban Stop/Go: Repeated accel/decel events (most realistic). Report: Wh/km plus regen capture per stop (Wh recovered).

5.3 Range Claims (Targets Clearly Separated from Measured)

Target Range (Pre-Validation): Range targets are expressed as conditional estimates only.

“Projected range will be reported in Wh/km under defined mission profiles. Preliminary targets will be replaced by measured Wh/km after rig validation of force-vs-gap and thermal limits, then confirmed in controlled field tests across representative surfaces.”

Measured Range (Post-Validation): Reported as measured usable energy (kWh), measured consumption (Wh/km), and measured range (km) per profile with full conditions metadata.

5.4 Efficiency and Loss Accounting

System efficiency is broken into: driver efficiency (electrical conversion losses in the inverter); coil losses (I squared R resistive losses plus proximity effect losses); substrate losses (eddy current heating and dissipative coupling in the surface and mist layer); auxiliary loads (mist system, cooling, compute).

“Efficiency” is reported as net mechanical effect per electrical input under logged conditions, not as an abstract percent. An efficiency number without its operating point is meaningless.

6. Technical Specifications and Performance Metrics (Measured-First)

6.1 Claim Maturity Statement

“All performance claims are categorized as Target (T), Measured (M), or Derived (D). No Target value is presented as achieved until verified under logged test conditions.”

6.2 Performance Metrics Table

Legend: M = Measured, T = Target, D = Derived

6.3 Footnotes (Credibility Armor)

• Payload vs Force: A 150 kg system mass requires approximately 1470 N of upward force (not including dynamic margin). Any payload claim must be backed by a measured force-vs-gap curve and thermal limits.
• Gap Targets: Start with mm-scale rig goals. cm-scale hover is only claimed if the same force can be held at that distance without runaway oscillation and within thermal limits.
• Range Reporting: Range is not stated as a standalone number. It is derived from measured usable energy and measured Wh/km under defined mission profiles.
• Regen Honesty: Regen is reported as Wh recovered per event and net benefit after losses.
• All “Measured” values require test metadata: surface type, mist state, ambient conditions, and thermal state must be attached to the measurement record.

7. Safety Systems, Redundancy, and Verification

7.1 Safety Philosophy

The Franklin Construct is designed so that any single fault (sensor fault, drive fault, power interruption, loss of coupling, software fault) results in a controlled transition to a safe state.

Safe state is defined as: stability is deprioritized in favor of landing; the platform executes a controlled descent to a passive support method (skids/wheels); high-energy electromagnetic actuation is disabled after touchdown.

No safety feature is considered implemented until it is measured, logged, and repeatable.

7.2 Safety-Critical Components (With Measurable Requirements)

A) Dead-Man Control (Primary Human Safety Interlock)

Requirement: Release triggers immediate transition to controlled descent and disables translation mode.

Verification (Rev A): Response time <= 100 ms from release to state change; tested with human-in-loop at bench, then at each gate involving human presence.

B) Emergency Stop (E-Stop)

Requirement: E-stop triggers coil drive shutdown immediately. Energy storage isolated via contactors.

Verification (Rev A): Command latency: coil drive disable <= 50 ms; contactor open initiated <= 100 ms; post-touchdown coil voltage near zero.

C) Mechanical Passive Support (Skids/Wheels)

Requirement: Board remains stable and controllable on passive supports at low speed and during landing events.

Verification (Measured): Drop/landing tests: ___ trials, max measured deceleration ___ g, no structural failure.

D) Overcurrent / Overtemperature Protection

Requirement (Rev A): Inverter current limited to 1.25x intended steady hold current; thermal derating ladder enforced; hard LAND/shutdown at critical temperature.

7.3 Fault Handling Rules (Stop-Rule Behaviors)

• Gap estimate confidence loss: if Conf < threshold → controlled descent to skids/wheels.
• Driver fault / overcurrent: immediate clamp + state transition; disable translation.
• Power sag / undervoltage: controlled descent; isolate pack if needed.

Each fault is injected in test (software-in-the-loop and bench), and the system must demonstrate correct detection, correct state transition, bounded touchdown behavior, and logged diagnostic record.

7.4 Exposure and Operational Safety

Nanoparticle / Aerosol Controls: sealed reservoirs, filtered vents, fill procedure, PPE, and operating restrictions. Operational aerosol/dust monitoring results recorded as available.

Electromagnetic Field Caution Zone: define a conservative no-go radius for pacemakers/implanted devices until field mapping is complete. Field strength vs distance mapped in test, recorded and summarized in an exposure appendix. No medical claims. Just measured fields plus conservative rules.

Thermal Contact and Burn Hazards: external surfaces must remain below ___ degrees C during normal operation. Hot surfaces shielded and labeled. Verified by measured surface temps in a standardized duty cycle test.

7.5 Safety Verification Matrix

7.6 Power Loss Language (Revised)

“On power loss, the system transitions to passive supports (skids/wheels). Any residual electromagnetic effects are treated as incidental; safe landing is guaranteed mechanically.”

8. Prototype Development Path (Gated Validation Program)

8.1 Overview

Development proceeds through gated milestones. Each gate has: an objective (what we must prove); a test setup (how we prove it); pass/fail criteria (what “success” means — measured, not hoped); and artifacts (documentation/data that must exist before advancing).

No rider tests occur until Gates 1-3 are passed with logged evidence. This is a structural rule of the program. The temptation to skip ahead and “just try it with a person” is the single most dangerous failure mode of any novel mobility prototype. The gating program exists to prevent that.

8.2 Gate 0 — Bench Bring-Up and Electrical Safety (Rev A)

Objective:
Prove the power electronics, sensing chain, and safety interlocks function reliably under non-load conditions. This gate is about the electronics, not the physics. No magnetic interaction with surfaces is required. The question is: does the electrical system behave predictably and safely?

Test Setup:
Bench inverter + coil dummy loads / sacrificial coils; instrumented power supply or pack emulator; data logger; thermal sensing; safety circuits. No rig, no surfaces, no mist. Pure electrical validation.

Pass/Fail Criteria (Rev A Targets):

Precharge & Contactors:

• Precharge complete time: <= 1000 ms
• Inrush peak current after precharge: <= 2x steady-state current (log both values)
• Contactor open time (command to confirmed open): <= 150 ms
• Contactor close time (command to confirmed closed): <= 250 ms

Protection:

• Overcurrent threshold (software + hardware clamp): set to 1.25x intended steady hold current
• Overcurrent clamp response time: <= 5 ms (measured)
• Overtemp derate: Level 1 at 70C; Level 2 at 85C; forced LAND/shutdown at 95C (verified)
• Dead-man → LAND logic: <= 100 ms (measured)
• E-stop → coil drive disable: <= 50 ms; energy isolation: <= 100 ms (measured)

Logging Standard:

• Log rates verified: state/g_est/IMU >= 200 Hz; V_pack/I_pack/temps >= 25 Hz
• Absolute must: event markers for E-stop, dead-man, derate level change, fault code set, LAND start, touchdown

EMI Sanity:

• No uncontrolled resets or sensor corruption at full switching operation
• Voltage measurement accuracy: <= 2%
• Current measurement accuracy: <= 5%
• Data logger captures V/I/P/E streams with correct units and timestamps

Gate 0 PASS (Rev A):

• 0 unexpected MCU resets in a 5-minute “full switching / representative load” bench run
• Safety events are correctly logged and transition to safe state deterministically
• All protection thresholds verified with logged evidence

Artifacts Required:

• Power/thermal logs (raw + summary)
• Safety circuit timing report (precharge, contactor, E-stop, dead-man — all with measured timestamps)
• Driver efficiency measurement under representative waveforms (bench)
• Current and voltage calibration records
• EMI sanity report (sensor noise comparison: coils off vs full switching)

8.3 Gate 1 — Static Force Validation (Core Physics Gate) (Rev A)

Objective:
Demonstrate repeatable force-vs-gap behavior over representative surfaces, with controlled thermal rise. This is where we stop theorizing and start counting newtons. If the force is not there, the architecture is revised before scaling.

Test Setup:
Instrumented static rig: force sensor/load cell, controllable gap mechanism, surface samples (asphalt/concrete/wood/metal reference), optional mist application, temperature sensors. The rig must allow controlled, repeatable gap positioning.

Pass/Fail Criteria (Rev A Targets):

Force Targets:

• Key operating point: F >= 50 N at g = 10 mm for >= 60 s (static hold, control off)
• Short hold: F >= 80 N at g = 7.5 mm for >= 30 s
• Why these numbers: big enough to prove you are not measuring noise or magnetic “tingle,” but not so big you are implicitly claiming rider-scale

Repeatability:

• Same-condition repeated runs: >= 3 runs
• Force variance at the key operating point: <= 10% across those runs

Thermal Limits (Rev A):

• Over the 60 s hold at the key point: coil temp rise: Delta T <= 15C; driver temp rise: Delta T <= 15C
• If coil or driver temps exceed Rev A limits at any gap point, terminate the sweep and mark the run FAIL (thermal stop rule)

Mist (Optional in Rev A):

• Sheet resistance measurement points: t = 10 s, 30 s, 60 s, 120 s
• Uniformity grid: 9-point minimum
• Acceptable uniformity variance (Rev A): <= 25% (tighten later)
• Persistence window goal: >= 60 s within your defined sheet-R threshold

Gate 1 PASS (Rev A):
You produce at least one repeatable force-vs-gap curve and hit the 50 N @ 10 mm hold without overheating.

Artifacts Required:

• Force-vs-gap plots with full conditions metadata (all 7 gap points)
• Thermal rise plots (coil and driver) for the 60 s key-point hold
• Mist characterization report (if enabled): sheet resistance vs time, uniformity grid, persistence window
• Load cell calibration record
• Gap mechanism repeatability verification
• >= 3 repeated run overlays at the key operating point showing <= 10% variance

This gate is the “truth gate.” No amount of software sophistication, control theory, or presentation polish substitutes for measured newtons.

8.4 Gate 2 — Closed-Loop Gap Hold (No Translation) (Rev A)

Objective:
Achieve stable closed-loop height control (“hover hold behavior”) on rig without lateral motion. Now you are proving hover-hold as a control problem — not just a force measurement, but a system that can maintain a commanded gap using sensor feedback and degrade gracefully when sensors fail.

Test Setup:
Same rig as Gate 1, but now with full sensor suite and estimator running. Disturbance inputs (small shocks/tilt within safe rig limits). Sensor-blind simulations via software or physical occlusion. Fault injection capability for undervoltage, overcurrent, and sensor loss.

Pass/Fail Criteria (Rev A Targets):

• Gap hold band: +/- 2 mm (rig positioning repeatability must be tighter than control stability targets)
• Hold duration: >= 60 s
• Operating point: whatever was validated in Gate 1 (start with the 50 N @ 10 mm point)
• Minimum estimator confidence during stable hold window: Conf >= 0.70
• Coupling quality metric: Cq >= 0.60 (Rev A) if computing
• Hard fault injected (overcurrent/undervoltage): safe state reached within timing requirements

Gate 2 PASS (Rev A):
Stable gap hold + successful sensor-blind landing + clean logs.

Artifacts Required:

• Control logs (gap estimate, confidence, actuator commands) for full duration
• Fault injection report (sensor loss, current fault, undervoltage) with timestamps and state transitions
• Verified state machine diagram + transition timings
• Disturbance response plots (showing <= 500 ms recovery to +/- 2 mm)
• Estimator Conf and Cq traces for the full hold window

8.5 Gate 3 — Tethered Translation (Low Energy, Low Speed)

Objective:
Demonstrate controlled lateral motion under tether with conservative limits. The tether prevents runaway; the test proves that the system can translate without losing stability or thermal compliance.

Test Setup:
Tethered sled or gantry support (prevents runaway); controlled surface track; speed limited; passive skids/wheels engaged as backup. Safety spotters positioned. E-stop accessible.

Pass/Fail Criteria (Measured):

• Stable translation at v = ___ m/s for >= ___ m with no oscillatory divergence
• Braking mode produces controlled decel; stopping distance measured and repeatable
• No thermal runaway in coils/drivers over defined duty cycle
• E-stop yields safe touchdown within envelope under motion
• Drift measured: ___ cm over ___ s (within limit ___ cm)

Artifacts Required:

• Wh/km consumption under Profile B-like conditions
• Braking curves (distance vs speed vs payload)
• Thermal + fault logs
• Translation drift measurements

8.6 Gate 4 — Untethered, No Rider (Dummy Payload)

• Safe stop on command; repeatable braking envelope
• Controlled descent behavior verified under multiple fault triggers
• Remote kill verified (range, response time ___ ms)
• No uncommanded translation beyond ___ m
• Field exposure mapping begins (distance vs field strength)

8.7 Gate 5 — First Human Stand (Stationary Only)

Objective:
Demonstrate human-safe stationary operation with maximum conservative limits. Everything is biased toward caution.

Preconditions:

• Gates 0-4 passed and documented
• PPE complete (helmet, eye protection, gloves, ankle support footwear, optional elbow/knee pads)
• Human test protocol written and signed
• Harness/spotting plan (if used) ready
• Speed disabled (Tier 0 enforced)

Pass/Fail Criteria (Measured):

• Stationary stability for >= ___ s within gap band
• No unexpected thermal rise beyond limits during short run
• Dead-man release → safe landing, measured timing
• No excessive EMI issues (sensor stability preserved; no resets)
• E-stop tested with human aboard (protocol-defined)

8.8 Gate 6 — Controlled Human Mobility (Incremental Speed Steps)

Objective:
Expand from stationary to low-speed motion with measured braking envelope and conservative scaling. Speed limits raised in steps (e.g., 1 m/s → 2 m/s → …), each step requiring repeatable braking distance and thermal compliance. No step is skipped.

Progression Rule:
Tier increases only after >= ___ passes at current tier with no stop-rule triggers.

8.9 Test Stop Rules (Mandatory)

Testing halts immediately if any of the following occur: unexpected oscillation growth; estimator confidence collapse; thermal runaway; uncommanded translation; safety interlock failure; field exposure beyond defined conservative limits; structural anomaly; unexplained sensor corruption or reset.

8.10 Why This Gating Structure Is Worth It

It forces the document to say: “We will earn each claim.” And it gives a clean narrative when someone asks, “What have you proven?” — you point to gate artifacts, not vibes. Each gate produces logged evidence. Each claim traces back to a Run ID. The program is defensible because it is structured around proof, not promise.

9. Risk Register and Failure Mode Mitigation (FMEA-Lite)

9.1 Scoring

Each risk is scored 1-5 on three axes: Severity (S): 1 = negligible, 5 = severe injury / major damage; Likelihood (L): 1 = rare, 5 = frequent; Detectability (D): 1 = easy to detect early, 5 = hard to detect before effect.

A composite internal score (often expressed as S × L × D) may be used for prioritization, but closure requires verification by logged test artifacts.

Higher-risk items are prioritized for mitigation before advancing gates. Priority is determined by Severity, Likelihood, Detectability, observed test outcomes, and gate relevance.

9.2 Risk Register Table (Selected Key Risks)

Note: Full risk register includes R-08 (slip/translation drift), R-09 (brake instability), R-10 (power loss mid-operation), R-11 (EMI corrupts sensors/MCU), R-14 (surface damage / unintended deposition), R-15 (mechanical structural failure), R-16 (thermal contact burn).

9.3 Top Priority Items

• R-01 Gap estimation loss: If the system cannot reliably know where it is relative to the ground, nothing else works. This is the single most critical sensing risk.
• R-07 Mist nonuniformity: The mist track is a core enabler but also a core uncertainty. If coupling varies unpredictably across the surface, the control system faces a constantly shifting plant model.
• R-12 EM exposure mapping: Until fields are mapped, exclusion zones are guesswork. This must be addressed before any human-involved testing.
• R-13 Aerosol exposure controls: Lungs are not replaceable parts. Material handling must be taken seriously from day one, not retrofitted after someone gets sick.

9.4 Risk Closure Rules

A risk is not closed by intent language. A risk is closed by demonstrated mitigation effectiveness under logged test conditions relevant to the gate where the risk matters.

Appendix I — Gate Checklist Sheets (Printable Pass/Fail)

Gate 0 Checklist — Bench Bring-Up & Electrical Safety (Rev A)

• [ ] Precharge complete <= 1000 ms (measured)
• [ ] Inrush after precharge <= 2x steady-state current (measured; both values logged)
• [ ] Contactor open time <= 150 ms (measured)
• [ ] Contactor close time <= 250 ms (measured)
• [ ] Overcurrent clamp set to 1.25x steady hold current; reaction time <= 5 ms (measured)
• [ ] Overtemp derate Level 1 triggers at 70 degrees C (verified)
• [ ] Overtemp derate Level 2 triggers at 85 degrees C (verified)
• [ ] Forced LAND/shutdown triggers at 95 degrees C (verified)
• [ ] Dead-man → LAND logic <= 100 ms (measured)
• [ ] E-stop → coil drive disable <= 50 ms; energy isolation <= 100 ms (measured)
• [ ] Log rates verified: state/g_est/IMU >= 200 Hz; V_pack/I_pack/temps >= 25 Hz
• [ ] Event markers firing correctly for E-stop, dead-man, derate, fault, LAND, touchdown
• [ ] 0 unexpected MCU resets in 5-minute full-switching bench run
• [ ] Sensor streams remain coherent under switching load
• [ ] Fuse strategy documented (ratings, locations)
• [ ] Watchdog reset behavior verified (no unsafe outputs on reboot)

Gate 0 PASS / FAIL — Notes: _______________________________________________

Gate 1 Checklist — Static Force Validation / Truth Gate (Rev A)

• [ ] Load cell calibrated (record on file)
• [ ] Gap mechanism verified (repeatable position within +/- 0.5 mm or better; rig positioning repeatability must be tighter than control stability targets)
• [ ] Thermal sensors placed and reading plausibly
• [ ] Force measured across 7 gap points (g = 5, 7.5, 10, 12.5, 15, 20, 25 mm; 10 s hold each)
• [ ] F >= 50 N at g = 10 mm for >= 60 s (static, control off)
• [ ] F >= 80 N at g = 7.5 mm for >= 30 s (short hold)
• [ ] Repeatability: variance <= 10% across >= 3 runs at key operating point
• [ ] Coil temp rise Delta T <= 15 degrees C over 60 s hold
• [ ] Driver temp rise Delta T <= 15 degrees C over 60 s hold
• [ ] (If mist enabled) Sheet resistance measured at t=10s, 30s, 60s, 120s
• [ ] (If mist enabled) Uniformity measured (>= 9-point grid) with variance <= 25%
• [ ] (If mist enabled) Persistence window >= 60 s within sheet-R threshold

Gate 1 PASS / FAIL — Notes: _______________________________________________

Gate 2 Checklist — Closed-Loop Gap Hold / No Translation (Rev A)

• [ ] Gap hold achieved: +/- 2 mm for >= 60 s at validated operating point (50 N @ 10 mm)
• [ ] Estimator confidence Conf >= 0.70 during stable hold window
• [ ] Coupling quality Cq >= 0.60 during stable hold window (if computing)
• [ ] No oscillation detector triggers during stable interval
• [ ] Disturbance applied; system returns to +/- 2 mm within <= 500 ms
• [ ] No thermal derate beyond Level 1 during test
• [ ] Primary gap sensor blinded → conservative mode entered <= 200 ms (logged)
• [ ] Conservative mode → LAND to passive supports within <= 2 s
• [ ] Hard fault injected (overcurrent/undervoltage) → safe state reached

Gate 2 PASS / FAIL — Notes: _______________________________________________

Gate 3 Checklist — Tethered Translation (Low Speed)

• [ ] Stable translation at v = ___ m/s for distance ___ m (no divergence)
• [ ] Drift measured: ___ cm over ___ s (within limit ___ cm)
• [ ] Yaw stability acceptable (no uncontrolled yaw events)
• [ ] Controlled stop from v = ___ m/s with stop distance <= ___ m
• [ ] Repeatability: stop distance variance <= ___% across ___ runs
• [ ] Regen recorded (if enabled): ___ Wh per stop (net)
• [ ] E-stop under motion → safe touchdown within envelope
• [ ] No thermal runaway; tempsc within limits

Gate 3 PASS / FAIL — Notes: _______________________________________________

Gate 4 Checklist — Untethered, No Rider (Dummy Payload)

• [ ] Stable low-speed motion within g___mm; no runaway translation
• [ ] Safe stop on command within defined stop distance
• [ ] E-stop tested under motion; touchdown within envelope
• [ ] Remote kill verified (range + response time)
• [ ] Thermal compliance under duty cycle
• [ ] Logs saved and backed up

Gate 4 PASS / FAIL — Notes: _______________________________________________

Gate 5 Checklist — First Human Stand (Stationary Only)

• [ ] Speed disabled (Tier 0 enforced)
• [ ] Gap hold +/- ___ mm for >= ___ s
• [ ] Dead-man release → safe landing within envelope (measured touchdown time)
• [ ] E-stop tested with human aboard (protocol-defined)
• [ ] Inspection passed (no overheating, no residue hazards)
• [ ] Logs saved and backed up

Gate 5 PASS / FAIL — Notes: _______________________________________________

Gate 6 Checklist — Incremental Human Mobility

• [ ] Stable controlled motion at tier speed for distance ___ m
• [ ] Braking distance within limit at that tier
• [ ] Thermal compliance within duty cycle
• [ ] Tier increases only after >= ___ passes at current tier with no stop-rule triggers
• [ ] Wh/km measured (Profile B/C)
• [ ] Range derived from measured usable kWh and measured Wh/km
• [ ] Claims register updated with Run IDs

Gate 6 PASS / FAIL — Notes: _______________________________________________

Appendix K — Force/Power Scaling Plan (From Rig Module to Rider Scale)

K.0 Module Definition

A module is one independently driven coil zone with local current sensing and associated field/gap sensing inputs, capable of producing a measurable force contribution that can be summed with additional modules under coordinated control.

This definition anchors all scaling discussion. When this document refers to “module-scale” testing, it means testing a single instance of this unit. When it refers to “multi-module” testing, it means two or more of these units operating under coordinated control on the same platform.

K.1 Scaling Philosophy

We do not “scale by hope.” We scale by measured curves and controlled replication of modules.

The core measured primitives are: force vs gap (under defined surface + mist state); power vs force (average and peak); thermal rise vs power (for coils and drivers); and stability margin (allowable disturbance before LAND triggers).

A rider-scale system is achieved by replicating a validated module across area, while ensuring power delivery, thermal management, and control coordination remain stable.

K.2 Define the Unit Module (The Thing You Scale)

A “module” is the smallest independently controlled lift/stability unit (one coil zone with local sensing and current control).

For each module, record (Measured): active area (m sq); maximum stable force at gap F(g); average power at that operating point P(F,g); peak transient power (and duration); thermal rise vs time at operating point.

K.6 Thermal Scaling (The Silent Killer)

When scaling, ensure adequate heat spreading, airflow or conduction paths scale with module count, and no “hot island” forms.

Scaling rule: if any module hits derate Level 2 during a duty cycle at a given tier, do not advance tiers — fix thermal design first.

K.7 Control Scaling (Multi-Module Coordination)

As N increases, avoid “fighting controllers”: each module has a local inner loop; a mid-level coordinator allocates force demand across modules (load balancing); a top-level estimator and state machine governs modes and safety.

Required tests before rider-scale: two-module interference test (force scaling + stability); four-module load-sharing test (no oscillatory coupling); full-array test with dummy payload, then tethered motion.

K.8 Scaling Gates

• Scale Step 1: 1 module → prove F(g), P(F,g), Delta T (Gate 1 & 2)
• Scale Step 2: 2 modules → verify near-linear force scaling (new mini-gate)
• Scale Step 3: 4 modules → verify coordinated stability + thermal distribution
• Scale Step 4: Full array + dummy payload → validate mission profiles B/C
• Scale Step 5: Human stand (Gate 5) only after dummy payload stability and stop behavior are repeatable

K.9 Range Scaling

Range is derived from measured consumption: Profile A: Wh/min; Profile B/C: Wh/km.

Then: Range = E_usable / (Wh/km), where E_usable is measured usable energy.

Appendix E — Data Dictionary (Logging Standard)

E.1 Required Fields (Minimum Viable Logging)

All logs must include timestamped streams for:

• Electrical: V_pack (V), I_pack (A), P_pack (W), E_used (Wh), SOC (%), T_pack (degrees C)
• Thermal: T_coil_i (degrees C), T_driver_i (degrees C)
• Control / State: state (enum), derate_level (0-3), fault_code (enum list), g_cmd (mm), g_est (mm), dg_dt (mm/s), est_conf (0-1)
• Inertial: roll / pitch / yaw (deg), gyro (deg/s), accel (m/s sq)
• Sensors: g_sensor_primary (raw + mm), g_sensor_secondary (raw + mm), hall_i (raw units or mT if calibrated)
• Mist (if enabled): mist_enabled (0/1), flow_cmd (mL/s), flow_est (mL/s), nozzle_fault (0/1)

E.2 Units and Sampling Rates (Rev A)

• Control-critical fields (state, g_est, IMU): >= 200 Hz
• Electrical/thermal (V_pack, I_pack, temps): >= 25 Hz (faster during transients if logger supports it)
• Any resampled data must record the resample method

E.3 Event Markers

Logs must include event markers: EVENT_ESTOP, EVENT_DEADMAN, EVENT_DERATE, EVENT_SENSOR_BLIND, EVENT_LAND.

Appendix N — Glossary & Symbols

N.1 Symbols (Math / Control)

• F — Supported force / lift (Newtons, N)
• F_req — Required force to support system weight (N)
• F_target — Required force plus dynamic margin (N)
• g — Gap (distance from board reference plane to surface), millimeters (mm)
• g_cmd — Commanded gap (mm)
• Delta g — Allowed gap error band (mm)
• g_dot — Gap rate of change (mm/s)
• P — Power (Watts, W)
• E — Energy (Watt-hours, Wh)
• V — Voltage (Volts, V)
• I — Current (Amps, A)
• B — Magnetic flux density (Tesla, T) or milliTesla (mT)
• alpha — Dynamic margin fraction (e.g., 0.3 for +30%)

N.2 Control and Estimation Terms

• Estimator — The software that fuses sensors to estimate gap, attitude, and confidence metrics
• Conf — Estimator confidence score (0-1). Low Conf means “distance/attitude estimate is unreliable.”
• Cq — Coupling quality score (0-1). Indicates how predictable the surface/mist interaction appears right now.
• Derate Level — Thermal/power safety level (0-3) controlling how aggressively the system is allowed to operate
• PLL Lock — Phase-Locked Loop lock state indicating resonant drive tracking is stable
• Stop Rule — A condition that forces the system into LAND

N.3 Modes / States (State Machine)

• IDLE — Powered but inactive; coils and mist disabled
• ARMED — Sensors verified, logging on, safety live, awaiting command
• GAP_HOLD — Stationary gap stabilization mode (no translation)
• TRANSLATE — Controlled motion mode
• BRAKE — Controlled deceleration / damping mode

N.4 Surface / Mist Terms

• Sheet Resistance (Ohm/sq) — Electrical resistance of a thin film independent of size
• Persistence Window — Time interval after deposition during which film properties remain within thresholds

N.5 Evidence Terms (Measured-First)

• T (Target) — Intended value, not yet verified
• P (Provisional) — Observed once successfully (needs repeat runs)
• M (Measured/Verified) — Verified with >= ___ runs, full metadata, and no stop-rule triggers
• D (Derived) — Computed from measured inputs (e.g., range = usable kWh / Wh/km)

Appendix O — Minimum Instrumentation List (Gates 0-2)

This is the “don’t try to prototype blind” list. It’s the smallest set of tools that makes early data defensible.

O.1 Gate 0 (Bench Bring-Up) — Minimum Instruments

• Voltage measurement (pack voltage, inverter bus voltage)
• Current measurement (pack current; inverter/coil current if available)
• Temperature sensors on: coil(s), driver heatsink/device area, battery pack
• Event logging for: E-stop, dead-man, contactor state, derate level
• Data logger (any platform that timestamps and exports CSV reliably)
• Nice-to-have (not required): oscilloscope for debugging switching and noise

O.2 Gate 1 (Static Force Validation) — Minimum Instruments

• Load cell / force sensor with calibration record
• Gap measurement method independent of field inference (ToF or ultrasonic preferred)
• Repeatable gap positioning mechanism (even manual, but must be measurable and repeatable)
• Thermal sensors (coil and driver at minimum)
• Power logging (V/I/P/E as in Gate 0)

If mist is enabled (minimum add-ons):

• Sheet resistance measurement method (4-point probe preferred; otherwise clearly labeled proxy)
• Simple uniformity sampling plan (grid points + record sheet resistance or proxy)

O.3 Gate 2 (Closed-Loop Gap Hold) — Minimum Instruments

• Everything from Gate 1, plus:
• Two independent gap inputs (primary + secondary), OR one direct gap sensor plus a validated conservative LAND trigger when it is blinded
• IMU (preferably dual) for attitude disturbance detection and estimator robustness

End of Document

The Franklin Construct Mark II — Complete Engineering Document
Sections 0-9 + Appendices A-O | Rev A Targets Integrated

This document is a traceable engineering program where every claim points to data. It is structured so that progress is earned through logged evidence, scaling is validated through measured replication, and safety is guaranteed mechanically when all else fails.

The measured-first doctrine applies to every number, every table row, and every public statement. No Target is presented as achieved. No Derived value stands without its measured inputs. No gate is passed without its artifacts.