Build Log

Mechanical Assembly Complete — 6-DOF Shake Table Ready for Electronics

The mechanical phase is done. After weeks of precision fabrication, hundreds of screws, cuts, bruises, and late nights — the StabilityCore 6-degree-of-freedom shake table is mechanically complete and ready for electronics and firmware.

What’s built:

• X-axis translation — NEMA 23 stepper on linear rails with GT2 belt drive and idler wheels for maximum grip
• Y-axis translation — duplicate of X-axis, rotated 90°
• Z-axis vertical — two ball-screw linear actuators in diagonal opposite positions, connected by 2020 lifting frame
• Yaw rotation — NEMA 23 stepper driving aluminum lazy susan via belt and idlers, ±30° with mechanical stops
• X-axis pitch — Savox servo with bevel gear assembly, precision brass couplers, red Loctite on all joints
• Y-axis pitch — Savox servo with identical bevel gear assembly

Passive isolation system:

• Multi-layer spring isolation with matched rates per level — stiffer springs at bottom, softer at top
• Rubber stoppers at spring limits secured with M5 screws through 2020 T-nuts
• Extension springs with ball-bearing fishing swivels for Z-axis counterbalance — free rotation in all directions
• Cast iron payload plate on 2020 spacers with pulley clearance underneath

Professional wiring infrastructure:

• GX12 4-pin aviation connectors for all motor connections
• Color-coded nylon cable sleeves with heat-shrink terminations
• Patch bay on both electronics rack and shake table frame
• Fully disconnectable for transport — unplug 8 connectors and the table separates from the rack

Next phase: Electronics — connecting stepper drivers, wiring the GX12 patch bay, mounting ESP32 boards, sensor installation, and firmware development. Then PID tuning on all 6 axes and historical earthquake waveform playback.

80+ hours of build time. One person. Every screw, every bracket, every alignment decision. The mechanical foundation is solid. Now the machine learns to move on its own.

WaveForge OIMH First Voltage — Wave Energy Validated on the Shake Table

Historic day. First measured voltage from the Orbital Inertial Mass Harvester (OIMH) — WaveForge’s wave energy prototype tested on the StabilityCore shake table platform.

Measured DC voltage data:

• Gentle tilt rocking (~1 RPM): 3V
• Realistic ocean swell (30 BPM): 4V
• Moderate swell (40 BPM): 5–6V
• Active seas (60 BPM): 8V
• Storm conditions (100 BPM): 13V

Key discovery: the offset eccentric mass automatically converts linear rocking into circular orbital motion without any perpendicular nudge input. Pure physics — the asymmetric gravitational pull on the off-center weight self-converts the motion. This confirms the “resonant orbital pumping” mode documented in the WaveForge technical paper.

Two patented technologies validating each other on one platform: StabilityCore simulates the ocean, WaveForge harvests the energy. The shake table isn’t just a seismic simulator — it’s a wave energy research platform.

Shake Table Provisional Patent Filed

Filed provisional patent for the shake table itself as a standalone research instrument.

• Application #64/021,085 — 28 claims
• Covers: 6-DOF platform, distributed ESP32 PID control, spring isolation architecture, volumetric laser visualization, historical earthquake playback, physical fader tuning, solar-powered portable operation, science kit educational deployment
• Filing fee: $130 (small entity)

This is the second StabilityCore patent (the first covers the full-scale building isolation system, #63/986,480). The shake table patent protects the bench-scale instrument that validates the technology and generates revenue through science kit sales.

Total StabilityCore IP portfolio: 2 provisional patents, 102 combined claims.

Intensive Mechanical Build — From Frame to Nearly Complete

Two weeks of intensive daily build sessions transforming the 2020 aluminum extrusion frame into a precision 6-DOF motion platform:

• March 16: Outer frame nylon nut/washer upgrade, sensor architecture planning
• March 17: Registered 3 LLCs (WaveForge, StabilityCore, DayLux), filed WaveForge provisional patent
• March 21: Y-axis 80% complete, Z-axis progress, Govee lighting installed
• March 22: Continued Y-axis assembly, Z-axis mounting
• March 24: Spacers installed, GX12 patch bay ordered, bird app security hardened
• March 25: Y-axis springs installed, X-axis pulleys tensioned, all axes smooth
• March 26: Frame alignment, cable sleeves started
• March 27: Cast iron payload plate installed, inner frame in outer frame complete
• March 28: Top layer complete, weaker springs on Y-axis, Savox servos arrived

Lessons learned: manufactured holes can’t be trusted for precision alignment — always measure. Lock washers + nylon nuts + red Loctite on every critical joint. Spring rates must be matched to actual load at each level. Patience is the most important tool in the box.

Hybrid Magnetic Bearing & Maritime Expansion

Major conceptual breakthrough today. We solved the biggest objection to electromagnetic seismic isolation at building scale: power consumption.

The solution: a hybrid bearing with permanent magnets carrying the static building load (zero power, always on) and electromagnets handling only the dynamic PID trim. The permanent magnets do 95%+ of the heavy lifting — literally. The electromagnets barely have to work.

Added a mu-metal shielding layer between the two magnetic sources to prevent field crosstalk — the same principle used in audio transformers and MRI machines. This enables independent tuning of each layer.

Five-layer bearing stack finalized:

• Permanent magnet half-sphere (passive lift)
• Mu-metal shielding (field isolation)
• Electromagnet (active PID trim)
• Non-magnetic cable tethers (safety limits)
• VFML gas bearing (backup)

Also opened up an entirely new market: maritime stabilization. Same technology, continuous operation. The tagline: "Perfect sleep in stormy seas." Envisioning a full deck of cruise ship sleeping quarters floating on magnetic bearings, plus individual bed isolation.

Conceived a feedforward sensor system using LiDAR, sonar, and machine vision to detect incoming waves and pre-calculate PID corrections before impact. Ships already carry most of this equipment.

Planned a maritime prototype demo: glass of wine on a kayak on the Willamette River, friend drives a waterski boat past, wake hits — wine doesn't move. Two kayaks side by side for comparison. That's the shot.

Finally, wrote the ESP32 stepper motor test sketch for tomorrow's bench test. First real motor spin of the project.

Levitation Demo, VFML Pneumatics, and Parts Flood

Shot the hero photo and video of the electromagnetic levitation demo — stainless steel dome floating over the coil base with blue LED underglow. Dark background, clean composition. This is our first real piece of investor-ready content.

Sourced and ordered the complete VFML pneumatic kit (~$82): Schrader fill valve, 12V solenoid dump valve, glycerin pressure gauge, brass manifold tee, push-to-connect fittings, PU tubing, and PTFE tape. All 1/4" NPT threading for compatibility. Also ordered an AIRBANK electric bike pump for clean, button-press pressurization during demos.

Parts are flooding in: load cells, new ESP32s, ball head joints, microSD card, anti-static bags for ESD-safe storage. The workbench is getting crowded.

Designed and ordered SG90 micro servo winch spools in OpenSCAD for the Space Needle prototype cable system. Press-fit onto SG90 spline shaft, 3 guide grooves, line anchor hole.

Servo Speed Problem & Shake Table Redesign

Setback: Bench-tested the Miuzei 20kg servos for PID correction — way too slow. At 0.16 seconds per 60 degrees, they can't track earthquake frequencies. The HS-422 was responsive but still not fast enough for dynamic loads with momentum.

After researching servo specs and calculating momentum requirements for a ~2 lb building model, selected Savox SC-1258TG (0.08s/60°, 12kg, titanium gears) at ~$85 each. Four ordered for the corners. Expensive, but this is where speed matters most.

Bigger pivot: Redesigned the entire shake table from relay-driven linear actuators to a CNC-style dual-axis stepper system. Two NEMA 23 steppers on MGN12H linear rails with GT2 belt drive. This gives precise position control for replaying actual seismograph data — not just random shaking, but "this IS the 1994 Northridge earthquake at 1/25 scale."

Designed a two-layer 20"×20" HDPE enclosure: bottom plate ("bedrock") holds steppers and rails, top plate ("ground") rides on carriages. Three-sided box hides all mechanics, open back for ventilation and access. Sandbags inside for mass.

Also verified all Pololu controllers (4× Simple Motor, 2× Jrk 21v3) over USB. The Jrk's built-in PID with GUI tuning might be useful for correction axes later.

OMSI Earthquake Challenge Game

Designed the interactive exhibit for OMSI Science Fair (5,000+ attendees). Two stations: a main control panel with LED push buttons and slide faders, and a kid-facing joystick station with arcade buttons.

The game: Kid grabs joystick, tries to keep a ping pong ball on the shaking platform. Timer counts up. Ball bounces unpredictably. Ball rolls off = game over.

Then you toggle PID on. Hand the joystick back. The computer counteracts everything they throw at it. Ball stays centered no matter how hard they shake. Eyes go wide.

Added a PvP mode: Red joystick = earthquake attacker, Blue joystick = defender trying to stabilize. Then PID takes over and beats both humans. Ordered a matching blue joystick for the defender station.

The pitch: "You just felt how hard it is. Now imagine a whole building. Our computer does this 200 times per second."

Modular Cartridge System & Retrofit Revolution

Developed the modular spring cartridge concept — "jack in the box." Pre-compressed, self-contained cartridges with two-stage safety release. Ships locked, installs in an hour, deploys on earthquake detection like an airbag.

Key insight: instead of going under the foundation, insert cartridges directly at post-and-beam connection points. Same install process contractors already use for adjustable jacks. Zero behavior change, same tools, same crew.

Built-in analog pressure gauge on each cartridge — zero-power, always-visible load indicator. Combined with internal force sensors, every cartridge becomes a distributed seismograph node.

Rendered 6 STL files for 3D printing prototypes. Ordered from JLC3DP in MJF PA12-HP (structural) and binder-jetting metal (locking pins).

Patent Filed

USPTO Provisional Patent filed. Application No. 63/986,480. 35 claims, 23 technical figures. $130 filing fee paid. One year to convert to non-provisional (deadline: February 19, 2027).

Title: "Active Seismic Isolation System with Real-Time PID-Controlled Force Cancellation, Electromagnetic Friction Reduction, Multi-Tier Early Warning, and Self-Sustaining Seismic Energy Harvesting."

The patent covers the full technology stack: PID control algorithms, sensor fusion, electromagnetic friction reduction, Zigbee feedforward prediction, energy harvesting, refrigerant cooling, and adaptive gain scheduling.

Also added "Patent Pending" badges to the stabilitycore.io website. It's real now.

Website Launch & Technical Paper

Launched stabilitycore.io on Cloudflare Pages. Four pages: Home (investor pitch), Technology (deep dive), About (founder), Contact. Dark theme, red accents, professional.

Wrote a 10-section academic technical paper with equations, experimental design (6 experiments), and scaling analysis. Expanded patent from 12 to 28 claims including compliant pylon interface, 5-layer defense-in-depth, predictive phase cancellation with Zigbee mesh, and electromagnetic bearing friction reduction.

Designed and ordered first 3D printed parts: pendulum bearing dishes in resin + one in SLM stainless steel (~$255). The steel dish will demo the real bearing surface at scale.

The Church Moment — "Stop Fighting the Vertical"

The VFML (Vertical Force Mitigation Layer) concept didn't come from a whiteboard or a textbook. It came in church.

I was sitting in the pew, thinking about the hardest unsolved problem in the system: vertical earthquake forces. Horizontal isolation has elegant solutions — pendulum bearings, cables, PID correction. But vertical is different physics. You can't just let a building swing sideways on a vertical axis.

Then it hit me: stop fighting it. Just let it go.

An office chair gas cylinder. You pull the lever, the chair drops. It doesn't fight your weight — it accepts it and lets you sink. Then it pumps back up slowly. That's the entire VFML concept: pressurized columns that vent during an earthquake, let the building sink a few inches as the energy is absorbed, then slowly re-inflate afterward.

The sinking IS the absorption. Gravity is the restoring force. The earthquake doesn't need to be fought — it needs to be allowed to pass.

I went home and designed the whole system that afternoon. Four pneumatic posts, O-ring sealed sleeves, a solenoid dump valve triggered by accelerometer. Mechanical relief valves as backup — zero electronics required for basic function. The system degrades from smart to dumb, never from working to broken.

Sometimes the best engineering insights come from the quietest moments.

Day One

StabilityCore began as an idea: what if you could cancel earthquake forces in real time, like a drone flight controller for buildings?

10 hours of independent research and concept development preceded this day. Today we designed the firmware architecture (PID controller, IMU sensor fusion, actuator manager, earthquake waveform generator), planned the 1/25 scale prototype, created the investment pitch, and drafted the first patent claims.

Platform design: 22"×10" clear acrylic, 45 drilled holes, 9 springs, 8 actuators (4 simulation + 4 stabilization), ESP32 with MPU6050 IMU. Side-by-side demo: two platforms, same earthquake, one with PID, one without.

The journey starts here.

What Went Wrong (And What We Learned)

• OONO relay modules + ESP32 = incompatible. Relays need 4-6.5V signal; ESP32 outputs 3.3V. Wasted a session troubleshooting. Lesson: Check signal voltage compatibility BEFORE wiring. Fix: Switched to MOSFET trigger modules (3.3V compatible).
• JQML linear actuators too slow for earthquake simulation. High torque, very low speed. Can't oscillate at earthquake frequencies. Lesson: Actuator speed matters more than force for dynamic simulation. Fix: Shelved for VFML/slow-correction roles. Redesigned shake table with NEMA 23 steppers.
• Miuzei 20kg servos too sluggish for PID correction. 0.16s/60° can't track dynamic loads with momentum. Lesson: Bench test before committing to a design. Fix: Upgraded to Savox SC-1258TG (0.08s/60°).
• SG90 spool walls too thin for 3D printing. JLC3DP rejected first design — 1mm flanges would crack in SLA resin. Lesson: Minimum 2mm walls for SLA. Fix: Redesigned with 2mm flanges, 8mm drum, 14mm flange diameter. Accepted with thin-wall warning.
• Arduino Uno brownout from motor current spike. Driving relay + actuator from Arduino caused resets. Lesson: Separate motor power from logic power. Always.
• TAP Plastics didn't have black HDPE in stock. Ordered white instead. Same material, same strength, just not the color we wanted. Lesson: Cosmetic preferences shouldn't hold up progress.

Every setback pushed us toward a better design. The stepper + belt shake table is far superior to the relay + actuator approach we started with. The Savox servos will outperform the Miuzei by 2×. The MOSFET modules are cleaner than relays. Failures are data.