Engineering notes from the Gleagle R&D lab — wireless power, control algorithms, LED efficiency, and levitation manufacturing, documented in full.
Gleagle Hardware R&D | November 2026 — Our upgraded PU500G transmitter keeps the 12 V single-switch quasi-resonant platform and delivers 125 mW at a 35 mm levitation gap with about 2 W input. The measured series reaches 315 mW at 15 mm.
The key loss was the galvanized steel base plate, not the Y35 magnet ring. We document the 0.5 mm aluminum isolation ring, assembled-product frequency tuning, receiver matching and Hall-signal checks — with the test limits stated plainly.
Gleagle Hardware R&D Team | October 2026 — "It's only two watts, a phone charger will do" — the two-watt figure is correct and completely irrelevant. Levitation is a peak-power load: 1.5–2.5 W steady, but 12 V × 3 A = 36 W during catch-and-lock, a 15:1 crest factor. And 12 V is a legacy optional USB PD profile — PD 2.0 replaced it with 15 V; only 5/9/15/20 V are mandatory, and fewer than half of retail adapters expose 12 V.
Force is quadratic in voltage (F/F₀ = (V/V₀)²): 9 V fallback leaves 56% of lifting force, and a hard-reset adapter drops the floater — a recoverable electrical problem becomes an irreversible mechanical one. The fix: fixed 12 V / 5 A adapter, or negotiate 15 V and buck down on-board.
Gleagle Levitation R&D | October 2026 — Float height is bought, not tuned. Our field-tested playbook for pull-down (top-attraction) systems: permanent-magnet bias (power falls with (1−η)² — a 50kg / 40mm worked example drops from 1.2kW to 3W), closing the magnetic circuit with back iron (+41% height, and why the same trick buys only +19% on push-up), a top bias magnet for the last few millimeters, high-voltage switching drive, and 4-axis balance — plus the power-off holding problem and the 1.15× over-bias scheme that solves it.
Field reminders included: stacked-plate thickness scans, polarity-reversal tests, why you run the power-cut test ten times, and why simulated absolute heights are not to be trusted.
Gleagle Levitation R&D | October 2026 — Our next-generation push-up (repulsion) platform splits the two hardest problems apart: magnets buy float height (pole area, bore span, energy product), while denser coil arrays hold stability — whose demand grows with a high power of h. Includes the ring-magnet field-arch explainer, the computable area-bore-height trade table, and the magnet-matrix concept that also slows floater spin.
Key insights: why the vertical axis is free in push-up systems (static power → zero in theory), why “height is bought, not tuned”, and why a floater smaller than the ring bore is not wasting the field.
Gleagle Levitation R&D | October 2026 — An architecture deep-dive into one-handed float: why single-direction drives (F = k·i, one polarity only) inevitably overshoot and run hot, how bidirectional drive creates restoring force on both sides of the levitation point so overshoot is braked electrically, and how the slow-servo loop integrates the static bias current to zero — load-independent.
Key insights: the positioning aid exists to do the “catch” the system cannot do itself; the potential well turns from a cliff into a symmetric well; e_ss = 0 means the optimum is found, not designed — and it re-converges for any floater weight.
Gleagle Levitation R&D | October 2026 — Pull-down (attraction) levitation hits a square-law wall: P ∝ h². Our next-generation approach breaks it twice — a permanent-magnet ring carries the static load (power falls with (1−η)²; 80→95% bias is a 16× cut), and a multi-matrix coil array splits correction work (power falls as 1/N). Full derivation chain, bias and channel tables, and the cost ledger.
Key insights: why required B has nothing to do with the gap, how bias also turns the actuator linear and bidirectional, and the honesty clause — bias approaches zero static power but never reaches it; push-up does.
Gleagle Levitation R&D | October 2026 — Why can't a levitating lamp take a brighter LED? Geometry, not engineering: coupling falls with distance cubed (k ∝ (d₀/d)³), eddy currents burn power in the levitation magnet (Pe ∝ f², skin depth ≈ 1.5 mm at 172 kHz), and a 2 W transmit budget reaches the LED as ~100 mW at a 20 mm gap. Our tuned link lands ~500 mW — and why simply cranking transmit power hits four walls: eddy heating, EMC, loop stability, and the cube law itself.
The levers that actually work: closing the gap (20→18 mm = +37% coupling), ferrite field shaping, tracked resonance, and LED efficacy at the knee.
Gleagle Application Engineering | October 2026 — Sourcing levitating displays for the US market? This buyer's guide covers push-up vs top-type architectures, payload classes and float-gap honesty, 8 manufacturer evaluation criteria, 6 supplier red flags, the real procurement timeline from inquiry to mass production, MOQ and US logistics, and the five factors that actually drive cost.
Key insights: why "maximum payload" quotes mislead, why battery-powered floaters are a red flag for professional displays, how to verify a real factory in 10 minutes, and market price reference points from $129–$205 US retail to factory-direct volume pricing.
Gleagle Hardware R&D Team | October 2026 — A rainbow looks like the easiest feature to build — in a power-locked levitating system it is the hardest. Three engineering decisions made the difference: full-hue continuous color (no 64-step jumps, 100% of the color wheel at maximum saturation), gamma correction tuned to the human perception curve, and channel weighting that keeps all 21 curated colors equally bright at constant power.
Key insights: why most RGB rainbows look steppy, why perceived brightness is a perception problem rather than a wattage problem, and how channel weighting fixes muddy mixed colors without overheating the base.
Gleagle Hardware R&D Team | October 2026 — "Can it run on batteries?" We have the technology — production-proven designs exist for commercial displays — but for consumer products we do not recommend it. The math: steady levitation draws 1.5–2.5 W, yet catch-and-lock peaks hit 36 W, forcing a small lithium cell to deliver 10–15 A through a boost converter.
Key insights: why a dead battery means a falling floater (impact force ≈150 N on PU systems, upward slam on PD systems), the "clack" that wakes households at night, how battery aging makes every risk worse over time, and the UN38.3 / dangerous-goods shipping surcharge. Verdict: manageable risk for commercial displays with low-voltage alarms — strongly not recommended for home use.
Gleagle Hardware R&D Team | October 2026 — Some objects sit perfectly still on a levitation display; others spin the moment you let go. The magnetic field is rotationally symmetric — it neither drives nor brakes rotation, like a frictionless perfect bearing. Spin is decided by your object and the air around it, not by the base.
Key insights: how tiny imbalances act like a hula hoop to build visible rotation from a whisper of torque, three object-side tricks to tune spin speed (center it, round it, stick it), and the four warning signs of spinning too fast — heat, wobble, creep, and noise.
Gleagle Hardware R&D Team | October 2026 — Levitation is not "place it and forget it." A floater drifts every moment; the controller corrects it every moment. This article breaks down the loop that turns levitation from a brief phenomenon into a stable, finely adjustable state: position sensing, force computation, and a power stage that must execute the smallest command faithfully.
Key insights: why magnetic force is inherently unstable (steep nonlinear force-vs-distance), why stability comes from working at the right gain rather than more gain, and how the system continuously seeks the operating point that delivers the same stability with less force — lower energy, less heat, no manual re-tuning.
Gleagle Hardware R&D Team | October 2026 — Like poles of a magnet repel. We turned that single line of physics into a stable, contact-free levitation platform: permanent magnets carry the load passively while a closed-loop stabilization system senses the smallest drift and corrects it thousands of times per second.
Key insights: why magnetism alone can never hold anything still, the deliberate split between passive load-bearing and active stability, fail-safe descent as designed behavior, and the four things you actually experience — silence, stillness, grace, and nothing to wear out.
Gleagle Levitation R&D | September 2026 — Buyers keep asking for dimmable levitating lamps. The physics says no: with an industry receiver budget under 100 mW (ours reaches ~500 mW), brightness steps are below what the eye can resolve — our own three-step moon lamp proved it in the market. The industry workaround, an always-listening remote module, burns 30–50 mW of standby power and takes half the lamp’s light with it.
We also built true stepless dimming over the power carrier (FSK, no RF receiver) for a Canadian retail program — it works, and it still doesn’t matter, because the perceptual problem is upstream of the control channel. Our recommendation: switch color, not level. Zero-power µA-level color switching keeps full brightness in every color — standard on our bulbs and moon lamps.
Gleagle Hardware R&D Team | August 2026 — A levitating speaker sits on its base more than it plays. That means the battery spends most of its time nearly full, caught in a loop that a regular speaker never sees. This article documents the complete battery management system: 4.0V charge ceiling (5x cycle life), 50% storage voltage (2x calendar life), dynamic energy management during levitation, and firmware-based usage detection using 12-bit ADC voltage curve analysis.
Key technical insights: Why lithium cells rot fastest at 4.2V full charge, how Toyota's hybrid strategy applies to consumer electronics, and how our proprietary energy management keeps wireless charging coils cool when the battery tops off.
Gleagle Hardware R&D Team | August 2026 — This article documents the complete thought process behind developing the rainbow color transition algorithm for our next-generation smart shoe display LED strips. From the initial failure of the textbook HSV approach through three failed patch fixes, to the breakthrough sine-wave mathematical solution that delivers constant-brightness, stepless rainbow effects using only a 256-byte lookup table and 4 lines of core code.
Key technical insights: Why HSV produces 100% brightness oscillation (255–510 range), the fundamental coupling problem between saturation and brightness, and how a high-school trigonometric identity — sin(θ) + sin(θ+120°) + sin(θ+240°) ≡ 0 — provides a mathematically provable constant-brightness solution with R+G+B ≡ 384 at all times.
Gleagle Technology R&D Center | August 2026...
Gleagle engineering notes on why a specific coil, resonance and levitation-control architecture uses a 160–180 kHz working range. This is an architecture-specific design choice, not a universal industry frequency...
Imagine this: you're standing in a hotel lobby at 11pm, and you want to change the color of the floating moon lamp behind the reception desk. There's no remote control. No smartphone app. No touch sensor anywhere on the floating unit. How do you do it?...
Wireless power transfer for levitating products has always been a one-way street. The base transmits power to the floating unit — enough to light an LED, charge a small battery, or drive a Bluetooth speaker. But there was no way for the floating unit to talk back. No data channel. No feedback loop. ...
There's a hard physical limit that every wireless power engineer knows about: when you push current through a coil, it heats up. Push enough current through a small coil inside a sealed acrylic base, and you hit 80 degrees Celsius within about an hour. At that temperature, the coil's resistance incr...
Most people assume that a levitating lamp's brightness is determined by how much power the wireless coil can deliver. That's only half the answer. The other half — and in many cases, the more important half — is what the LED does with that power once it arrives....
The physics of magnetic levitation is deceptively simple on paper. A permanent magnet repels another magnet. An electromagnetic coil provides active correction. A Hall sensor detects position. An MID slow-servo loop trims coil current only when needed. The object floats. The problem is that this works perfectly in a textbook ...
If you are a brand, distributor, or procurement manager looking to source magnetic levitation products from China, you have probably encountered the same problem: every factory looks the same on Alibaba. Same product photos. Same specifications. Same "ISO 9001 certified" badge. The only difference i...
Gleagle Levitation R&D | October 2026 — The ring magnet carries the weight, four coil groups only correct — so the hover height is a pure balance problem: the float settles where push force equals weight, and heavier loads must sit closer to the ring. Includes PU2000G bench data (48mm empty → 20mm at full 2kg) and why we recommend sizing to about 50% load.
Here is something you will not find on most datasheets for levitating light bulbs: the real reason almost every floating bulb on the market is single-color. It is not because manufacturers do not want to offer RGB. It is not because customers prefer white light. It is because the power budget simply...
If you've ever tried to control the color or brightness of a levitating lamp remotely, you already know the problem. The floating part has no wires. No battery. No physical buttons. It's just spinning in mid-air, held up by a magnetic field, and somehow it needs to receive color-change commands with...
Here's a problem that haunted our engineering team for about two years: how do you build a single magnetic levitation base that can float a 200-gram sandal and an 1800-gram boot with the same stability? Every levitation device has a sweet spot — a narrow weight range where the electromagnetic field,...
Let's talk about the single biggest reason people return levitating products: the floating object drifts. It starts out fine — perfectly centered, spinning slowly, looking magical. Then after a few hours, or after someone bumps the table, or just because the room temperature changed by a few degrees...
Gleagle Technology R&D Center | WPT Series Part 4 — What turns a resonant link into a shippable product: two-layer foreign object detection (power accounting + Q-monitoring), SELV-constrained receiver voltages, EN 55011 EMC design, and the measured anchors we ship against — 3 W at 160 mm on the Ø200 long-range platform, 150–200 mW across a levitation float gap.
The human eye doesn't see brightness linearly. A 50% PWM duty cycle on an LED doesn't look half as bright as 100% — it looks more like 70-80% bright. This is called the Stevens power law of perceived brightness, and it's been known in vision science since the 1950s. Despite being common knowledge, m...
If you've ever used a magnetic levitation product that came with a plastic positioning jig and a multi-step instruction sheet, you already know the most frustrating part of the experience. You have to hold the floating object at exactly the right height (not too high, not too low), wait for the base...
There's a sound that every levitation engineer dreads. It's a high-pitched whine — somewhere between a mosquito and an old CRT television. It comes from the coil. Specifically, it comes from the magnetic core of the coil vibrating under the PWM current that drives the levitation field. If that vibra...
A practical engineer's guide to why iron is ferromagnetic but copper is not — explained from electron configuration through exchange interaction and domain theory. Includes a permeability reference table for common engineering materials and a design guide for magnetic shielding with mu-metal, steel, and silicon steel. Covers the saturation problem in multi-layer shields and practical rules for selecting shielding materials.
Gleagle Technology R&D Center | WPT Series Part 3 — SS/SP/PS/PP compensation compared: why parallel-parallel is the default for weak-coupling long range (Q-amplified induced voltage), and the six-stage efficiency chain — driver, TX cavity, space link, RX cavity, rectifier, regulator — whose product no single stage can rescue.
Gleagle Technology R&D Center | WPT Series Part 2 — The distance–power scaling law: at a 50 mm gap a Ø200 coil delivers 8× the power of a Ø80 coil. Plus frequency engineering: skin depth and Litz wire selection, the frequency-squared eddy-loss penalty (0.75 W bare vs 2 W installed, measured), and how the industry converged on the 110–230 kHz window.
Gleagle Technology R&D Center | WPT Series Part 1 — Wireless power from the Biot–Savart law up: why coil radius alone sets the field decay rate, the resonance condition, and the closed-form link efficiency η = f(k·Q) — the design philosophy that structure serves k and materials serve Q. Includes critical coupling and frequency splitting ("too close burns").