Roschin & Godin’s ‘Simple SEG’ Replication Reports 35% Weight Reduction

October 9, 2025
|In Searl Effect
|By Tim Ventura

On a spring-mounted platform in a Moscow lab, a one-meter ring of permanent magnets eased through a few hundred revolutions per minute and a displacement transducer began a slow, unmistakable drift. Near 550 rpm the drive motor’s ammeter crossed zero—“back current”—and the team pulled the motor out of the circuit altogether. What remained kept climbing into a narrow operating window just shy of 600 rpm, steady enough that the researchers could step resistive loads onto the shaft and watch two needles move in concert: electrical power climbing while the platform’s apparent weight swung far beyond what ordinary vibration could explain:  a 35% weight reduction overall (~271 lbs).

Roschin & Godin’s Converter: A Simplified SEG

In the late 1990’s, Vladimir Roschin and Sergei Godin built what’s described as a “Magnetic Energy Converter” to test a concrete proposition: that a dynamic lattice of magnetized rollers orbiting a stator without contact might couple to its environment in ways not captured by textbook electrodynamics. What they built was a essentially a simplified Searl Effect Generator: a stationary stator made from magnetized segments and a ring of magnetized rollers riding on air bearings around its rim, all skinned in conductive copper so the moving lattice of fields and eddy currents behaved like one continuous surface. They simplified John Searl’s multilayered “Law-of-the-Squares” stacks into a one-ring converter.

The stator was a circular track assembled from rare-earth magnet segments; the rotor was a necklace of magnetized rollers constrained to orbit the stator with minimal friction on air bearings. Copper foil, 0.8 millimeters thick, wrapped both stator and rollers in a continuous conductive skin. To emulate Searl’s surface “imprinting” without reproducing his full process, they inserted cross-magnetized elements oriented at ninety degrees to the base magnetization in both stator and rollers, held a one-millimeter air gap between moving and fixed elements, and chose the ratio of stator diameter to roller diameter as an integer at least twelve—geometry they said favored a resonant interaction among the moving magnetic elements.

The build was heavy: about 110 kilograms of rare-earth magnets in the stator and 115 kilograms in the rotor, all mounted on a platform that could move a centimeter vertically on springs and shock absorbers. The platform’s motion, read continuously by an inductive displacement transducer, served as their proxy for apparent weight.

Lab Setup and Procedure

A starter motor, coupled through an overrunning clutch, brought the rotor through the lower regimes and was then dropped out entirely so any “critical” behavior would be self-governed. A conventional electrodynamic generator on the same shaft fed stepwise resistive loads—ordinary electric water heaters—so the team could probe the machine’s response in one-kilowatt increments.

A high-voltage system ringed the rotor with electrodes ten millimeters from the periphery and biased the machine at up to twenty kilovolts, positive on the stator and negative on the electrodes, allowing the team to ask how electrostatic polarization altered stability and the platform trace.

The test cell was ordinary and thoroughly documented: a ground-floor room of roughly one hundred square meters with a three-meter ceiling, ferromagnetic hardware nearby, and a reinforced concrete deck overhead leading to a second floor—features they noted because field mapping would soon carry them beyond the platform.

Operating Signatures

The operating signatures arrived in sequence and, according to the authors, in repeatable fashion. At around two hundred rpm the platform trace began to deviate from baseline. Near five-hundred-fifty rpm the motor ammeter indicated current reversal and the motor was disconnected. From there the rotor continued to accelerate on its own into what the authors called a critical regime, a narrow shelf where speed stabilized just below six hundred rpm.

In that window, the team could add kilowatt-sized chunks of resistive load and watch the speed sag and recover to its preferred number, behavior that vanished if they pushed the load beyond about seven kilowatts or tried to force the speed higher. The details matter because they anchor the most controversial claim in procedure rather than lore: the weight signal tracked not just speed but also load and electrical bias in a coherent way.

Apparent Weight Reduction Effect

At the peak of their loading ladder—six to seven kilowatts drawn from the shaft—the apparent weight of the 350-kilogram platform shifted by roughly thirty-five percent relative to the initial reading. The sign could go either way—“−G” for a decrease or “+G” for an increase—and the transition exhibited hysteresis and a memory of rotation direction, as if the system retained a directional imprint when spun clockwise versus counterclockwise.

High-voltage polarization at twenty kilovolts, with negative electrodes facing the moving rollers, appeared to tighten the effect: so long as speed stayed above roughly four hundred rpm, the weight deviation persisted even as the team stepped load beyond six kilowatts. Remove the bias and the window narrowed. Add more than about seven kilowatts, and the rotor fell out of self-running altogether. The strongest behavior lived inside a bounded operating envelope, not in runaway.

“For a maximum consumed power (7 kW), a change in the total platform weight reached 35% of the initial value.”

Roschin and Godin stated that the magnetic energy converter (the active rotor + stator assembly) had a combined mass of roughly 225 kg — about 110 kg in the stator magnets and 115 kg in the rollers. A reduction in weight of 35% for this value alone would be remarkable (about 79kg) – however, the entire experimental platform, including frame, suspension, sensors, and instrumentation, totaled about 350 kg. According to their description of the experiment, “the total platform weight changed by 35 % of its initial value”, suggesting the total weight reduction was approximately 123kg, which is about 271 pounds.

Spatial Phenomena

Around that same shelf, other observables rose and fell in lockstep. In a darkened lab, a blue-pink corona hugged the millimeter gap between rollers and stator and the air took on the sharp smell of ozone—suggesting strong electric fields without the erosive pitting one would expect from arcing. A faint, high-frequency whistle emerged near onset.

These sensory cues aren’t proof of anything exotic; charged, fast-moving hardware talks to the air. But they mattered because they coincided with the quantitative signals: the back-current crossing near 550 rpm, the self-governed plateau just under 600 rpm, the stepped loads, the tightening of the weight effect under high-voltage polarization.

After runs, the researchers left the platform and walked the room with a Hall-sensor magnetometer. They reported vertical, concentric “magnetic walls,” crisp cylindrical shells of enhanced static field extending out to about fifteen meters from the device and rising through the ceiling to the floor above.

The walls were sharply bounded, with a thickness on the order of a few centimeters and spacing of roughly half a meter to eighty centimeters that widened with distance from the center. Within those same shells they measured thermal anomalies: against a room background near twenty-two degrees Celsius, a stable local cooling of six to eight degrees appeared not only around the device but within the mapped walls themselves.

The thermometer they used was sluggish, with about a ninety-second response time, yet the thermal boundaries were distinct enough, they wrote, to feel by hand when passing in and out. The picture they describe—discrete magnetic shells co-located with cool volumes—carried upstairs and even outside the building.

“This is accompanied by … a decrease in the surrounding air temperature, and the formation of concentric ‘magnetic walls’ at a distance of up to 15 m.”

Key Specifications & Experimental Details

Roschin and Godin conducted the one-ring converter tests in a Moscow laboratory at the Institute for High Temperatures and later presented their findings at the 37th AIAA Joint Propulsion Conference in 2001. The working body was roughly one meter in diameter and, together with its base, sat on a spring-isolated platform totaling about 350 kilograms.

The stator ring and free-orbiting magnetized rollers were skinned in copper and separated by an air gap near one millimeter. A starter motor with an overrunning clutch brought the rotor to the “back-current” threshold around 550 rpm, at which point the motor was disconnected and the device settled into a self-running plateau just under 600 rpm. Electrical load was drawn through a conventional electrodynamic generator in one-kilowatt steps, with stable operation reported up to roughly six to seven kilowatts.

Ring electrodes set about ten millimeters from the periphery were biased up to twenty kilovolts—positive on the stator, negative on the electrodes—an arrangement reported to stabilize the regime at higher loads. Within this window, the team observed an apparent platform-weight change as large as thirty-five percent, reversible with hysteresis and dependent on rotation direction. Additional phenomena included a blue-pink corona with an ozone odor and concentric vertical “magnetic walls” extending to roughly fifteen meters, often co-located with cool air pockets about six to eight degrees Celsius below ambient.

Interpreting the Data

For all the bold numbers, the account reads less like a free-energy manifesto than a builder’s autopsy of a stubborn operating point. The authors itemize the knobs they could turn and report how those knobs moved the needle: geometry set by a diameter ratio of at least twelve, a one-millimeter gap, cross-magnetization at right angles to the base flux, a copper skin binding each segmented surface into a single conductor, a twenty-kilovolt bias that seemed to stabilize the “−G” mode at higher loads.

They do not claim to have closed an energy balance to a skeptic’s satisfaction; rather, they argue that the machine repeatedly entered a state in which mechanical and electrical observables behaved as if a new coupling had switched on, one that might plausibly be read as a local change in gravitation or as a propulsive interaction with an unseen medium.

They explicitly note that a direct thrust measurement was not performed, and they frame their conclusions in terms of effects received—self-governed rotation, weight change, structured magnetic and thermal fields—rather than a completed theory.

Skeptical Questions

Skeptical questions follow naturally and the authors anticipated many of them. Could elastic coupling between the spinning mass and the spring platform fake a gravity signal? Could torque reactions from the generator, acoustic standing waves, electrostatic forces from the twenty-kilovolt bias, or field interactions with nearby steel explain the platform’s behavior?

Their response was to remove the motor once the critical regime appeared, to isolate the shaft with an overrunning clutch, to use discrete resistive loads, and to map magnetic and thermal fields with instruments not tied to the platform.

They also catalogued the lab’s ferromagnetic clutter and the ceiling’s reinforcement because those features could have distorted measurements; indeed, they reported that the concentric shells passed through the reinforced slab to the level above with little distortion, a detail that strengthened their conviction that something more than building magnetization was at work. Whether that interpretation is justified is precisely what a modern replication should settle.

Looking Forward

In the years since, the provenance has remained stable enough for careful readers to follow. The AIAA conference paper anchors the operating thresholds, voltages, and load protocol; the narrative versions expand on the lab setting, audible and luminous phenomena, and field mapping.

Whatever one’s priors, the numbers are specific enough that a modern replication—armed with calibrated torque sensing, synchronized power analyzers, optical encoders, and independent force channels—could decisively narrow the possibilities.

If the shells and the cooling are real, they will yield to higher-fidelity mapping; if the platform weight change is a measurement artifact riding on conventional couplings, a hardened protocol will expose it quickly. Either outcome would be valuable.

References with Hyperlinks