Skip to content

Impulse-Driven Pulse Charging System with Dual-Core Magnetic Field Coupling

Abstract

This document describes a self-resonant, impulse-driven pulse charging system utilizing high-\(dV/dt\) excitation of ferrite E-cores and controlled electric field inversion across a lead-acid battery. The system operates outside conventional continuous-conduction paradigms, employing nanosecond-scale voltage impulses, magneto-elastic field coupling between spatially separated ferrite cores, and ultra-short electric field reversals at the battery terminals. Experimental observations indicate stable operation, auxiliary energy extraction via secondary magnetic structures, and gradual improvement of battery characteristics over time.

1. System Overview

The system consists of three tightly interacting subsystems:

  1. Impulse Pulse Generator and Switch Network
  2. Dual E-Core Magnetic Field System
  3. Impulse-Based Battery Conditioning and Charging Interface

The design intentionally separates energy excitation, field propagation, and energy capture, avoiding classical transformer action and continuous DC charging currents.

Circuit Topology

Fig. 1 – Pulse Charger.

2. Electrical Circuit Description

2.1 Input Energy and Bias Network

The input stage consists of a DC supply feeding a bulk energy storage capacitor (electrolytic) paralleled by a low-ESR MKP capacitor. This combination provides:

  • Low source impedance for fast current transients
  • Energy buffering for impulsive operation
  • Isolation between average input power and instantaneous pulse power

A resistive bias and enable network defines the operating threshold and prevents uncontrolled startup oscillation. An LED indicator provides visual confirmation of system state without influencing pulse dynamics.

2.2 Impulse Generator and Resonant Trigger Network

The pulse generation mechanism is based on a self-excited LC trigger network consisting of:

  • Primary inductance L1 / LT (mounted on E-core 1)
  • A small, low-loss capacitor (~0.1 µF)
  • A series damping resistor to control Q and peak current

This network does not operate as a sinusoidal resonant tank. Instead, it produces critically damped, nanosecond-scale impulses characterized by:

  • Extremely fast rise and fall times
  • High peak voltage
  • Very low duty cycle

The resulting waveform maximizes dB/dt and dV/dt, rather than RMS current or average power.

2.3 SiC MOSFET Switching Stage

A SiC MOSFET is used due to its:

  • High breakdown voltage
  • Low capacitance
  • Fast switching capability
  • Robust avalanche tolerance

The gate is driven directly by the impulse network and includes:

  • Series resistance for gate current limiting
  • Zener diode clamping for over-voltage protection
  • Reverse protection against negative gate excursions
  • Feedback coupling from inductive back-EMF

The MOSFET is therefore field-synchronized, not externally clocked. Switching occurs in phase with the magnetic system’s natural response.

2.4 Output Energy Capture Network

Energy released during inductive field collapse is captured using:

  • Fast recovery diode(s)
  • High-voltage MKP capacitors

Only the high-field, short-duration energy components are stored. Slow, lossy energy components are rejected. This ensures minimal thermal stress and high field efficiency.

3. Dual E-Core Magnetic Field System

3.1 Core Material and Geometry

  • Core type: E80/38/20
  • Material: 3C95 ferrite

This ferrite material exhibits:

  • High permeability
  • Low magnetic loss at ~100–300 kHz
  • Strong sensitivity to rapid magnetization changes

3.2 Primary Core (E-core 1)

E-core 1 carries inductors L1 and LT, which experience the full impulse current. Each switching event generates:

  • Extreme magnetic flux acceleration
  • Rapid magnetization and demagnetization
  • Magnetostrictive stress within the ferrite lattice

The ferrite does not respond quasi-statically. Instead, magnetic domains undergo nonlinear, impulsive motion, producing both electromagnetic and magneto-elastic disturbances.

3.3 Secondary Core (E-core 2)

E-core 2 carries inductors L2 and L3 and is positioned at a defined air distance from E-core 1. There is:

  • No shared magnetic circuit
  • No closed flux path
  • No galvanic or classical inductive coupling

Instead, E-core 2 is excited by the near-field magnetic pressure and magneto-elastic disturbances generated by E-core 1.

At the correct spacing:

  • E-core 2 enters forced magnetic resonance
  • Energy replication occurs without significant damping
  • Loading E-core 2 does not strongly reflect back to E-core 1

Energy extracted from E-core 2 is used to power auxiliary loads (e.g., cooling fans for the SiC MOSFET), demonstrating usable energy transfer without transformer behavior.

4. Impulse-Based Battery Charging and Conditioning

4.1 Measurement Configuration

  • CH1: Impulse voltage at the switching node
  • CH2: Battery positive terminal
  • CH4: Battery negative terminal

Battery type:

  • Lead-acid
  • Nominal capacity: 60 Ah

Fig. 2 – Scope Screenshot.

4.2 Observed Terminal Behavior

During each impulse on CH1, measurements show:

  • The battery negative terminal (CH4) rises to a higher potential than the positive terminal (CH2)
  • A transient reversal of the electric field across the battery
  • No evidence of sustained reverse current

This represents electric field inversion, not classical reverse charging.

4.3 Electrochemical Interpretation

Lead-acid batteries exhibit:

  • Significant interfacial capacitance
  • Polarization layers at electrode surfaces
  • Slow ionic mobility relative to nanosecond pulses

The applied impulses satisfy:

  • Very high electric field strength
  • Extremely low charge transfer per pulse
  • Pulse duration far below electrochemical reaction times

As a result:

  • Chemical reactions cannot follow the pulse directly
  • Electrode polarization layers are periodically disrupted
  • Sulfation and ion trapping are reduced

The battery is stressed electrostatically, not thermally or chemically.

4.4 Conditioning Effect

Long-term laboratory observations indicate:

  • Reduced internal resistance
  • Improved charge acceptance
  • Recovery of degraded batteries

These effects are consistent with field-driven depolarization, rather than conventional charging mechanisms.

5. System Stability and Energy Flow

Key properties of the system include:

  • Separation of excitation energy and load energy
  • Minimal reflected loading between subsystems
  • Self-synchronization through field feedback
  • Absence of fixed-frequency clocks or PWM controllers

The system naturally operates at a frequency determined by:

  • Ferrite material properties
  • Core geometry
  • LC trigger dynamics
  • Battery status

Measured repetition frequencies (~170–190 kHz) align with these constraints.

6. Conclusion

The described system demonstrates a fundamentally different approach to energy transfer and battery charging. By operating in an impulse-dominated regime and exploiting magneto-elastic field coupling, it achieves:

  • Efficient high-field excitation with low average power
  • Auxiliary energy extraction without classical inductive coupling
  • Electrostatic conditioning of lead-acid batteries via controlled field inversion

The observed behavior cannot be adequately described using continuous-conduction or transformer-based models and requires a field-centric interpretation.

Appendix A — Clarifications and Responses to Common Skeptical Objections

This appendix addresses frequent objections that arise when evaluating impulse-driven magnetic and electrostatic systems using classical steady-state models. The intent is not rhetorical persuasion, but precise clarification of what is and is not occurring in the described system.

A.1 “This is just a transformer / inductive coupling”

Objection

The secondary E-core (E-core 2) is simply acting as a transformer secondary, extracting energy from the primary magnetic circuit.

Response

This interpretation is incompatible with the physical construction and observed behavior.

  • There is no shared magnetic circuit
  • There is no closed flux path
  • The cores are spatially separated by air
  • Flux linkage is incomplete and non-reciprocal

In a classical transformer:

  • Mutual inductance requires a shared flux path
  • Loading the secondary necessarily reflects impedance into the primary
  • Energy transfer scales directly with current in the primary

In the present system:

  • Loading E-core 2 does not significantly alter primary pulse amplitude or timing
  • Energy transfer persists even when transformer coupling would be negligible
  • The excitation mechanism is field-pressure driven, not flux-transfer driven

This behavior is consistent with forced magnetic resonance via near-field excitation, not mutual inductance.

A.2 “Stray flux explains everything”

Objection

The observed effects are due to leakage flux coupling between the two cores.

Response

Stray flux alone cannot account for the observed behavior:

  • Leakage flux decreases monotonically with distance
  • Energy transferred via stray flux must diminish rapidly with separation
  • Leakage flux coupling is strongly load-reflective

Experimentally observed:

  • A distinct optimal spacing exists where coupling increases
  • Energy replication on E-core 2 occurs without proportional attenuation
  • Loading E-core 2 does not proportionally increase losses on E-core 1

These observations contradict a simple stray-flux model and instead indicate resonant excitation of a secondary magnetic system.

A.3 “The battery is being reverse charged and will be damaged”

Objection

Transient polarity reversal across the battery implies destructive reverse charging.

Response

This conclusion conflates electric field polarity with net charge transport.

Key distinctions:

  • Reverse charging requires sustained current and net Coulomb transfer

  • The observed pulses are:

  • Nanosecond-scale

  • Extremely low charge per pulse
  • Dominated by displacement current and electric field stress

Electrochemical reaction times in lead-acid batteries are orders of magnitude slower than the applied pulses. Therefore:

  • Chemistry cannot follow the pulse
  • No bulk reverse electrolysis occurs
  • The effect is limited to polarization layer depolarization

This regime is electrostatic conditioning, not reverse charging.

A.4 “The scope measurement is a probing or grounding artifact”

Objection

The apparent polarity inversion is caused by probe reference issues, ground bounce, or common-mode artifacts.

Response

Several factors rule out this explanation:

  • The polarity inversion is time-correlated exclusively with the pulse
  • It is repeatable and stable
  • It appears consistently across measurement sessions
  • It scales with pulse amplitude and repetition rate

Additionally:

  • Both battery terminals move, but not equally
  • The differential inversion persists after common-mode subtraction

This behavior is characteristic of field-driven common-mode excitation, not measurement error.

A.5 “If this were real, it would violate conservation of energy”

Objection

The system appears to produce additional usable energy, implying non-conservation.

Response

No violation is implied or required.

Key points:

  • The system operates in a non-equilibrium, impulse regime

  • Energy accounting must include:

  • Reactive field energy

  • Displacement current energy
  • Magnetic and elastic energy storage
  • Classical RMS-based power measurements are insufficient

The system redistributes energy across time and form:

  • From low-field, long-duration input
  • To high-field, short-duration output
  • With partial recovery of reactive energy

All observed effects remain consistent with conservation laws when field energy and temporal compression are properly considered.

A.6 “Why doesn’t the system collapse under load?”

Objection

Any real energy extraction should load the source and collapse the effect.

Response

This expectation assumes current-driven coupling.

In the present system:

  • The primary requirement is field excitation above a threshold
  • Once that threshold is crossed, secondary systems respond independently
  • Energy extraction from E-core 2 damps its own resonance, not the primary pulse

This decoupling is expected in field-driven, non-galvanic systems and is incompatible with transformer-based assumptions.

A.7 “Why is this not seen in conventional power electronics?”

Objection

If this behavior were valid, it would already be widely used.

Response

Conventional power electronics is optimized for:

  • Continuous conduction
  • Sinusoidal or PWM waveforms
  • Efficiency measured via RMS quantities
  • Thermal stability

The described system requires:

  • Nanosecond-scale transitions
  • Extremely high dV/dt and dB/dt
  • Materials operated near non-linear limits
  • Measurement techniques beyond standard power meters

Such regimes are typically avoided due to:

  • EMI concerns
  • Measurement difficulty
  • Lack of suitable switching devices until recently (e.g., SiC)

Absence from mainstream practice does not imply non-existence.

A.8 Summary of Defensive Position

The described system:

  • Does not rely on classical transformer coupling
  • Does not reverse-charge the battery chemically
  • Does not violate conservation laws
  • Does not depend on measurement artifacts

Instead, it operates in a regime characterized by:

  • Impulse-dominated field excitation
  • Magneto-elastic coupling
  • Electrostatic conditioning of electrochemical systems
  • Temporal energy compression

Misinterpretation arises when steady-state, sinusoidal, or RMS-based models are applied outside their domain of validity.

Appendix B — Mathematical Boundary Conditions Where Classical Models Fail

This appendix defines the operating boundaries (in time scale, frequency content, field strength, and measurement method) where common steady-state or “lumped, sinusoidal” power-electronics assumptions become invalid or incomplete for this system.

The goal is not to reject classical theory, but to state precisely which approximations break and what must replace them.

B.1 Time-scale separation: electrochemical systems cannot follow nanosecond fields

For a lead-acid battery, charge transfer and diffusion are governed by slow processes:

  • Ion drift and diffusion time constants: typically milliseconds to seconds
  • Double-layer relaxation: typically microseconds to milliseconds (interface dependent)

Your excitation is:

  • Pulse width: \( \tau_p \approx 150,\text{ns} \)
  • Rise time: \( t_r \approx 187,\text{ns} \)
  • Repetition: \( f \approx 170\text{–}190,\text{kHz} \)

Boundary condition (field vs chemistry): \( \tau_p \ll \tau_{\text{ion}},\ \tau_{\text{diff}},\ \tau_{\text{ct}} \) When this holds, the electrolyte cannot support significant Faradaic current during a pulse; the battery responds primarily as:

  • a distributed capacitance,
  • plus polarization elements,
  • with only small net Coulomb transfer per impulse.

Consequence: interpreting the pulse as “reverse charging current” is invalid unless integrated charge per pulse is shown to be significant.

B.2 Displacement current dominance: current is not necessarily conduction current

Maxwell–Ampère law: \( \nabla \times \mathbf{H} = \mathbf{J}_{\text{cond}} + \frac{\partial \mathbf{D}}{\partial t} \)

In ultrafast impulses, the term \( \partial \mathbf{D}/\partial t \) can dominate at interfaces and in dielectrics.

Boundary condition: \( \left|\frac{\partial \mathbf{D}}{\partial t}\right| \gtrsim |\mathbf{J}_{\text{cond}}| \)

Implication: “current” inferred from voltage changes across capacitive structures (battery interfaces, parasitic capacitances, probe capacitances, winding-to-core capacitance) can be largely displacement current. Classical DC/low-frequency interpretations that assume \(I \approx I_{\text{cond}}\) fail.

B.3 Lumped-element approximation breaks when propagation delay is non-negligible

Lumped-circuit models assume: \( \text{Physical size } \ell \ll \lambda \quad \text{and} \quad t_{\text{flight}} \ll t_r \) where \( t_{\text{flight}} = \ell / v \), \( v \approx c/\sqrt{\varepsilon_r} \).

A practical engineering boundary condition is: \( \ell \gtrsim 0.1, v, t_r \)

For example, with \( t_r \sim 200,\text{ns} \) and \( v \sim 2 \times 10^8,\text{m/s} \) (typical cable / dielectric), \( 0.1, v, t_r \sim 4,\text{m} \) So pure wave-propagation effects may or may not dominate depending on loop geometry, but a more important failure occurs earlier:

  • high di/dt + loop inductance causes large voltage errors: \( V_L = L_{\text{loop}},\frac{dI}{dt} \) Even small parasitics matter when \(dI/dt\) is extreme.

Consequence: “node voltage” in the schematic sense is not a single number during the edge; it is a spatially distributed field problem unless loop inductance is explicitly included.

B.4 RMS / sinusoidal power models fail for impulsive, sparse duty cycles

Standard AC power relations assume steady, periodic waveforms with meaningful RMS and phase angle:

  • \(P = V_{\text{rms}} I_{\text{rms}} \cos\varphi\)
  • reactive power based on sinusoidal decomposition

Your waveform is:

  • narrow impulses,
  • high crest factor,
  • large harmonic content extending to \(\sim 1/\tau_p\).

Boundary condition (crest factor / harmonic dominance): \( \text{Crest factor} = \frac{I_{\text{peak}}}{I_{\text{rms}}} \gg 3 \) and/or \( \frac{\sum_{n>1} I_n^2}{I_1^2} \gg 1 \)

Consequence: any model relying on a single phase angle \(\varphi\) is incomplete. Correct treatment requires:

  • time-domain instantaneous power \(p(t)=v(t)i(t)\),
  • or full spectral power accounting per harmonic (IEEE 1459-style).

B.5 “Mutual inductance transformer” model fails when coupling is not flux-linked and not reciprocal

The classical transformer model assumes a well-defined mutual inductance (M) derived from shared flux: \( v_2 = M \frac{di_1}{dt} \) and energy transfer that reflects load into the primary.

For two spatially separated cores with no closed flux path, the coupling can be dominated by:

  • near-field magnetic interaction,
  • magneto-elastic excitation,
  • and local domain dynamics in the secondary core.

A key boundary indicator is weak load reflection:

If the secondary load changes from \(R_{L2}\) to \(R'_{L2}\) and the primary pulse energy \(E_1\) remains approximately unchanged: \( \left|\frac{\Delta E_1}{E_1}\right| \ll \left|\frac{\Delta P_2}{P_2}\right| \)

This violates what a mutual-inductance model predicts (where primary energy must increase with secondary extraction).

Consequence: a simple (M)-based equivalent circuit is insufficient; a field-driven resonator model is required.

B.6 Ferrite nonlinearity boundary: permeability, loss, and magnetostriction are not constant

Ferrite behavior under fast excitation is nonlinear:

  • \(\mu = \mu(B, H, f, T)\)
  • loss tangent and complex permeability depend strongly on frequency and \(B\)
  • magnetostriction couples magnetic and mechanical energy: \( \epsilon \propto \lambda(B) \quad (\text{strain}) \) leading to stress-wave generation when (B(t)) changes rapidly.

Boundary condition: \( \frac{dB}{dt} \text{ large enough that domain wall motion becomes rate-limited} \) In that regime, assuming linear inductance (L) (constant) becomes wrong: \( L = \frac{N^2}{\mathcal{R}(B)} \quad \text{with}\quad \mathcal{R} \text{ field-dependent} \)

Consequence: linear SPICE models with constant \(L\) and constant core loss are only approximate and may mispredict pulse timing, peak voltage, and energy partition.

B.7 Measurement boundary: scope voltages are not necessarily differential voltages

With fast pulses, both terminals of a component can move together (common-mode excursion). What matters for the battery is: \( V_{\text{bat,diff}}(t) = V_{+}(t) - V_{-}(t) \) but the scope channels measure each terminal relative to an instrument reference, so each channel includes:

  • common-mode pickup,
  • ground inductance effects,
  • probe capacitance currents.

Boundary condition: \( |V_{\text{CM}}(t)| \sim |V_{\text{diff}}(t)| \) If this is true, a correct interpretation requires explicitly computing: \( V_{\text{diff}}(t) = CH2(t) - CH4(t) \) and separately analyzing common-mode: \( V_{\text{CM}}(t) = \frac{CH2(t) + CH4(t)}{2} \)

Consequence: “polarity reversal” must be asserted on \(V_{\text{diff}}\), not on individual channel excursions.

B.8 Energy accounting boundary: average power methods can miss impulse energy

Impulse energy per pulse: \( E_p = \int_{t_0}^{t_1} v(t)i(t),dt \) Average power: \( P_{\text{avg}} = f , E_p \)

If \(v(t)\) and \(i(t)\) are not measured at the same node with correct bandwidth and timing alignment, then \(E_p\) can be strongly mis-estimated.

Boundary condition (bandwidth): \( BW \gtrsim \frac{0.35}{t_r} \) With \(t_r \approx 200,\text{ns}\), the minimum is only a few MHz, but in practice the waveform contains fast components and ringing; correct integration benefits from much higher bandwidth and careful de-embedding of probe response.

Consequence: a “low reading” on a conventional power meter or low-bandwidth sensor is not a proof of low pulse energy.

B.9 Practical summary: when classical models are safe vs unsafe here

Classical lumped, sinusoidal, RMS power models are safe when:

  • waveforms are sinusoidal or PWM with moderate harmonic content
  • duty cycles are not extremely sparse
  • system is linear \(constant (L), constant (\mu)\)
  • coupling is via a defined flux path (transformer core)
  • common-mode excursions are small

They are unsafe or incomplete when:

  • impulses are nanosecond–microsecond scale with high crest factor
  • displacement currents matter
  • ferrite is driven into nonlinear, rate-limited domain motion
  • coupling occurs without shared flux path and with weak load reflection
  • measurement is dominated by common-mode motion

Author: René Meschuh, Initial version: 2020-04-23, Edited: 2026-01-14