Bedini-John: Primer

John Bedini's Simplified School Girl (SSG) pulse motor, the "radiant energy" battery charger, and how to audit a "battery COP" claim.

Source tags: [read] = from a document in the project folder; [web] = Internet source, URL given; [memory] = recalled from training, with a confidence level. Verdicts come from arbiter.md. Numbers in section 5 were recomputed by me and reproduced by example.py.

Source gap (repeated from claims.md and arbiter.md). The folder holds no Bedini book, no patent text, no lab notes and no measured data. The only substantial readable works are the 2008 BM3 Group build manual (volunteer-written, no results in it) and a saved web page about a science-fair motor. Most other files are diagrams. Everything below about whether the effect exists rests on web sources (the patent page, one 2018 journal paper, a battery datasheet) or on [memory], not on data in the folder. To go further, get: Bedini, Bearden and Bedini, Free Energy Generation: Circuits and Schematics (Cheniere Press, 2006); the full text of US 6,545,444; the BM3 Yahoo group posts (mostly lost when Yahoo Groups closed).

Correction to arbiter.md found while writing this primer (now adopted in arbiter.md, "QE revision", section 1). The arbiter originally read Table 2 of the Sriphan et al. (2018) paper as giving the primary (supply) battery's power, and from it derived a "whole-system ratio of about 0.63". I re-extracted the paper [web: http://gratisenergi.se/mgbedini.pdf] and checked the arithmetic. Table 2's "running battery 12.71 W" is exactly the average discharge power of the charged battery in Table 3, and its "charging battery 4.94 W" is exactly the average charging power in Table 3. Both numbers are on the same (charge) battery. The 0.63 is roughly the ratio of discharge time to charge time, not a system efficiency. The paper reports no measurement of the primary battery at all. So the arbiter's "single most useful finding" does not stand; its verdicts do not change, because none of them rested on that number alone (Claim 2's whole-system verdict was already Never properly tested). Section 5 walks through this.

Updated after quality review (QE). All 17 diagrams in the folder have now been read from their page images, and the manual's test pages (pp.19-24) were read in depth. This added: the only builder-recorded input/output numbers in the folder (section 5, step 7), the manual's "1-ohm resistor test" and matched-bulb power method (sections 3 and 5, step 8), the manual's own conventional-charger comparison (section 7), the exact wording of the "backpopping" note (section 4), and corrected dates for the circuit lineage (section 6). No verdict changed [read: arbiter.md, QE revision].

---

1. The idea in one paragraph

John Bedini (1949-2016 [memory: medium]) was an American audio-electronics designer who built small motors whose rotor is a wheel of magnets spun by short electrical pulses into a coil. Each time the switch cuts the coil off, the coil throws a high-voltage spike, and Bedini sent those spikes into a second battery to charge it. His followers, notably the BM3 (Bedini Monopole 3) Yahoo group, wrote a build manual for a version called the Simplified School Girl (SSG) circuit. They said the second battery is charged mostly by "radiant energy" from the vacuum rather than by ordinary current, that it can later deliver more energy than went into it, and that the coil pulse "opens the Bloch wall" of the magnets to let this energy in [read: manual pp.4-5, 20-21]. The mainstream view is that the SSG is a real, working, fairly ordinary flyback (inductive-kick) charger, and that the "more out than in" result comes from where the energy accounting draws its boundary: it counts the battery being charged and deliberately leaves out the battery that powers the machine. The interesting part for a learner is that the claim is cheap to test properly, and to my knowledge nobody has published a neutral, whole-system test.

2. What you need to know first

Energy, power and the joule. Power P (watts, W) is the rate of energy flow; energy E (joules, J) is power times time. For electricity, P = V I, where V is voltage (volts) and I is current (amperes, A). If V and I change with time, the energy over a time T is E = ∫₀ᵀ v(t) i(t) dt, the area under the instantaneous power curve. This integral is the only correct way to count energy into or out of anything driven by pulses. Read: Halliday, Resnick and Walker, Fundamentals of Physics, the chapters on current and circuits.

Charge and ampere-hours. Charge Q (coulombs, C) is current times time. Battery makers quote charge in ampere-hours (Ah): 1 Ah = 1 A for 1 hour = 3600 C. Charge is not energy. To get energy from charge you multiply by voltage: E ≈ V Q. A 12 V, 17 Ah battery holds about 12 x 17 x 3600 ≈ 734 kJ.

Inductors and flyback. A coil (inductor) of inductance L (henries, H) resists changes in its current. The voltage across it is v = L di/dt. A current I flowing in the coil stores energy E = ½ L I² in its magnetic field. If a switch suddenly cuts the current, di/dt becomes huge and negative, so the voltage spikes (hundreds or thousands of volts) until some path, here a diode into a battery, lets the current keep flowing and the stored energy drain away. This is the "inductive kick", "flyback" or "back EMF" (EMF = electromotive force, a voltage generated inside a component). Car ignition coils and the flyback converters in phone chargers work this way. Read: Griffiths, Introduction to Electrodynamics, chapter 7 (inductance and magnetic energy); Horowitz and Hill, The Art of Electronics, the section on inductive kick.

Average versus RMS (root mean square) values. For a current i(t) that repeats with period T, the average is Ī = (1/T) ∫₀ᵀ i dt and the RMS value is I_rms = √((1/T) ∫₀ᵀ i² dt). The average is what moves charge into a battery (charge = Ī x time). The RMS is what heats a resistor or a lamp filament (heat power = I_rms² R, R the resistance in ohms). For steady DC they are equal. For short pulses they are not: a current that is 10 A for 10% of the time and zero otherwise has Ī = 1 A but I_rms = 10 x √0.1 ≈ 3.2 A. A cheap DC meter reads roughly the average; a lamp's brightness follows roughly the RMS. Read: Horowitz and Hill, The Art of Electronics, chapter 1.

Lead-acid batteries. A lead-acid battery stores energy chemically. Three facts matter here. (i) Coulombic efficiency (charge out / charge in) is below 1, typically 85-95%, and energy efficiency (energy out / energy in) is lower, roughly 70-85% [web: https://www.sciencedirect.com/topics/engineering/ah-efficiency, as cited in mathematician.md]. (ii) The resting (open-circuit) voltage indicates the state of charge (SOC, the fraction of full charge remaining): the BM3 manual itself gives about 12.65 V full and 12.0 V empty [read: manual p.12]. That whole range is only 0.65 V, so small voltage errors are large SOC errors. (iii) A new battery's usable capacity usually rises over its first several cycles ("conditioning"); the manual says so too [read: p.12]. Read: Linden and Reddy, Handbook of Batteries, the lead-acid chapters.

Conservation of energy and system boundaries. Energy is conserved for any closed boundary you draw: energy in = energy out + change in stored energy + losses. A "coefficient of performance" (COP) is just a ratio of energy out to energy in across some boundary. Change the boundary and you change the ratio. The textbook COP above 1 is a heat pump: it moves 3 J of heat for 1 J of electricity because the other 2 J come from outside air, which is inside the energy balance once you draw the boundary around air and pump together. Read: Schroeder, An Introduction to Thermal Physics, the chapter on engines and refrigerators.

Magnetic domains (needed for claim 4). A permanent magnet is made of regions (domains) in which the atomic magnetic moments point the same way. The thin boundary between two domains is a domain wall; one common type is a Bloch wall. A wall costs energy; it is not a source. Read: Kittel, Introduction to Solid State Physics, the ferromagnetism chapter.

Zero-point energy (needed for claim 4). In quantum mechanics a harmonic oscillator of angular frequency ω (radians per second) cannot have zero energy; its lowest level is E₀ = ½ ħω, where ħ = h/2π is the reduced Planck constant. This "zero-point energy" is the floor. You cannot extract energy from a system already in its lowest state, because there is no lower state to drop into. Read: Griffiths, Introduction to Quantum Mechanics, chapter 2 (harmonic oscillator).

3. The textbook view

The SSG circuit as an engineer reads it. From the manual's parts list and schematic [read: manual; Circuit diagram1.gif page image]: a bicycle rim with ceramic magnets all facing outward with the same pole; a coil wound with two windings (a thick "power" winding and a thin "trigger" winding, about 850 turns, on a core of cut welding rod); one power transistor (2N3055 or MJL21194); a resistor and potentiometer to set the trigger current; diodes 1N4001/1N4007; a neon bulb; a 12 V primary battery and a 12 V charge battery.

Operation, one pulse: 1. A magnet approaches the coil and induces a small voltage in the trigger winding. This switches the transistor on. 2. Current I rises in the power winding, drawn from the primary battery. The coil's field pushes the magnet onward, so the wheel keeps spinning. Energy now stored in the coil: E = ½ L I², where L is the power winding's inductance and I the current at switch-off. 3. As the magnet passes, the trigger voltage reverses and the transistor switches off. The coil voltage leaps (v = L di/dt) until the diode to the charge battery conducts. The stored ½ L I² flows into the charge battery as a brief, high-voltage current pulse. The neon bulb, which strikes near 90 V [read: Circuit diagram1.gif], protects the transistor if no battery is connected.

Energy balance per pulse (symbols: E_primary is the energy drawn from the primary battery during the on-time, E_mech the energy given to the wheel, E_loss the heat in wire, transistor and diode, E_charge the energy delivered to the charge battery):

E_primary = E_mech + E_loss + E_charge, with E_charge ≤ ½ L I².

The wheel's kinetic energy is roughly constant at steady speed, so E_mech is ultimately friction and air drag. Every joule that reaches the charge battery came from the primary battery. Whole-system efficiency is

η_system = (energy recoverable from charge battery) / (energy drawn from primary battery) < 1,

and with battery losses on top it will typically be well below 1.

What permanent magnets can do. A static magnet's field is conservative for a rotor moving through it: over one full turn the magnet does zero net work on the rotor, unless the magnet's own magnetization changes, which would show up as the magnet weakening over time [memory: high].

Pulse charging is real. Charging a lead-acid battery with pulses is a legitimate technique, and pulse "desulfators" are a commercial product category, though evidence for their benefit is mixed [memory: medium]. So "it charges batteries well" and "it charges batteries from nowhere" are different claims.

Measuring pulsed power: why the manual's methods mislead. The manual offers two ways to measure the charging power [read: manual p.21]. Method 1: an analog ammeter in series with the charge battery, read once, multiplied by the battery voltage, then removed "before the start of the charge cycle". Method 2 (matched bulbs): replace the charge battery with a small bulb, run an identical bulb from the primary battery through a potentiometer and ammeter, turn the potentiometer until the two glow equally by eye, and take volts x amps on the reference bulb as the SSG's output power. The textbook objection to Method 2 is the average-versus-RMS point from section 2. A filament glows according to the heat it receives, which goes as I_rms² R; a battery charges according to the average current Ī. For spiky pulses I_rms is well above Ī, so a bulb that glows like a 100 mA DC bulb is receiving much less than 100 mA of average current. Brightness matching by eye is also coarse, because light output rises steeply and nonlinearly with filament temperature. And the bulb replaces the battery, so the circuit runs at a different operating point. Neither method measures the primary battery: the manual says "just try and forget it" [read: p.21]. The correct method is to sample v(t) and i(t) fast enough to see the pulses and integrate ∫ v i dt.

4. Where this researcher departs

The exact point of departure is the system boundary. Everything up to "the coil spike charges a battery" is textbook. The departure is to draw the energy balance around the charge battery only, find out > in for that battery, and conclude that an outside source exists. The textbook reply is that the charge battery's "out > in" can only mean the battery started fuller than it finished, or that the "in" was mis-measured, because a battery stores and releases energy; it does not make it. And the primary battery, which really is the source, has been left out of the account on purpose.

Bedini's patent, US 6,545,444 (granted 8 April 2003), describes the machine in conventional back-EMF terms but also uses Bearden's language of an "open dissipative system ... not in thermodynamic equilibrium with the active vacuum" [web: https://patents.google.com/patent/US6545444B2/en]. So the vacuum-energy reading is the inventor's own, not only his followers'.

5. Worked example: auditing the one published COP table

There is no data in the folder, so the best real numbers available are from the only peer-reviewed paper on an SSG-type machine: Sriphan et al., "Coefficient Performance of Battery Running and Charging by Magnet ...", Journal of Electrochemical Energy Conversion and Storage 15(4), 041002 (2018) [web: http://gratisenergi.se/mgbedini.pdf]. They built a six-pole, eight-neodymium-magnet Bedini generator, charged a 12 V 17 Ah sealed lead-acid battery, then discharged it into a load, and reported COP = (energy out of the charged battery) / (energy into it). Every step below is reproduced in example.py. Steps 7 and 8 then use the few numbers that are in the folder.

Symbols: I_in = charging current (A), V̄_in = average of resting voltages before and after charging (V), T_in = charge time (s); I_out, V̄_out, T_out the same for discharge. E_in = V̄_in I_in T_in and E_out = V̄_out I_out T_out (J). Q = I T / 3600 is charge in Ah.

Step 1. Recompute the paper's own arithmetic (Table 3, original design, row 1).

• Charge: P_in = 12.58 V x 0.485 A = 6.10 W. E_in = 6.10 W x 39,600 s = 241,600 J (paper: 241,560 J).
• Discharge: P_out = 12.62 V x 1.01 A = 12.75 W. E_out = 12.75 W x 25,200 s = 321,200 J (paper: 321,300 J).
• COP = 321,200 / 241,600 = 1.33 (paper: 1.33).

All four original-design rows reproduce (1.33, 1.95, 1.88, 1.34; average 1.63). So the arithmetic is honest. Note, though, that if you used the BM3 manual's table formula literally (Input / Output), row 1 would give 0.75: the manual's printed formula would turn this "success" into a loss.

The replication (DC-DC boost) design does not reproduce: its own joule columns give COP 2.90, 3.82, 3.35, 3.34, while the printed COP column says 2.13, 2.19, 1.73, 1.44. Something in that table is mislabelled or mis-entered. That alone is a reason not to lean on it.

Step 2. Ask what boundary the COP is drawn around. Both E_in and E_out are measured at the terminals of the charge battery. The machine's power source, the primary battery, does not appear. A COP above 1 here says: "this battery gave out more energy than the current meter showed going in." It says nothing yet about the machine.

Step 3. Check whether Table 2 supplies the missing number. Table 2 lists "Running battery average power consume 12.71 W" and "Charging battery average power consume 4.94 W" for the original design. It is tempting to read "running battery" as the primary. But average the four rows of Table 3:

• mean P_out = (12.75 + 12.50 + 12.75 + 12.85) / 4 = 12.71 W
• mean P_in = (6.10 + 4.80 + 4.06 + 4.80) / 4 = 4.94 W

Exact matches. Table 2 is a summary of Table 3; "running" means the charged battery running a load. The primary battery's draw is not reported anywhere in the tables. (The quantity 1.63 x 4.94 / 12.71 = 0.63 that the arbiter used is just the ratio T_out/T_in, about 0.5-0.75 row by row; it is not a system efficiency.) The replication's Table 2 figures (23.68 W, 8.02 W) do not match its Table 3 averages (5.88 W, 1.08 W), another unexplained inconsistency.

Step 4. Charge balance: can a battery return more ampere-hours than it received? Row 1:

• Q_in = 0.485 A x 39,600 s / 3600 = 5.34 Ah
• Q_out = 1.01 A x 25,200 s / 3600 = 7.07 Ah
• Q_out / Q_in = 1.325.

A lead-acid battery has coulombic efficiency below 1. So 1.73 Ah more came out than went in. Either the battery started this cycle fuller than it ended (state of charge not matched), or the charging current was under-read (an average or analog meter on a spiky pulse current is a known weak point; the manual itself calls pulsed current "a little trickier than normal to measure" [read: p.21]), or there is a real anomaly.

How big a state-of-charge mismatch would explain the whole surplus? Energy surplus = 321,200 - 241,600 = 79,600 J. A full 17 Ah battery at about 12.6 V holds 12.6 x 17 x 3600 = 771,000 J. So 79,600 J is 10.3% of the battery's capacity. On the manual's own scale (12.0 V empty to 12.65 V full), 10% of SOC corresponds to about 0.067 V of resting voltage. The paper reports only the average of start and end voltages for each half-cycle, not whether the battery came back to the same voltage at the end of the cycle. A 67 mV drift, below what many hobby meters resolve reliably at 12 V, would account for the entire "COP 1.33". Allowing for a realistic 90% coulombic efficiency, the battery would need to start about 2.3 Ah (13%) fuller than it finished.

This is not proof that drift happened. It shows the size of the effect is within the measurement's blind spot, which is why the test needs matched start and end states.

Step 5. What the primary battery must have supplied (illustrative). The paper does not give coil inductance or primary current, so this step uses an assumed value, clearly labelled. At 1300 rpm with 8 magnets, and assuming one pulse per magnet pass, the pulse rate is f = 1300/60 x 8 = 173 pulses per second. To deliver row 1's 6.10 W into the charge battery, each pulse must carry 6.10 / 173 = 35 mJ. With an assumed L = 20 mH, ½ L I² = 0.035 J needs I = √(2 x 0.035 / 0.020) = 1.88 A at switch-off, drawn from the primary battery on every pulse, plus whatever drives the wheel and heats the transistor. That is the term the COP leaves out. Measuring it is one more current probe.

Step 6. Units check on the 9 V science-fair motor. The saved web page says a 9 V battery ran the motor for 5 days (120 h) at 22 mA, or 13.2 mA with the generator coil loaded [read]. Charge required: 0.022 A x 120 h = 2.64 Ah (1.58 Ah at 13.2 mA). A 9 V alkaline battery holds about 0.6 Ah at a 25 mA drain [web: https://data.energizer.com/pdfs/la522_ap.pdf, via the arbiter]; the page itself says 400 mAh. That is a shortfall of 4.4-6.6x at 22 mA, or 2.6-4.0x at 13.2 mA. A 600 mAh battery would last 27 h at 22 mA. The page's own editor doubted the account [read]. The likelier explanation is a reporting or measurement error (for example, an averaging meter on a pulsed current, or a battery swap), but no log exists to check.

Step 7. The only input/output pair in the folder. One drawing, "SCH-illust. 2coil 4inch" (a 30 Dec 2005 redraw by Erwin Badertscher of an SSG "replication"), carries a handwritten block: "2 COILS - SWEET SPOT", "15V IN AT 2.2 AMP", "CHARGING 0.3 AMP OUT" [read: page image p.10; deep/-5058f1.md]. The drawing shows a bench power supply ("PS") as the source, not a battery. Input power: P_in = 15 V x 2.2 A = 33 W. The page gives no charge-battery voltage; assume a typical 12.5 V while charging (an assumption, not on the page). Output power: P_out ≈ 12.5 V x 0.3 A = 3.75 W. Ratio: 3.75 / 33 ≈ 0.11. Even if every unknown is taken in the SSG's favour (say the charge battery sat at 15 V), the ratio is 4.5 / 33 ≈ 0.14. This is one uncontrolled builder note with no duty cycle or meter type stated, so it is a hint, not evidence; but it is the only place in the folder where someone wrote down the primary side, and it points the conventional way. (The redrawer also wrote "SCALAR ENERGY" across the sheet; the numbers on it do not support the label.) A second sheet, N-Pole-Motor-6b (Oct 2005), records "CHARGING RESULT: SCR -> 13.51V -> 0.8A -> 0.2A" and "5-IN4007s -> 12.96V -> 1.0A -> 0.25A" [read: page image p.8]; the arrows are not explained, so it cannot be turned into an efficiency.

Step 8. What the manual's "1-ohm resistor test" can and cannot show. The procedure [read: manual p.19]: remove the charge battery, put a resistor of about 1 Ω across its terminals, run the machine, and pass if the meter reads 1 V or less and the resistor stays cool; if not, retune the base resistor. Ohm's law (I = V/R) and P = V²/R give the bound: at 1 V across 1 Ω, an average current of at most about 1 A and, for steady DC, 1 W. The manual then concludes that conventional energy "is not enough to account for the observed charging" [read: p.20], but gives no number linking the test to the charge battery. Three problems, each checkable with the arithmetic above: - The output is pulsed. If the meter reads the average, 1 V average can hide much more heating. Example: pulses of 10 A at 10% duty give 1 A average (1 V on the meter) but I_rms ≈ 3.2 A and heat = 3.2² x 1 = 10 W. Only the "resistor not hot" clause guards against this. - A resistor is not a battery. With a 12-14 V battery in place, the coil must push its current against that voltage, so the circuit's operating point is different. Even 1 A average into a 12.5 V battery would already be 12.5 W of ordinary charging power. - The operator tunes until the meter reads 1 V or less, so the 1 V figure is set by the procedure, not measured from the charging. So the test is a sensible tuning check (it keeps the transistor from overheating), but it carries no information about "two kinds of energy".

What the example teaches. The published numbers are arithmetically sound for the original design, and they are compatible with ordinary physics once you notice (a) the boundary excludes the source and (b) a 10% state-of-charge drift is enough to produce COP 1.33. They neither prove nor refute an anomaly, because the measurements that would decide it were not taken. The folder's own scraps of primary-side data (step 7) and its "proof" test (step 8) do not change that picture.

6. Evidence table

No reviewer or arbiter found any neutral laboratory test of the SSG, positive or negative. None of the claims has been "Tested, not supported", and none was "dismissed without testing" in the sense of a refused experiment; the protocol simply never measured the source.

7. What a learner should take away

Solid. The SSG is a real, buildable flyback motor-charger. Its spike, its neon protection, its pulse charging and its energy recovery are all textbook and worth building to see an inductor's kick for yourself. Ceramic-magnet tuning sensitivity is real.

Wrong, and why the mistake is instructive. The over-unity inference fails at the boundary, not in the circuit. "More out than in" for a storage device is a statement about its state of charge, and the protocol never pins that state down. Three classic measurement errors stack on top: resting voltage used in place of charging voltage, a single meter snapshot used in place of a time integral of a pulsed current, and a COP formula printed upside down. Each is a common error in amateur and professional energy measurement, and recognising them is a transferable skill. Also instructive: the one peer-reviewed paper looks like a whole-system test until you average its columns and find that every power figure is on the charge battery. Check what was measured before reading what was concluded. (The arbiter in this review made the same misreading, which shows how easy it is.) A fourth error is worth learning on its own: judging pulsed power by lamp brightness or by an averaging meter. Brightness follows the RMS current, charging follows the average current, and for spikes the two differ by a large factor (section 2 and section 5, step 8).

Genuinely open. Nobody has published the cheap decisive test: two matched batteries brought to the same state of charge, v(t) and i(t) logged on both batteries with a two-channel scope or logger sampled well above the pulse rate, integrated, compared with a plain-charger control, over 20 or more cycles. A plain-charger control is not a new idea: the manual's own "Further Testing" protocol already has one [read: manual pp.23-24]. Two batteries from the same batch, one charged by the SSG and one by a conventional charger, are cycled on the same load and their run-times compared. That is a fair comparison of chargers, and worth doing. It still cannot test over-unity, because it records only run-time on the charge battery, says outright that it is "not presently concerned with ... what amount of energy is consumed on the front end", and uses one battery per arm with no blinding. The missing piece is the same as everywhere else in this project: the energy drawn from the primary battery, integrated over time. Whether the community's repeated "COP above 1" reports are entirely SOC drift and meter error, or contain a pulse-charging effect on battery chemistry worth studying (for example, recovered capacity in sulfated batteries), is unsettled by data. Note that a pulse-charging benefit to battery capacity would be interesting chemistry; it would still not be energy from the vacuum.

8. Check yourself

1. A coil with L = 50 mH carries 1.2 A when the transistor turns off. What is the most energy one pulse can deliver to the charge battery, and where did it come from?
2. The BM3 manual's table computes Battery COP = Input Joules / Output Joules. If a builder records 200 kJ in and 250 kJ out, what does the table report, and what would it mean taken literally?
3. A 12 V, 7 Ah battery is charged with 30 kJ and later delivers 33 kJ. On the manual's 12.0-12.65 V scale, about how many millivolts of resting-voltage drift between the start and end of the cycle would explain the 3 kJ surplus? (Use 12.6 V average.)
4. Why does a heat pump's COP of 3 not violate energy conservation, and what is the analogous "outside" term in the SSG?
5. In the Sriphan paper, how can you tell that "running battery 12.71 W" is not the primary battery?

9. Further reading

Mainstream: - Griffiths, D. J., Introduction to Electrodynamics, chapter 7 (Faraday's law, inductance, energy in magnetic fields). - Horowitz, P. and Hill, W., The Art of Electronics, 3rd ed., sections on inductors, inductive kick and switching converters. - Linden, D. and Reddy, T. B. (eds.), Linden's Handbook of Batteries, lead-acid chapters (charge efficiency, open-circuit voltage vs state of charge). - Kittel, C., Introduction to Solid State Physics, ferromagnetism and domains. - Griffiths, D. J., Introduction to Quantum Mechanics, chapter 2 (harmonic oscillator and zero-point energy). - Schroeder, D. V., An Introduction to Thermal Physics, heat engines and refrigerators (COP done properly).

Researcher and community: - BM3 Group, Bedini Monopole 3 Group Experiment, V1.01 (2008), in the project folder [read]. - Bedini, J. C., US Patent 6,545,444 B2 (the manual prints it as 6,454,444 on p.1 and p.24, a digit transposition [read]), "Device and method for utilizing a monopole motor to create back EMF to charge batteries" (2003) [web: https://patents.google.com/patent/US6545444B2/en]. - Bedini, J., Bearden, T. and Bedini, J., Free Energy Generation: Circuits and Schematics, Cheniere Press, 2006 [web: cited as ref. 5 in Sriphan et al.]. Not in the folder. - Rick Friedrich, "Monopole Mechanical Oscillator Energizer" Self-runner diagram (May 2007), and Erwin Badertscher's redraws of the N-pole and SSG circuits (2005), in the project folder as images [read: page images]. - Sriphan, S. et al., J. Electrochem. Energy Convers. Storage 15(4), 041002 (2018) [web: http://gratisenergi.se/mgbedini.pdf]. - "Bedini Motor test 5 day motor run on 9VDC battery science fair project" (saved Energy21 web page), in the project folder [read].

---

Answers

1. E = ½ L I² = 0.5 x 0.050 x 1.2² = 0.036 J (36 mJ). It came from the primary battery, which drove the 1.2 A through the coil during the on-time.
2. The table reports 200/250 = 0.80. Read with the manual's own text definition of COP (out/in, p.10), 0.80 would mean the battery returned only 80% of its input, the opposite of what was recorded. The correct out/in value is 1.25. Because the printed formula is inverted, a table value can't be interpreted until you know which formula the builder actually used.
3. Full capacity ≈ 12.6 x 7 x 3600 = 317.5 kJ. 3 kJ / 317.5 kJ = 0.94% of SOC. Times 650 mV (12.65 - 12.0) ≈ 6 mV. A few millivolts of drift is enough, which is why matched start and end states must be measured carefully (or the cycle repeated many times so drift averages out).
4. The heat pump's boundary includes the outside air: 1 J of electricity moves 3 J of heat, 2 J of which came from the air, so total energy is conserved. In the SSG the "outside" term is the primary battery, which supplies every joule that reaches the charge battery, and which the BM3 protocol deliberately leaves out.
5. Averaging Table 3's discharge powers (V̄_out x I_out for the charged battery running its load) gives exactly 12.71 W, and averaging the charging powers gives exactly 4.94 W. Both Table 2 numbers are summaries of measurements on the charge battery; no primary-battery measurement appears.