Bedini-John: Mathematician review (Step 2)

Reviewer role: PhD mathematician. Checks derivations, units, definitions, hidden assumptions. Reader level: self-taught undergraduate.

Source gap (repeated from claims.md): The folder holds no Bedini book, patent text, lab notes or measured data. The only substantial readable works are the 2008 BM3 Group build manual (volunteer-written) and a saved web page about the "science fair" motor. No SSG test results (meter logs, scope traces, COP tables) exist in the folder. Therefore every verdict below is about the arguments and accounting as written, not about measured data. Where I say "no test on record" that means none in the folder and none I could find; I did not find any independent replication (I did not do a literature search beyond the checks cited).

What I read myself: the full text layer of the BM3 manual (pp. 4-5, 10-11, 12, 14, 18-22 in detail) [read]; the full saved HTML of the 5-day motor page [read]. I did not re-open the image files; image-derived statements are Step 1's [read: page image] and are used only as stated there.

---

Corrections to Step 1

1. Test-protocol details omitted in facts.json that matter. The manual (p.21) has the builder place the analog ammeter in series with the charging battery, read amps, then "Remove your analog amp meter before the start of the charge cycle" [read]. p.10 also says the meter should not stay connected "as it reduces charge rate" [read]. So the logged "Input (Amps)" is a spot reading taken with a meter that perturbs the circuit, and the actual charge runs without it. facts.json does not mention this.
2. Overstated criticism of the analog meter. facts.json says pulsed current "measured with an analog meter ... is an unreliable measure of true power." Nuance: a moving-coil meter reads the time-average current, which for a battery being charged is the quantity that matters for charge (amp-hours). The real defect is not the meter type; it is (i) the meter being removed during the real charge, (ii) power being computed as (rest voltage) x (average current), and (iii) the primary battery being excluded (see Claim 2).
3. Inverted COP formula: confirmed, and more serious than stated. Verified at manual p.10 (definition: "total energy performed as work / total energy paid for") versus p.11 (table: "Battery COP is the result of Input Joules / Output Joules") [read]. Step 1 only notes this. I show below (Claim 2) that with the table's formula an ordinary lead-acid battery gives COP above 1. I cannot see the linked online calculator, so I do not know whether it implements the table text or the definition.
4. Patent description. facts.json cites US 6,545,444 only for its number. [web] https://patents.google.com/patent/US6545444B2/en : title "Device and method for utilizing a monopole motor to create back EMF to charge batteries", inventor John C. Bedini; rectifies back-EMF through a bridge into a capacitor then a recovery battery; the page summary I retrieved reports no experimental data in the patent and an expiry for non-payment of fees (that last detail I did not independently confirm). The patent text is a flyback-recovery circuit, which is mainstream electronics; the "radiant energy" language belongs to the manual and the community, not to a measurement.
5. claims.md claim 3 merges two different devices. The "backpopping" claim comes from the Rick Friedrich annotated schematic (a 2-transistor SSG with a bridge and top coil); the 5-day, 22 mA figure comes from a different circuit (single-transistor all-North-pole 9 V motor with LED generator). They share no data and should be treated as two claims. I treat them as 3a (backpopping) and 3b (5-day run).
6. Minor: facts.json says the 5-day motor "drew less (13.2 mA) when the generator load was attached" - confirmed [read]. The page states the battery "measured 8.9 volts" before the start; no end-of-run voltage [read].

---

Background the reader needs (brief)

• Inductor energy. A coil of inductance L carrying current I stores E = (1/2) L I^2 joules. When a transistor cuts the current, the coil forces current to continue (v = L dI/dt), producing a voltage spike ("flyback" or "back EMF", EMF = electromotive force). This energy came from whatever drove I in the first place. Textbook: Griffiths, Introduction to Electrodynamics, the chapter on inductance and magnetic energy.
• Power in a battery. Instantaneous power is p(t) = v(t) i(t). Average power is the mean of the product, which is not in general the product of means.
• Energy efficiency vs charge (coulombic) efficiency. Energy = V x Q. A lead-acid battery returns about 85% of charge (Ah) and about 70-80% of energy (Wh) [web] https://www.sciencedirect.com/topics/engineering/ah-efficiency , https://en.wikipedia.org/wiki/Roundtrip_efficiency (search summaries; ranges vary by source).

---

Claim-by-claim

Claim 1: SSG charges the second battery by "radiant energy" rather than conventional electron current from the primary battery.

a. Mathematically consistent? There is no mathematics to check: "radiant energy" is never defined by an equation, unit or measurable quantity. The manual divides energy into "Classical Electromagnetic" and "Radiant Energy / Energy from the Vacuum" [read p.21] but gives no way to separate them. The argument offered is by elimination: the 1-ohm test "helps us see that there are two kinds of energy ... The 'conventional' kind of energy is not enough to account for the observed charging. Therefore there must be something else ... that is hard to measure" [read p.20]. That is an argument from the failure of a crude measurement to a new entity; the "therefore" does not follow, because the measurement chain (see Claim 2) is unsound. Also p.4: "Only an insignificant amount of conventional charging could take place" [read] is asserted with no number. By contrast, the standard circuit analysis is clean: transistor switches coil current from the primary battery; on turn-off the coil EMF (v = L dI/dt) drives current through the diode into the second battery. Energy stored (1/2) L I^2 each pulse is the upper bound on what that pulse can deliver.

Worked bound (illustrative values, NOT from the document, which gives no inductance or peak current): L = 20 mH, peak current 2 A: E = 0.5 x 0.020 x 4 = 0.04 J per pulse. At 100 pulses per second that is at most 4 W, and every joule was drawn from the primary battery's 12 V supply during the on-time. No extra source is needed to explain charging.

b. Consistent with experiments? The weak form (a pulsed inductive circuit can charge a battery with the spikes) is ordinary and consistent with all known circuit behaviour. The strong form (the energy comes from somewhere other than the primary battery) has no measurement against it or for it in the folder.

c. Tested? The manual's own protocol deliberately does not account for the primary battery ("as hard as it is, just try and forget it" [read p.21]), so it cannot test the strong form. No neutral test on record. I found no independent replication in this review (not searched exhaustively).

d. What would change? Conservation of energy in a closed circuit (or an identified external source: if there were one, a measurable one). Quantum field theory's zero-point energy is not extractable as work from a ground state; I flag that last sentence as [memory] high confidence, standard result.

Verdict (split): - Weak form (inductive flyback charging from the primary battery): Established. - Strong form (non-conventional "radiant" energy): Not testable as stated (undefined quantity) and Never properly tested (the protocol omits the primary battery; no neutral test on record; not shown to have been dismissed without testing).

---

Claim 2: Measured on the charge battery alone, energy out exceeds energy in ("battery COP" above 1); repeated cycles improve it.

a. Mathematically consistent? No, in four respects:

1. Inverted definition. p.10: COP = work out / energy paid in, "1.00 means energy output equals energy input" [read]. p.11: Battery COP = Input Joules / Output Joules [read]. If the table is used as written, COP above 1 means more energy went in than came out, the ordinary, lossy case. Conversely the text on p.5 and p.22 treats COP above 1 as the desired anomaly. The two cannot both be right.

Worked consequence: for a normal lead-acid battery with 85% charge efficiency and the same average voltage used on both sides (the table averages resting start/end voltages in both sections, so the voltage factors nearly cancel), Output J / Input J ~ 0.85. The table's Input/Output = 1/0.85 = 1.18, "above 1" with no anomaly at all. Whether anyone's reported ">1" figures came from this is unknown; no data is in the folder.

2. Resting voltage used for power. Input Joules = (average of start and end resting voltages) x (current) x time [read p.11]. True energy is the integral of v(t) i(t) with v the terminal voltage during charge, which exceeds the resting voltage. With the same approach on discharge (terminal voltage is below resting voltage), the voltage factors largely cancel, so the table's quantity is essentially ampere-hours out/in, not energy. Ah efficiency is the wrong measure for an energy claim. Pulsing adds a small further difference: for v = E + r i, mean(v i) = E mean(i) + r mean(i^2). Example: E = 12.3 V, r = 0.1 ohm, 2 A spikes at 5% duty: mean i = 0.1 A, mean i^2 = 0.2 A^2, so E mean(i) = 1.23 W and r mean(i^2) = 0.02 W, about 1.6%. Small; I do not claim this is the main error.
3. Unmatched battery state. The discharge cutoff is a voltage threshold (12.0-12.2 V, p.12 [read]); the charge stops at a voltage or a time. Nothing forces the battery to return to the same state of charge, so energy "out" can include energy stored in earlier cycles. A COP computed over one cycle is not a closed energy balance unless the state of charge at the start and the end is the same. The manual itself says new batteries are sulfated and need "5 to 10 charge/discharge cycles" of conditioning, and that repeated cycling "should improve the COP" [read p.12, p.22]. A rising capacity of a new lead-acid battery with conditioning is a known battery effect (reader should check: [memory] medium; not verified on the web) and does not require an external source.
4. Measurement perturbation. The current is read with a series ammeter that "reduces charge rate" and is removed before the real charge [read p.10, p.21]. Input current is therefore measured under different conditions than it was delivered. The sign of the error is understated input, which inflates a true out/in ratio.

Boundary error (the fundamental one): the energy balance is drawn around the charging battery only, with the primary battery's draw excluded. Conservation of energy applies to the whole system. The p.5 sentence "Instead of looking at the battery that powers the system, we look at the charging battery" [read] changes the system boundary, so the result cannot bear on over-unity even if the numbers were exact. Illustration (assumed values): primary draws 0.5 A at 12 V = 6 W; charge battery receives 0.1 A at 13 V = 1.3 W. Charge battery in/out may look "COP 1.0 or above" while the system efficiency is at most about 0.2 x (battery return fraction).

b. Consistent with experiments? Lead-acid energy efficiency of 70-80% is well documented [web, see above]. A true out/in above 1 would contradict that; a table-defined ratio above 1 does not.

c. Tested? The manual contains no data [read]. No independent replication with total-system accounting is in the folder, and none that I found. The manual asks for 20 cycles posted to a Yahoo group; the group's archives are not in the folder (Yahoo Groups content was removed, [memory] high).

d. What would change? If a closed total-energy balance (primary + charging battery, with matched start/end state of charge, integrated v(t) i(t) from a scope or logger) showed out above in, conservation of energy would fail. Nothing in the documents supports that.

Verdict: As a statement of over-unity, Internally inconsistent (the COP formula is inverted between definition and table, and the system boundary excludes the primary battery). As an experimental claim about real energy gain, Never properly tested (no neutral, controlled, total-system measurement on record; not shown to have been dismissed without testing). The simple "conditioning improves cycle capacity" part is Plausible, untested by this folder (ordinary battery conditioning).

---

Claim 3a: Backpopping, the primary battery is recharged while powering the machine (Friedrich schematic).

a. Consistent? The schematic notes ("primary battery recharges while it powers everything", "capacitor across the DC side significantly reduces effect") are assertions with no data [read per Step 1: page image]. A battery can show a rising terminal voltage when its load current falls or when charging current from a spike path flows into it; terminal voltage is not state of charge. A recovery path (coil spike back to the supply through a bridge) returns a fraction of the energy drawn, which would reduce net draw (and is mainstream: regenerative switching), but cannot return more than was drawn. b. Consistent with experiments in the weak form (partial recovery, energy bookkeeping saying net draw is positive). No data for the strong form. c. No test on record in the folder; no independent replication found. d. The strong form (net-negative draw) would violate energy conservation. Verdict: Weak form (partial recovery): Established as circuit behaviour [memory] high. Strong form (net recharge, i.e. self-running): Never properly tested.

Claim 3b: The 5-day, 22 mA, 9 V motor run.

a. Arithmetic (done by me): 22 mA x 120 h = 2.64 Ah. At 8.9 V, energy = 8.9 x 2.64 = 23.5 Wh (about 84.6 kJ) [read inputs; arithmetic mine]. The page itself rates such a battery at about 400 mAh [read]. Energizer datasheet listings give about 625 mAh at 25 mA drain for a 9 V alkaline (Energizer 522) [web] https://www.newark.com/energizer/522bp/alkaline-zn-mno2-battery-9v/dp/50F120 , search summary of the Energizer datasheet at https://data.energizer.com/pdfs/la522_ap.pdf (I did not open the PDF itself). So the claim requires 2.64/0.625 = 4.2 to 2.64/0.4 = 6.6 times the battery's capacity. At the 13.2 mA with load: 1.58 Ah, still 2.5 to 4 times. Internal oddity: the draw fell from 22 to 13.2 mA when a generator load was added [read]. Adding a load and seeing less input current would, if the 22 and 13.2 mA are both real mean currents from the same battery, mean input power fell while output rose. That could arise from an instrument artifact or changed pulse timing (an uncontrolled mundane explanation I cannot test) but is not documented. b. Inconsistent with standard battery capacity, if both the duration and the current are true. No end-of-run voltage, no data log, no independent witness report [read]; the account is second-hand ("John says"). Also a possible alternative: the page editor notes a slot-car motor appears to drive the wheel, which would draw 200-500 mA by his estimate [read]; the page does not reconcile this. c. No test on record; a classroom exhibition is not a test. Never replicated by a documented controlled run. d. Conservation of energy would be violated if the figures were accurate and complete. Verdict: Never properly tested (anecdote; unauditable). Taken at face value, the numbers conflict with the battery's rated capacity by a factor of about 4-7, which makes a measurement or recording error far more likely than new physics, but I cannot assert a calculation-refutation because the underlying measurements are not in evidence.

---

Claim 4: The coil pulse "opens the Bloch wall" of the magnet and pumps in zero-point energy.

a. Consistent? The quote is Bedini's, relayed in the manual [read p.4]. In mainstream physics a Bloch wall is the thin transition layer between two magnetic domains in a ferromagnet where the magnetization rotates (domain wall) [memory] high confidence. It is not "the place in the center of the magnet where the energy comes from"; its energy is a positive wall energy per unit area that costs energy to create. "Opening and closing the window pumps in zero-point energy" has no equations, no coupling constant, no units. Zero-point energy is the lowest energy of a quantum system; extracting net work from it would need a lower state to fall into, and there is none [memory] high. b. No experimental results in the folder support it. Ceramic (strontium ferrite) magnets, a hard magnet with high coercivity, are specified, whereas neodymium magnets are said to be unsuitable [read p.14]; a coil pulse of a few amp-turns is unlikely to move domain walls in a hard ceramic magnet at all (my reasoning, [memory] low-medium; not checked on the web). c. No test. d. Would require a mechanism for tapping vacuum energy; none given. Verdict: Not testable as stated; terminology is misapplied (Bloch wall), no measurable prediction is made.

---

Claim 5: Only exact builds (ceramic magnets, specified coil, no changes) reproduce the effect.

a. Not a mathematical claim. Logically it immunizes the claim: any failure can be attributed to a deviation from the build. The manual states a policy "build exactly as directed before changing anything" and invites only conforming contributors to "advanced groups" [read p.5]. It is a social rule, not a test condition. b/c. The manual does give a concrete protocol and a clear parts list, which is a step toward testability. But nothing defines what counts as "the effect" numerically (the COP is defective, Claim 2) and no pass/fail criterion exists. d. None. Verdict: Not testable as stated (no measurable success criterion; failure explained away by construction). It can become testable if a recording rule is fixed (see "What would settle it").

---

Claim 6: Collapse spikes can be harvested through a bridge and capacitor into a second battery (1984 N-pole lineage).

a. Consistent? Yes. Pulsed coil, diode or bridge, capacitor, second battery is standard flyback energy recovery (as in the patent, [web] https://patents.google.com/patent/US6545444B2/en ). The 1984 drawing is described as a "test motor for energy recovery", 25-300 mA [per Step 1, page image]. Efficiency less than one is implied. b. Consistent with normal circuit theory (energy per pulse at most (1/2) L I^2). c. Not a contested result; nothing here needs replication, only an efficiency number, which the folder lacks. d. Nothing. Verdict: Established as energy recovery. Any reading as net gain is covered by Claims 2 and 3.

---

What one measurement would settle it

Use two identical batteries; log v(t) and i(t) of the primary battery and of the charging battery with a two-channel scope or data logger sampling well above the pulse rate; integrate p = v i on both over a full cycle. Start and end both batteries at the same measured state of charge (rest voltage after the same rest time, or an Ah-count from a calibrated discharge such as with the West Mountain CBA II in the folder). Compare total energy out of the system (stored energy gained by the charging battery plus whatever is recovered) with energy drawn from the primary battery. Do the same with a plain constant-current charger to calibrate lead-acid return. If total out exceeds in beyond the instrument error, the claim stands; otherwise the "extra" energy is accounting.

Summary table

Web checks used (budget about 6): US6545444B2 (patent text), Energizer 9 V datasheet capacity, lead-acid efficiency. Not web-verified: Bloch wall definition, lead-acid conditioning effect, Yahoo Groups removal ([memory], confidence stated above).