Bedini-John: Science Contrarian review

Reviewer role: Science Contrarian (PhD-standard, evidence-first; Mallove is a standards model, not a persona). Inputs: facts.json, claims.md (Step 1), source folder re-read directly: the BM3 manual (full text via pdftotext, pp.4-5, 9-11, 19-21), the 5-day motor web page (full text), and the Self-runner1.JPG schematic (page image).

Source gap (repeated from Step 1). The folder has no Bedini book, no patent text, no lab notes and no measured data (no meter logs, scope traces or COP tables). The one document with a protocol (the BM3 manual) is volunteer-written and contains no results. Anything below about whether the effect exists rests on [web] sources I name, and on [memory] where marked. Verdicts are therefore mostly "never properly tested", not "refuted".

Tag key: [read] from the folder documents; [read: page image]; [web] URL; [memory] with confidence.

---

Corrections to Step 1

1. The patent does not claim over-unity. claims.md claim 1 ("charges by radiant energy rather than conventional current") is the BM3 group's and Bedini's spoken gloss, not the patent's. US 6,545,444 B2, "Device and method for utilizing a monopole motor to create back EMF to charge batteries", filed 13 Mar 2001, granted 8 Apr 2003, claims a motor that captures back EMF for charging a recovery battery; it frames the machine as an "open dissipative system" in which a small trigger pulse is paid for and recovered back-EMF is stored, with no stated COP above 1 [web: https://patents.google.com/patent/US6545444B2/en]. facts.json correctly fixed the printed patent number (6,454,444 is a typo in the manual cover [read]) but did not note this gap between patent and manual rhetoric. The manual's "Version" footer says 1.00 on every page I checked, while the file is named V1.01 [read].
2. Book citation in claims.md is garbled. It reads "Bedini's own 1984 book, 'Free Energy Generation' (Bedini and Bearden, 2006)". The book is Free Energy Generation: Circuits and Schematics: 20 Bedini-Bearden Years, Cheniere Press, 2006 [web: https://books.google.com/books/about/Free_Energy_Generation.html?id=NUuuHwAACAAJ]. "1984" belongs to the N-pole motor drawing (N-Pole motor 84.jpg, 9-19-1984 [read: page image per Step 1]), not to a book. Not verified: whether the book is the 2006 or 2007 printing (a Scribd copy is titled 2007).
3. The inverted COP is confirmed, with detail. The text defines COP as "total energy performed as work / total energy paid for by the operator", where 1.00 means output equals input (p.10), but the table's last bullet says "Battery COP is the result of Input Joules / Output Joules" (p.11) [read]. As written, a table value above 1 would mean less energy out than in. A published replication (see claim 2) uses output/input, so the intent is clearly the standard definition and the table text is a typo. Step 1 left this open; I resolve it as a typo that matters to anyone computing from the manual literally.
4. Step 1 attribution detail confirmed, one gap: facts.json says the test ignores the primary battery. Confirmed in the manual: "At this point we will not be concerned with the current being used by the input battery... just try and forget it" (p.20-21), and later "We are not presently concerned with ... even what amount of energy is consumed on the front end" (p.22) [read]. The manual also says "Only an insignificant amount of conventional charging could take place" (p.4) [read], which is an unmeasured assertion.
5. Page 5 of the manual explicitly ties the effect to a measurement method: "by measuring with conventional meters we will see more energy leaving the receiving battery via a constant load than entered it" [read p.5]. This matters for claim 2 and was not quoted in Step 1.
6. 5-day motor page, extra details Step 1 left out: the page itself says "10 year old" in one place and "12 year old" in another; the page's editor suspects the wheel is driven by a slot-car motor; and the page is second-hand ("John says this circuit ran for 5 days") with no end-of-run voltage and no named witness other than an unnamed science teacher who measured 8.9 V at start [read].
7. Unopened files: Step 1's "[memory] low" entries for SSG.jpg, SCH-illust. 2coil 4inch.jpg, B&C update, trevor3p/ (2 of 3) and six N-pole files remain unverified; I did not open them either. No verdict below depends on them.

---

Claim 1: The SSG charges a second battery by "radiant energy", not conventional current

a. Mathematics. The circuit, as drawn in Self-runner1.JPG [read: page image], is a transistor-switched coil (bifilar, ~850 turns) with diodes: when the transistor turns off, the coil's stored energy, E = (1/2) L I^2 (L = inductance, I = current just before turn-off), must go somewhere, and the voltage rises until a diode conducts. That is the textbook flyback mechanism, and the neon bulb at about 90 V and the "over 2000 V with S1 open" note are both consistent with it [read: page image]. Nothing in the folder gives an equation for any non-conventional charging channel. The manual's contrast "electron current vs. something else" has no mathematical statement, so there is nothing to check for consistency. Illustrative number (my own, not from the sources): with L = 10 mH and I = 0.3 A, E = 0.5 x 0.01 x 0.09 = 0.45 mJ per pulse.

b. Consistency with experiment. Flyback charging from collapsing coil fields is routine engineering (boost converters, ignition coils) [memory: high]. A battery charged by a rectified pulse is charged by current by definition; any current through a battery in the charging direction is electron/ion current. So the weak form (pulse charging through diodes) is consistent with all known experiments. The strong form (a charging channel that is not conventional current) conflicts with nothing only because it predicts nothing measurable on its own.

c. Tested? The group's own text says the effect cannot be seen as electron current ("what makes it work isn't electron current", p.4 [read]), but the only measurement offered is conventional current and voltage meters. So the strong form has never been tested by anyone because no measurable differs from the weak form. Not dismissed without testing; it was never given a testable definition.

d. If true. Energy would have to enter the battery by a path not captured by P = V x I at the battery terminals, which would violate the definition of electrical power at a terminal pair. A genuinely new energy channel would also need a conserved-quantity bookkeeping scheme; none is proposed. A hidden source, such as the environment, is the minimal interpretation, and the patent itself invokes an "environmental source" while claiming thermodynamic compliance [web: patent URL above].

Strongest honest case. The patent and Bedini's backers argue that a coil's back-EMF spike delivers energy in a form (very short, high voltage) that ordinary averaging meters misread, so conventional measurement understates what reaches the battery; and that "recovery" of otherwise wasted back-EMF is real, which is true.

Verdict (split). - Weak form (flyback pulses charge the second battery through current): Established. - Strong form (a charging mechanism distinct from conventional current): Not testable as stated.

---

Claim 2: Charge-battery COP above 1, improving with cycling

a. Mathematics. The manual's own formulas: Input J = average V x I x t; Output J = average V x I x t; COP = Input/Output in the table, work/paid in the text [read p.10-11]. Problems, each [read] in the manual unless noted: - Inverted COP (Correction 3). - Average voltage is the mean of start and end resting voltages (p.11), not the voltage during charging. Terminal voltage during charging is above resting voltage in a lead-acid cell, so input energy is systematically underestimated. Size of bias: charging at roughly 13.5 V instead of 12.6 V is ~7% (my arithmetic; the voltage values are typical [memory: medium]). That is not large enough alone to produce COP 1.6, but it points the right way. - Pulsed current read on an analog meter, or by matching lamp brightness (p.20-21) [read]. A needle meter on a spiky unidirectional waveform reads some weighted average depending on damping; the manual itself says "a little trickier than normal to measure". Meter error in the direction of understating input is plausible and unquantified. - Primary battery excluded by design ("forget it"). - State of charge not tracked. The test starts and ends on resting voltage "at least one hour" after, which gives only a rough state-of-charge reading.

Charge-balance check (published data). The one paper I found reporting COP > 1 for an SSG-type machine uses the manual's method. Sriphan, Kerdchang, Prommas et al., "Coefficient of Performance of Battery Running and Charging by Magnet Generator Bedini", J. Electrochem. Energy Convers. Storage 15(4), 041002, 2018 (DOI 10.1115/1.4039504) [web: http://gratisenergi.se/mgbedini.pdf, copy of the ResearchGate paper]. Table 3 there lists, for the "original" design, runs such as 0.485 A x 39,600 s in, and 1.01 A x 25,200 s out. Recomputing: - Charge in: 0.485 A x 39,600 s = 19,206 A·s = 5.34 Ah. - Charge out: 1.01 A x 25,200 s = 25,452 A·s = 7.07 Ah. - Energy in: 12.58 V x 19,206 A·s = 241.6 kJ (paper: 241,560 J, matches). Energy out: 12.62 V x 25,452 A·s = 321.2 kJ (paper: 321,300 J, matches). COP = 1.33, matching the table. So the output charge exceeds input charge by 33%. A lead-acid cell can't return more ampere-hours than it received unless it began with stored charge or the current was mis-measured; coulombic efficiency for lead-acid is normally under 100% [memory: high]. The paper does not state the battery's starting state of charge in the table I read, so I can't tell which explanation applies. That check needs no assumption about voltage.

The same paper's Table 2 lists "Running battery average power consume: 12.71 W" (original) and 23.68 W (replication) [read, web]. If "running battery" is the primary battery, 12.71 W over 39,600 s is about 503 kJ, which exceeds the 321 kJ delivered by the charge battery, i.e. total-system COP below 1. I am not certain of the table labelling, so this is a flag for follow-up, not a finding.

b. Consistency with experiment. Lead-acid round-trip energy efficiency is well below 1 (roughly 70-85% [memory: medium; not web-checked]). So a measured COP above 1 on the charge battery alone requires either under-measured input, unaccounted starting charge, or an outside source. The manual's own text gives a mundane contributor: new batteries need "5 to 10 charge/discharge cycles" to condition and "Older batteries go through a de-sulfating process", so "repeated cycles should improve the COP" (p.12, p.22) [read]. Rising capacity with cycling in a new battery is expected from conditioning and does not distinguish the SSG from any charger. The manual asks for a conventional-charger control only as optional "Further Testing" (p.22) [read], and says it is "not presently concerned" with energy used on the front end.

c. Tested? Reports: BM3 group members (data not in the folder; the posts are missing since Yahoo Groups was shut down [memory: high]); Sriphan et al. 2018 (peer-reviewed journal, 4 runs per design, same charge-battery-only method). No independent replication of the paper that I found. I do not have a source for any neutral laboratory test of the SSG; a web summary I retrieved mentions a 2004 television test but I could not confirm it, so I do not use it. Not dismissed without testing in the sense of a refusal to look: skeptic forum threads exist, but I did not find a controlled measurement of total system power in or out [web search, no usable source].

d. If true. The charge battery's COP above 1 together with the primary's measured draw (counted) would be a net energy gain for the pair: breaks conservation of energy unless an external source is identified. With the primary excluded, nothing in mainstream physics is challenged at all, because the primary battery's discharge can supply the energy.

Strongest honest case. The method is the manual's; the paper's numbers are arithmetically consistent and repeated across 4 runs and a second circuit. It is exactly the sort of anomaly that deserves a proper instrument, and the folk method was designed by non-professionals who flag measurement difficulty openly. The right test is cheap and decisive (see below).

Critique of the researcher's side. Excluding the primary battery from the COP is not a minor omission: it drops the largest term. Calling a charge-battery-only COP above 1 "undeniable proof" (p.21) [read] is a logical error. The inverted formula went uncorrected through two versions.

Decisive test (not done, to my knowledge). Log voltage and current on both batteries at kHz sampling for the whole run, integrate v x i on each, include the primary, and check ampere-hour balance; run a conventional-charger control with matched time and batteries.

Verdict. - Charge-battery-only COP above 1 as reported: Never properly tested (reported by advocates with an input-excluding, averaging-meter protocol; not dismissed without testing, but no neutral controlled test on record). The cycling improvement is Plausible, untested as a conditioning effect that any charger would show. - Whole-system COP above 1 (primary included): Never properly tested.

---

Claim 3: Primary battery self-recharges ("backpopping"); the 5-day 9 V motor

a. Mathematics. From the page [read]: 22 mA at 4000 rpm, 120 h, 8.9 V at start, "400 millamps battery for one hour". Charge drawn = 0.022 A x 120 h = 2.64 Ah. Energy ≈ 8.9 V x 2.64 Ah x 3600 = ~84.6 kJ. A 9 V alkaline holds roughly 0.5 Ah [memory: high; the page itself says 400 mAh], i.e. about 16 kJ. So the stated figures need ~5 to 6 times the battery's capacity. With the generator attached, 13.2 mA over 120 h = 1.58 Ah, still ~3 to 4 times. Both numbers cannot be true with an ordinary battery and no extra source, so either a number is mis-stated (for example an averaging meter on a pulsed current), the run was shorter, or the claim is anomalous. The self-runner schematic notes: "primary battery is recharging between powering everything" and "backpopping of primary takes place once every rotation" [read: page image], with no numbers.

b. Consistency with experiment. A battery that recharges itself while powering a load and its generator is a closed loop and would be a perpetual-motion machine of the second kind (or first). All tested closed loops are known to lose energy [memory: high]. The note "capacitor across the DC of bridge reduces effect" (Self-runner1.JPG) is itself a test of the claim (a capacitor shouldn't change a net energy balance except by changing the waveform), but no data accompany it.

c. Tested? No. A teacher measured start voltage; there is no end-of-run voltage, no witness report, no instrument record. The page's own editor disbelieved the slot-car-motor possibility ("it appears ... driven by a slot car type motor"). The "12-year motor" is an unrecorded claim. The Rick Friedrich page says "built, tested, filmed and reported" and forbids posting without permission [read: page image]: no data, and sharing restricted. Not dismissed through a bad test: no test exists.

d. If true. The first law would fail for an isolated battery-motor system, or there would be a hidden input (e.g. an environmental source) whose mechanism needs naming. Either way, a published, instrumented 5-day run would be the first-class evidence; none exists in the folder.

Strongest honest case. The 22 mA figure could be right if the motor's real consumption is far smaller than a slot car's, and the page's "400 mAh" ignores that low-drain alkaline cells deliver more capacity per mAh at small currents; but even generous capacity (about 0.55 Ah at 25 mA [memory: medium]) gives ~25 h, not 120 h. I do not find a mundane explanation for 120 h.

Critique of the researcher's side. Second-hand, children's science-fair framing; a stated "12 years on the same batteries" that cannot be checked; the instruction to build exactly as Bedini says is the escape clause (claim 5).

Verdict. Never properly tested (anecdote; neither run nor refuted under controlled conditions; not dismissed without testing, simply never documented). The stated figures conflict with the page's own battery capacity, so as reported the numbers do not hold together. I do not call it refuted: the numbers are not a controlled measurement.

---

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

a. Mathematics. None offered; it is a one-line quotation from Bedini (p.4) [read]. A Bloch wall is the thin transition region between two magnetic domains in a ferromagnet, across which the magnetisation rotates; it is a boundary, not "the place in the center of the magnet where the energy comes from" [memory: high]. Moving domain walls with a pulsed coil is known (Barkhausen noise, hysteresis loss), and dissipates energy as heat. "Zero-point energy" is the ground-state energy of a quantum system; by definition no lower state is available, so net work cannot be drawn from it without a lower state to fall to [memory: high; standard QM]. The sentence thus uses two technical terms in ways their definitions don't support. It is not a calculation, so I cannot call it refuted by calculation.

b. Consistency with experiment. Nothing measured in the folder supports or opposes a Bloch-wall mechanism. Ceramic (strontium ferrite) magnets are hard magnets; a few-hundred mA pulse in a spool with a welding-rod core (R60) is far below what would be needed to change their magnetisation substantially [memory: medium].

c. Tested? No one has tested it. There's no measurable prediction (for example, magnet remanence or domain structure before and after a run).

d. If true. Extractable vacuum energy would need quantum field theory to allow a lower vacuum state accessible to a bar magnet, with no known mechanism [memory: medium].

Verdict. Not testable as stated. Where it can be read as a physical statement, it is contradicted by the standard meaning of the terms, but I do not rest the verdict on that.

---

Claim 5: Works only if built exactly as specified (ceramic magnets, exact parts)

a. Mathematics. Not applicable.

b. Consistency. The manual's rule: "Neodymium magnets are NOT suitable for the Monopole SSG!" (p.15) [read]. The 2018 Sriphan et al. paper used a six-pole, eight-neodymium-magnet prototype and reports COP above 1 [web]. Either the exact-build rule is not real, or the two reports mean different things by "COP".

c. Tested? There is no controlled variation (change one part, measure the effect on both batteries). The manual instead restricts changes until a builder "has done as directed" and says that only those who "follow directions" "will be considered for invitation to other advanced groups" (p.5) [read]. That is a social filter on who gets to report, and a failed replication can always be blamed on a deviation.

d. If true. Little: this is a claim about the researcher's reporting standard, not about physics.

Strongest honest case. Real circuits are tuned (the manual's "sweet spot", 1-ohm resistor test, [read p.18]) and beginners really do fail to tune pulse circuits; insisting on a reference build before varying anything is sound practice in replication.

Critique. A claim whose failures are always attributed to the builder is unfalsifiable unless the build is fully specified, with a pre-stated pass criterion. The manual states a pass criterion (COP above 1 in 20 cycles) but a weak one, as above.

Verdict. Not testable as stated in the "exact build" form: failures are explained away. The narrower, testable form (an SSG built to the manual's parts list yields charge-battery COP above 1 under a full-system measurement) is under claim 2: Never properly tested.

---

Claim 6: Collapse spikes harvested through diodes/capacitor into a second battery (energy recovery)

a. Mathematics. Consistent: E = (1/2) L I^2 per pulse; a bridge or diode and a capacitor store it; the 1984 N-pole motor drawing records 25 to 300 mA, "built as a test motor for energy recovery" [read: page image per Step 1], and the dual-battery drawing shows spikes to about 200 V into a 330 uF capacitor [read: page image per Step 1]. A 330 uF capacitor at 200 V stores (1/2) C V^2 = 0.5 x 330e-6 x 200^2 = 6.6 J (my arithmetic from the drawn values).

b. Consistent with experiment. Yes: inductive flyback recovery is standard (boost and flyback converters, regenerative drives). Recovered energy is a fraction of what the coil received, not more than the source supplies [memory: high].

c. Tested? Standard engineering is tested daily; the specific Bedini machines have no measured efficiency in the folder. Interest in claim 6 rests on whether recovery efficiency can exceed what conventional circuits achieve, which no data address.

d. If true. Nothing new in the weak form. A recovery efficiency above 100% would be claim 3.

Verdict. Established as energy recovery in principle (recovering part of the pulse energy). Any specific efficiency for the Bedini circuits: Never properly tested.

---

Summary table

Overall contrarian assessment

The strongest anomaly in this folder is a measurement claim, not a physical one: the charge battery appears to return more energy and more charge than was measured going in (published example: 7.07 Ah out against 5.34 Ah in), under a protocol that leaves out the primary battery and uses averaged meter readings. That is exactly the kind of anomaly worth a proper instrumented test, and it is cheap to run. It is not evidence for zero-point energy. The "exact build" rule and the advanced-group gate weaken the movement's own claim to be reproducible, and the 5-day figure does not survive its own battery arithmetic.