Physicist review (PhD physicist): Ainslie-Rosemary-Ann

Source gap (repeated from claims.md). The patent body (WO 99/38247) was not read from the scan, only via Patrick Kelly's digest. The raw oscilloscope data, the ABB / Cape Hope Metrology Laboratory / BP reports and any third-party replication are not in the folder. Conclusions about those items rest on the authors' own reports or on [memory]. Everything about the experiment below comes from the Quantum article (October 2002) and its 2009 reprint, which I read myself: the reprint by pdftotext, the Quantum pages 4-5 and the Pretoria News page as images [read: page image]. The Quantum pages 2-3 and the patent I did not re-read, so the 143-200 kHz and 3.7% / 1.3% duty statements are taken from Step 1.

Source tags: [read], [web] (URL), [memory] (confidence). Web check budget: two calls. WebSearch returned only a pointer to forum analysis (below) and WebFetch of the energeticforum thread returned HTTP 404, so the check of later replications FAILED. Nothing in this review depends on a replication having happened or not.

Corrections to Step 1

1. "1.22 MJ" is not the average-power energy. The Quantum Table 1 / 2009 Table 2 "Joules" column equals the instantaneous "Control Watts" at each row multiplied by total elapsed time (computed: 30,333 J / (29 min x 60) = 17.43 W; 956,978 J / 57,600 s = 16.61 W; 1,222,429 J / 59,820 s = 20.44 W) [read: page image Quantum p.7, and reprint Table 2; arithmetic mine]. That is not an integral of power over time. The stated average of 17.53 W over 997 min gives 1.049 MJ, not 1.22 MJ. So the ratio is 17.53 / 1.13 = 15.5 (or 1.049 MJ / 67.6 kJ), not the 18.1 implied by 1.22 MJ / 67.6 kJ. Step 1 reported "1.22 MJ" and "16 to 18" without flagging this. The headline "17 times" in the press rests on the larger number.
2. The load-power rise in the last rows is unexplained. Load temperature rise went from 48.7 C (960 min) to 54.3 C (983) to 59.9 C (997), i.e. calculated "watts" from 16.6 to 20.4, in the final 37 minutes, while battery volts barely changed (24.7, 25.1, 24.3) [read]. A steady-state device does not do this unless something in the circuit changed. The authors do not comment. The 1.22 MJ figure is driven by that last row.
3. claims.md item 1: "no measurable battery voltage fall" is too strong. Table: 24.8 V falling to 24.3-24.7 V, with a 25.1 V reading mid-run [read]. The article's own wording is "no evidence of any significant battery voltage reduction". Also the authors say they omitted evaluating the batteries in terms of rated capacity (see c below).
4. claims.md item 6: "a number of companies tested or supported it" overstates what the sources say. Quantum p.8: Cape Hope Metrology Laboratory confirmed that test equipment had been calibrated; Spescom (Fluke distributor) "attended demonstrations"; ABB's "independent tests were confined to the evaluation of instantaneous power delivered simultaneously by the battery and dissipated in the load" using four-channel oscilloscopes (so electrical, not calorimetric, with no result stated); BP applied tests "to measure the effect of the energy gains on battery duration" (no result stated); Sasol offers a bursary [read: page image Quantum p.8]. The Pretoria News adds that Fluke "provided sophisticated testing equipment and certified the results", a journalist's phrase with no document behind it [read: page image]. None of this is an endorsement of the 17x claim. The companies approved use of their names and recommended "a wider forum for systematic and statistical consideration".
5. Step 1 summary of the controlled-battery arithmetic is right (12.45 V x 0.039 A = 0.49 W) but the consequence should be stated: the stated 1.13 W excludes the 0.49 W gate-drive supply, so total input measured by the authors' own numbers is about 1.62 W and the ratio drops to about 11 even if all else is accepted [arithmetic mine].
6. Step 1 is otherwise accurate on what I could check (circuit, 13.32 V / 17.74 W calibration, 52 C, 997 min, 67.6 kJ).

Background calculation used below (all numbers mine; component values from [read])

• Load R = 10 ohm, shunt 0.25 ohm, supply 24 V, L = 8.64 microhenry. Time constant L/R = 0.86 microseconds, so the current reaches its steady value within a few microseconds of switch-on.
• Maximum current when on: I = 24 / 10.25 = 2.34 A. Power drawn while on: 24 x 2.34 = 56.2 W.
• At the stated 3.7% duty: average battery power = 0.037 x 56.2 = 2.08 W. At the 1.3% duty the article gives for the oscillating mode: 0.73 W. The reported 1.13 W corresponds to about 2.0% duty. So the reported battery power is plausible for a mostly-off switch.
• To dissipate 17.74 W in the 10 ohm load you need rms current I_rms = sqrt(17.74/10) = 1.33 A. With peak current limited to 2.34 A, that needs an effective duty D = (1.33/2.34)^2 = 0.32 (17.74 / 56.2 = 0.316). The battery then supplies about 0.32 x 56.2 = 17.7 W.
• Energy stored in the inductor at peak: E = (1/2) L I^2 = 0.5 x 8.64e-6 x 2.34^2 = 23.7 microjoules. Even if all of it were dumped once per cycle: at 2.4 kHz that is 0.057 W; at 200 kHz that is 4.7 W. This is energy the battery put into the inductor in the first place (already counted in the battery's power), so it cannot add to the total.
• Battery drain if load heat really were battery-fed: 17.53 W / 24 V = 0.73 A for 16.6 h = 12.1 Ah, about 60% of two 20 Ah batteries. At the reported 1.13 W: 0.047 A, 0.78 Ah, about 4%.

Claim 1. The circuit dissipated about 17.5 W in the load while the battery supplied about 1.13 W (gain 16-18x), over 16.5 h with no measurable battery voltage fall.

a. Mathematics internally consistent? Partly. The load side is simple (V^2/R = 13.32^2 / 10 = 17.74 W, correct) and 1.13 W x 997 min x 60 = 67.6 kJ is correct. But (i) the 1.22 MJ is a computation artefact (Correction 1); (ii) the scale of battery power depends on a sample window. Quantum says the sample range spanned 1.2 microseconds [read: page image p.7], while the oscillation in the quoted mode is 143-200 kHz (period 5-7 microseconds, per Step 1). A window shorter than one period cannot give a cycle average. The reprint says the time base was 1 microsecond/div (10 microseconds across), still only one or two periods. The method averages V x I with V fixed at 24 V and I = Vshunt / 0.25, with samples of negative shunt voltage counted as energy returned to the battery. With a ringing waveform at hundreds of kHz through a resistor built from four parallel carbon resistors and unspecified wiring, that sign convention can be corrupted by parasitic inductance and by the scope ground reference (the scope earth, the shunt and the gate-drive battery share a common node, a typical ground-loop situation) [reasoning, memory: high that such artefacts are standard in switching-power measurement]. b. Consistent with experiment? No, if taken at face value: it requires the load to dissipate 10 times the power the sources supply. It is consistent with the whole body of calorimetry and circuit measurement that every real switching converter obeys energy balance. The energy budget above is decisive (see verdict). c. Tested? By whom? Replicated? The only measurements are by the authors (Ainslie and Buckley / Martin) [read]. The article invited readers to replicate and says ABB, BP and others did tests, but no report is in the folder. The raw waveform data (spreadsheet) are not provided. The authors explicitly did not measure load electrical power, because of "protracted discussion" about phase lag [read], which means the central comparison mixes a thermal measurement of the load with a scope-sampled measurement of the source, with no electrical measurement on the load to cross-check. They also omitted a battery capacity/discharge test as "too many vagaries" [read: page image p.7], although a decisive test of how much charge left the batteries is a simple ampere-hour measurement. A DC energy meter or a battery discharge to cutoff would have settled it. Later replication: [memory, low] there were hobbyist attempts on internet forums, with analysis attributing the results to measurement artefacts; I tried to confirm this on the web and could not (WebFetch of https://energeticforum.com/archive/index.php/t-4314.html returned 404; a WebSearch only returned a pointer to forum analyses, e.g. https://www.youtube.com/playlist?list=PLml9VdOeqKa_5QWbelpO5Wp8MctLt5EdQ, which I did not open). I do not rely on it. d. If true? Conservation of energy (first law) for a circuit with only a battery and passive parts. Not just the second law: with only a battery and a gate-drive supply as sources, there would have to be an unidentified energy source.

Verdict. Split: - The measured numbers as a demonstration of net energy gain: Refuted by calculation as an energy-source claim. The load can only dissipate energy supplied by the battery plus the 0.49 W gate battery. At 2.34 A peak, 17.7 W in the load needs about 32% effective duty, which draws about 17.7 W from the battery, not 1.13 W. The stored-energy bound (4.7 W at 200 kHz) is recycled battery energy, not a new source. So the 1.13 W figure must be an error (probable causes: sample window under one cycle, shunt/ground-loop artefacts, sign convention). I cannot say which from the documents; the raw data are not available. - The measurement itself (what the instruments actually would read): Never properly tested by a neutral party. No controlled independent test is on record in the folder; the authors' own invitation to replicate was not accompanied by data. Dismissal: Ainslie says Stellenbosch refused to look (claim 6). The cheap decisive test (battery ampere-hours, or a true-RMS/DC power meter on the supply) was available to the authors and not done.

Claim 2. The gain comes from counter electromotive force (back EMF) of the switched inductor, with returning energy partly recharging the battery, growing with frequency.

a. Math consistent? Back EMF itself is textbook: V = L dI/dt. For a current falling from 2.34 A through 8.64 microhenry in 0.1 microsecond, V = 8.64e-6 x 2.34 / 1e-7 = 202 V (clamped in practice by D1 and the MOSFET). That is real. But no equation in either article connects back EMF to net energy; the "model" in the appendix states no mathematics (the authors say so) [read]. The claim that energy "returns" to the battery while the main flyback diode D1 recirculates the current locally (D1 is across the load, not the battery) is not explained. Current reaches the battery only through the MOSFET body diode when the drain swings below the source, as in ringing, which would be a tiny energy. b. Consistent with experiment? Flyback and inductive kick are among the best established phenomena (ignition coils, switching regulators). What is not established is a net gain from them. The classical energy balance of an inductor: energy out per cycle equals energy in (minus losses) [memory: high; Griffiths E&M ch. 7 magnetic energy 1/2 L I^2]. c. Tested? Only by the authors. The statement that gain grows with frequency is untested: no frequency sweep table is given [read]. The 2009 paper says gain depends on frequency, duty and inductance but gives no data. d. If true? Energy conservation in L-R circuits; Faraday/Lenz law as applied in circuit theory would have to give extra energy beyond the stored field energy.

Verdict. Split: back EMF existing and returning a transient current: Established. Back EMF producing net energy beyond the battery's: Refuted by calculation (the energy bound above: even 100% recovery per cycle at 200 kHz yields 4.7 W, all of it battery energy already counted). Frequency dependence of the gain: Never properly tested.

Claim 3. With the gate resistor tuned the circuit self-oscillates far above the 555 setting (143-200 kHz in 2002; over 500 kHz per Kelly), and this is what makes the gain appear.

a. Math consistent? The documents conflict. The 2009 reprint quotes only 2.4 kHz, 3.7% duty and "random oscillation" when the gate resistor is raised [read]. Quantum (per Step 1) gives 143-200 kHz at ~1.3% duty. Kelly says over 500 kHz at ~55% duty [read, per Step 1]. These are different operating states, with 40x different duty, yet all are said to give the same gain of "17". The 2009 text drops the separate 12 V gate battery that the 2002 article discloses (I confirmed the reprint circuit and parts table show only BAT 1) [read]. Note a 55% duty (Kelly) would give a battery power of about 31 W and cannot coexist with 1.13 W [calculation, from background]. So Kelly's description (if it were of the same circuit) is not consistent with the 17x-ratio figure and the two cannot both describe the same operating point. b. Consistent with experiment? Parasitic oscillation of a MOSFET with a gate resistor and a drain inductance is established engineering practice [memory: high]; it is also known that a 555 output driving a MOSFET gate with high series resistance can produce linear-region partial switching. That a MOSFET in partial conduction dissipates heat in itself is also standard. The claim that only NE555N chips do it is not tested (Kelly, anecdotal). c. Tested? No oscillation spectrum, no scope screenshots in the folder, no data on conduction losses. No independent replication known to me ([memory], low; web check failed). d. If true? Nothing: parasitic oscillation is mainstream. What would change is only the claim that it breaks energy balance.

Verdict. Split: self-oscillation under detuned gate drive: Plausible, untested (mainstream-credible, well known as a failure mode; no recorded spectra). That the oscillation yields extra energy: Never properly tested. The description across documents (2.4 kHz / 143-200 kHz / 500 kHz; 3.7% / 1.3% / 55% duty) is Internally inconsistent as a record of one experiment, which matters because the replicator cannot tell what state was measured.

Claim 4. The result breaches unity; conventional energy-transfer models and the second law do not constrain electromagnetic energy transfer; the extra energy comes from an unseen medium (Kelly's reading).

a. Math consistent? No mathematics. The 2009 paper says the circuit is "not constrained to the principles governing the Second Law" because "all magnetic fields comprise particles" [read]. That is a non sequitur: the second law is a statement about entropy that applies to systems of particles; and the issue in the measurement is the first law (energy conservation), which is not addressed. The paper's own sentence "there are no engineering publications available for citation. It is hoped that this lack will obviate the need for references" [read] states the work does not engage the literature. b. Consistent with experiment? The first and second laws, including for electromagnetic energy transfer, are among the most heavily tested; Poynting's theorem (energy balance of the E&M field) is an exact consequence of Maxwell's equations [memory: high; Griffiths E&M ch. 8]. "Environmental energy" has no measured source in these documents. c. Tested? The claimed environmental inflow has not been measured. A test: shield the apparatus and compare; none recorded. d. If true? Maxwell's equations/Poynting's theorem and the first law for circuits; thermodynamic limits.

Verdict. Split: unity breached by the circuit: Refuted by calculation (see Claim 1). That second law does not constrain EM energy transfer: Internally inconsistent (the argument given does not follow, and the real problem is energy conservation). Extra energy from an unseen medium (Kelly): Not testable as stated; no mechanism, no measured quantity, no prediction given.

Claim 5. Current is the movement of discrete magnetic dipole particles ("zipons", tachyons at about twice light speed); magnetism is primary; the model predicted the over-unity result and may resolve non-locality, dark matter and the proton/electron mass ratio (proton = three electrons).

a. Math consistent? There is no mathematics [read: "no part of the model is developed mathematically" in the 2009 reprint appendix; the Unifying Field Model is described in Step 1 as verbal and numerological, and the mass-ratio assignments (zipon 0.5, electron 1.5, proton = 3 electrons plus extra "truants") were not derived]. I verified the text layer states the universe is "a single magnetic dipole moving at twice the speed of light" and an experimental "proof" in the appendix [read: pdftotext, lines 8, 26-28]. A basic arithmetic check: three electrons have charge -3e (and 3/2 spin units combined), not +1e of a proton, so "a proton is three electrons" fails on charge alone unless extra "truants" cancel it, which is not worked out [calculation]. b. Consistent with experiment? No for specific parts. Deep inelastic scattering (SLAC, 1960s-70s) shows proton substructure of quarks with fractional charges [memory: high; standard textbook, e.g. Griffiths, Introduction to Elementary Particles]. Electrons have been found pointlike to below 1e-18 m [memory: medium]. Tachyons have never been observed. Entanglement correlations (Bell-test experiments, e.g. Aspect 1982, and loophole-free tests 2015) do not allow signalling, which is standard; the paper's claim that entanglement "shows superluminal communication" is contradicted by the no-communication theorem [memory: high]. The model also does not yield the magnetic-monopole-free Maxwell equation structure. c. Tested? The only claimed test is the heater experiment (Claim 1), which the paper itself says may "prove" rather than demonstrates. Since that experiment fails the energy budget, it cannot support the model. d. If true? Special relativity, the Standard Model, QED, charge conservation, Bell-test interpretation, much of particle physics.

Verdict. Split: that the model predicted over-unity and is supported by it: Refuted by calculation (the claimed supporting experiment fails the energy budget). Proton = three electrons: Tested, not supported (scattering experiments; charge). Zipons, tachyonic dipoles, resolving dark matter/non-locality: Not testable as stated (no equations, no predictions; vague; "in defiance of the prescriptive use of mathematics" [read]). Within the author's framework, the text is also Internally inconsistent where it says current is "movement of magnetic fields" while requiring electrical potential as a "quantum of imbalanced fields" without definition.

Claim 6. Mainstream institutions refused to test or evaluate the claim (Stellenbosch refused a demonstration), while companies (ABB, BP, Sasol, Fluke distributors) tested or supported it.

a. Math. Not applicable. b. Consistent with record? The Stellenbosch refusal is Ainslie's statement as quoted by the Pretoria News [read: page image]; no response from the university is quoted. The company roles are listed in Correction 4: calibration, attendance, limited tests of unnamed results, a bursary offer. MTN ScienCentre's Prof Bruton says "Physics is not my field of expertise" and that there is "a case for serious debate"; the centre also notes that Ainslie "couldn't give us the oscilloscope" or the computer on which she displays the waveforms [read], so the exhibit did not include the measurement itself. c. Tested? By whom? No test reports are in the folder. The newspaper's statement that "the experiment has been demonstrated to representatives of BP, Sasol, ABB, Spescom and Fluke" is a claim about demonstrations, not about measured confirmation. The editor of Quantum: "It works ... But I cannot yet say why or how" [read] is an opinion, not a measurement. d. If true (as suppression)? Nothing in physics; it concerns how anomalous claims are handled. (A refusal to look is still not a refutation: the energy-budget argument above is independent of who looked.)

Verdict. Whether it was tested neutrally: Never properly tested (there is no neutral controlled test on record here; the Stellenbosch refusal is unverified, so I cannot say it was "dismissed without testing" beyond Ainslie's own report). The corporate-support claim as stated: not supported by the sources (see Correction 4).

Summary table

What single test would settle it

Run the circuit with a true DC energy accounting at the battery: an integrating coulomb counter (ampere-hours) and a battery discharge to a defined cutoff alongside a calorimetric measurement of the load, with the gate-drive supply included. If the battery drains 12 Ah over 16.6 h at 17.5 W of load heat, the 1.13 W scope figure is the error. The authors did not run it and omitted it as "too many vagaries".