Stiffler: Physicist review (Step 2)

Reviewer: PhD physicist role. Date: 2026-10-08. Project: Stiffler (Ronald R. Stiffler, Stiffler Scientific, Humble, Texas; died 2018-11-12 per the Investigator's obituary lead).

Tags: [read] from the documents or the Step 1 deep reads; [web] with URL; [memory] with confidence; [calc] my own arithmetic.

0. Scope, stake and source gap

• Source gap (repeated from claims.md): all nine documents are the inventor's own pages and white papers. He is an interested party (he sold boards and offered licences through Stiffler Scientific [web: https://www.zpenergy.com/modules.php?name=News&file=article&sid=3221]). The folder holds no independent measurement. Figures in the saved pages (ECAT scope photos, SEC Special Coil Fig 11/13/16) were not rendered by Step 1; numbers in them are unknown. The Investigator's finds are leads, mostly forum posts and videos by replicators, and none contains an energy balance.
• Where Stiffler himself disclaims over-unity I say so. Energetic Forum thread 4067, post #1 (2009-03-12): he does "not believe in OU", CEC is a "conversion ... with a 'mathematical' gain. No creation of energy"; post #2122 (2011-05-30): a CEC above 1 cannot simply be looped back, because that "would require manufacturing energy" [web: https://www.energeticforum.com/forum/energetic-forum-discussion/renewable-energy/4067-high-voltage-from-thin-air]. The documents themselves nevertheless use "excess energy", "anomalous power gain", "COP" and "Alternative Energy Source" [read]. A reader should hold both statements.
• Web verification (budget 6; used 2). (i) Medium-confidence [memory] that Rosemary Ainslie published a claimed over-unity MOSFET heating circuit c.2002: a search returned only a second-hand excerpt calling the circuit "somewhat difficult and sensitive to properly adjust" to reach astable oscillation; no independent replication report found. Check failed to retrieve primary sources; I do not use Ainslie's own claims as evidence anywhere below. (ii) Single-wire transmission literature: Strebkov, Avramenko and Nekrasov, "Single-wire electric power system for renewable-based electric grid" (2000) exists, patent US6104107 (Uniline Ltd., S. and K. Avramenko) exists, and one study reports "the current is reactive and capacitive in nature" with a 90 degree phase shift between standing-wave current and voltage; I did not read their full texts [web: https://patents.google.com/patent/US6104107 ; https://iitf.lbtu.lv/conference/proceedings2008/Papers/13_VoitkansGreivulis.pdf]. Everything else I state is [read], [calc], or high-confidence textbook physics ([memory, high]).

1. The mainstream view, stated fairly

1. Energy conservation (first law) for an open circuit: average input power from the supply must be measured as the time average of v(t) i(t). If the supply, coil, MOSFET and load are the only energy paths, average output (heat, light, radiation) cannot exceed average input plus any stored-energy change. A device can in principle draw energy from an environmental source (a solar cell does), and mainstream physics does not forbid that; what it demands is an identified source, and a measurement that rules out error. Stiffler's "Energy Lattice" and "virtual energy" are not identified to that standard.
2. Resonant circuits give voltage magnification, not power gain. A series-LC or tuned-coil node of quality factor Q has voltage up to about Q times the drive, with current set by the source; delivered power is limited by what the source supplies. [memory, high]
3. High-voltage, high-impedance RF nodes light neon lamps and LEDs with milliwatts or less. A neon lamp glows at well under a milliwatt; an LED glows visibly at 1 mA. [memory, high]
4. Single-wire (one-conductor) transmission is real engineering, not a violation: the "return" path is displacement current through the capacitance of the terminals (here the "top hats") to the surroundings and to each other. This is Tesla's and Zenneck's territory and is the subject of the Strebkov/Avramenko literature [web, see 0(ii)].
5. Real power in an AC circuit is P = V I cos(phi), where phi is the phase angle between voltage and current. Multiplying RMS V by RMS I gives apparent power (VA), which is never less than P. In a high-Q resonant circuit cos(phi) is tiny, so apparent power can be 10 to 100 times the real power. [memory, high]

2. Worked checks used below

(A) Phase error in RF power measurement [calc]. At f = 13.56 MHz a timing skew of 1 ns between voltage and current channels (probe or cable difference) is 360 x 13.56e6 x 1e-9 = 4.9 degrees. Apparent real power then differs from the truth by V I sin(4.9 deg) = 0.085 V I. For a resonant load with true phase 89 degrees (true P = 0.017 V I), the skew error (0.085 V I) is five times the true power and can have either sign. The same skew could even make the power appear negative, as one replicator reported (Energetic Forum thread 9344, post #316, 2019-12-01 [web]; see Investigator note). So "negative" power computed from skewed V and I is a known artefact, not evidence of a generator.

(B) Stiffler's input formula [read: SEC Special Coil, Fig 14]. "Pin = Vin x Iin", magnitudes of RF meter readings, no phase. This returns apparent power. If the load phase is 80 degrees, Pin overstates P by 1/cos 80 = 5.8; at 89 degrees by 57. An input estimated this way would overstate input (the opposite error from over-unity). But the same method applied to the output side, or any step that mixes "power" computed this way with a true-power output, can produce any ratio. Which side Stiffler used for which quantity is not stated.

(C) ECAT accounting [read: ECAT page; calc]. Stated: input 8.25 W; lamp 12 W; matching transformer "wastes" 6 W; "total power produced 18.11 W"; gain 18.11/8.25 = 2.195. If the input is only 8.25 W and 6 W is dissipated in the transformer, at most 2.25 W reaches the lamp. The 12 W figure is the lamp's nominal rating at 120 V RMS 60 Hz sine, not a reported measurement. For 12 W + 6 W to be real, the supply would have to deliver 18 W, i.e. 1.5 A at 12 V. The page gives 90 mA quiescent (1.08 W). 12 + 6 = 18, not 18.11.

(D) ExHeatExp calorimetry [read: Fig 15; calc]. Q = c m dT = 4.187 x 25 x (41.25 - 26.15) = 1580.6 J. Input = 1198.759 J over 600 s (mean 2.0 W). CEC = 1.319. Sensitivities of that ratio: - Mass entered: ratio is proportional to m. The text says 20-25 mL; at 20 g the ratio is 1.055. - Temperature: for the ratio to be 1.00 the water's mean rise would be 15.10/1.319 = 11.4 K. The sensor sits in the top ~5 mL of an unstirred 20-25 mL tube heated from an outer wall coil; warm water rises, so the top reading can exceed the volume-mean temperature by several kelvin. A 4 K stratification would account for the entire excess. [calc; stratification size is my estimate, not measured] - Input: if the 555 driver ran from a separate 12 V supply (the "split supplies" run), its draw (say 10 mA x 12 V = 0.12 W, my assumption) is about 6% of 2 W and would be omitted from "PSP Charge Supply" Sum(V I T). That accounts for about 6%, not 32%. - Heat sinks I can not use to explain it: heat loss and the glass heat capacity make true heat larger than Q, so they push the ratio up, not down. A genuine excess, if the thermometer and the input measure are both right, would therefore be somewhat larger than 1.32; equally, the reading may be wrong in either direction. This is why a calibration run (known DC power into the same coil and tube) is the decisive control, and the page reports none. - RF pickup on the LM335 leads (a toroid filter is shown in Fig 18) can add a spurious offset to a temperature reading. Whether the "calibrated before each run" LM335 was calibrated with the circuit running is not stated.

(E) Capacitance of a top hat [calc]. A conductor of radius 2.5 cm in free space has C = 4 pi eps0 r = 2.8 pF. At 1 MHz its reactance is 57 kOhm; a voltage of 140 V RMS drives about 2.5 mA of displacement current through it. This is consistent with a 48-LED series string glowing dimly from a coil, with no power gain. The guide gives no frequency, so the numbers are illustrative.

(F) The 62-LED arithmetic [read; calc]. 168 V / 2.7 V = 62. For the string to light at 1 mA it needs about 0.17 W. The paper never reports that a 62-LED string was lit, and the only measured input implied is about 47 mW dissipated in the 50 Ohm resistor R1 (4.32 V p-p). So the sentence "correlates to a maximum series LED string of approximately 62" is a voltage-capacity statement only; it is not a power claim.

3. Claim by claim

Verdict scale: Established / Plausible, untested / Never properly tested / Tested, not supported / Refuted by calculation / Internally inconsistent / Not testable as stated / Historically superseded but instructive.

Claim 1. ECAT: air-core MOSFET coil circuit gains power by "Poly-Phasing" (8.25 W in, 18.11 W out, gain 2.194)

a. Mathematics consistent? No equations exist for "cross phasing", "virtual energy pump" or "Poly-Phasing" [read]. The only mathematics is the energy table, and it is internally inconsistent: (C) shows that if 6 W of an 8.25 W input is lost in the transformer then the lamp cannot receive 12 W, so 18.11 W is formed by adding a loss to a load; 12 + 6 = 18, not 18.11; 8.25 W at 12 V is 0.69 A but 90 mA is quoted (the deep read says "under load"; the page does not make clear whether the 8.25 W and the 90 mA belong to the same operating condition). b. Consistent with experiment? A real COP of 2.2 would contradict energy conservation unless a source is identified. Nothing on the page excludes the ordinary explanation: pulsed input into a 2000 uF/0.01 uF filtered rail measured by an unspecified method, and a lamp "12 W" taken from its label. A tungsten filament lit by rectified or pulsed high-frequency waveforms does not follow its 60 Hz nameplate power, and the page notes "high voltage arcing at the base of the bulb". The author himself reports that ferrite or metal-core versions lost the effect [read]; that is what one expects if the effect comes from the circuit's resonant or ringing behaviour (core losses damp ringing), but it does not discriminate an error from a gain. c. Tested? By the author only, 1993-96 (first tries 1990 per his 2009 forum post [web, post #369, page 25 of thread 4067]), on a single figure for one operating point; no meter types, no waveform integration, no calorimetry, no error bars. He reports that Internet attempts by "Free Energy" builders failed because they changed the design (no attempts named) [read]. No replication of the intact design exists in the record I have. Not replicated. d. If true: energy conservation would need a new source with a mechanism (a field or medium that can be tapped and is not accounted for). Stiffler's own 2009 text says the "math could not support the (conversion) concept" and that he changed the name from Conversion to Coherence [web, post #369]. Taken as written, that is the inventor's report that the theory did not close.

Verdict: Never properly tested (strong form: a COP of 2.2 from ordinary 12 V input). The figures as printed are internally inconsistent (accounting in (C)); that is a defect in the evidence, not a refutation of every possible version of the circuit. The claim was not dismissed untested by others in the record; it was largely ignored, and the one third-party mention in the record is the author's own statement that copies failed.

Claim 2. SEC: a coil with an Avramenko plug taps an "Energy Lattice"; output is "cohered" energy, CEC = Uout/Uin; 130-168 V at the free end

a. Mathematics consistent? CEC% = (Uout/Uin) x 100 is stated with "Joule Seconds" in the equation but energy units in the note; joule-seconds are units of action, not energy [read]. U is not defined as a symbol. CEC is never evaluated in the white paper. The claim in the Measurement Methods paper that UCEC > 1 implies PCEC > 1 holds only if the two energies are accumulated over the same interval. Eq 9 there (0.62 sqrt(D^3/pi)) is dimensionally wrong as printed (should contain wavelength, not pi) [memory, high]. The scope data are a clean 4.32 V p-p at points A and B and 26.5 V p-p at C (ratio 6.1) [read]. That is voltage magnification at a series-resonant node. The resonant frequencies the author computes for his listed parts (2.68, 2.06, 1.27 MHz [calc by deep reader]) do not explain a 10-12 MHz peak; stray capacitance and the diode capacitance are unmeasured, so this is a gap in the model, not an inconsistency proof. b. Consistent with experiment? Everything reported is consistent with ordinary RF behaviour: step-up of voltage in a resonant coil, a neon lamp lit by body capacitance from a 130 V open end, a diode-plug rectifier lighting an LED. A white LED glows "intensely" at a few milliwatts. No result conflicts with energy conservation because no energy balance is reported. The Fig 18 "frequency response" plots five points with the axes swapped and a unitless 0-4 "relative power" axis whose values sit exactly on 0,1,2,3,4 [read: page image p.14], which look like index labels, not measurements; it supports at most a broad LED-brightness maximum near 12 MHz. Output measurement by LED light calibrated against DC-driven LEDs assumes light output tracks average electrical power at the LED; with a rectified 12 MHz waveform of unknown peak and duty this is an unvalidated assumption (for LEDs the peak-to-average ratio matters for efficiency). Input measurement as defined (Eq 3, integration across a 1 Ohm shunt) is the better half of the methodology; the output half (Eq 1: P = Vr^2/R; Eq 2: LED V x I) is DC-style and the paper does not say how RF at the load is converted. c. Tested? By the author (no energy balance published); informally by replicators. Investigator leads: Lidmotor's builds (SEC replication 2009-04-07; diode electrolysis 2010; 100K-resistor partial replication 2011) and forum reports (Raui 2011, Inquorate SRF measurement, zeos71's ~130 V p-p on L1 and ~300 V p-p on L2 of an SEC15-3) [web, Investigator notes]. They show that the circuit builds and lights lamps and LEDs, so the described effects are reproducible by others in the sense of "it glows". None reports input and output power. A replicator reading of the SEC loop as "an Avramenko plug equivalent" and one of "power always negative" at some frequency (post #316) are consistent with a phase-measurement artefact (2(A)). Stiffler's 2018 late measurement relayed by a replicator is input only (35 mA at 18 V = 630 mW for a CREE LED unit). Stiffler's 2010 plan to publish a paper from replicators' data was not fulfilled in any source I can see; the Investigator's analysis (death on 2018-11-12 is the obvious reason the record stops) stands, and I add that the paper was announced in 2010 and the white papers continued to 2018, so the gap is not explained by death alone; the record does not say why no energy-balance paper appeared in the eight years between. d. If true: a medium (the "Energy Lattice") that stores recoverable energy in space and releases it to a diode plug would have to exist. Stiffler's own descriptions of this ("Spatial Gate", "negative resistance" per the Panacea course snippet [web, Scribd]) are not defined quantitatively anywhere. The claim "an SEC exciter is a Negative Resistance" is testable: it requires measured V and I of opposite phase at the port with true average power flowing out; see 2(A) for the main artefact.

Verdict: Split. (i) The phenomena actually shown (voltage step-up, lamps lit, LEDs lit, one-end excitation): Established as ordinary resonant/capacitive behaviour, no anomaly demonstrated. (ii) The strong form (output energy exceeds input, drawn from an "Energy Lattice"): Never properly tested (no input/output balance published; CEC defined but never evaluated). (iii) The "Energy Lattice" as stated: Not testable as stated (no measurable prediction distinguishes it from resonance and rectification).

Claim 3. Stacking AV Plugs multiplies "excess energy": two plugs double it, three triple it

a. Mathematics consistent? The worked example is ordinary capacitor arithmetic E = C V^2/2. Check: 20 uF at 135 V = 20e-6 x 135^2 / 2 = 0.18225 J, not 1.8225E-2 J; 10 uF at 270 V = 0.3645 J, not 3.645E-2 J. Both printed values are 10 times too small; the ratio of 2 is unaffected [calc, confirms deep read]. The example is internally coherent as stored-energy bookkeeping: two 20 uF capacitors each at 135 V store 2 x 0.18225 = 0.3645 J, the same as the series pair at 270 V. It contains no input measurement and assumes 135 V. b. Consistent with experiment? Stacking diode-capacitor stages to multiply voltage is standard (Cockcroft-Walton multiplier). In an ideal multiplier the energy delivered per cycle to the stack is bounded by what the oscillator supplies; N stages with N capacitors store N times the energy because they are N capacitors charged to full voltage, which requires N times the charge to be delivered, not a gain. The statement "each AV Plug is capable of providing what one single plug would" is a statement about the output of a source with a fixed energy per cycle only if the oscillator supplies N times the energy; the page gives no measurement that it does not. The "empirical formula" announced for the stacking limit is not on the page (raw HTML checked by the deep reader). The three-stage claim rests on a photograph of neon brightness, which depends on camera exposure. c. Tested? Only by the author's photograph and stored-energy arithmetic. No replication of a stacked unit appears in the record I have. d. If true: only the "excess" part would require new physics; the doubling alone requires none.

Verdict: Split. Doubling of stored capacitor energy for two capacitors: Established (trivial arithmetic, exponent misprinted). "Multiplication of excess energy by the number of plugs": Never properly tested (no input measurement, formula absent).

Claim 4. Energy through one 12 ft wire (24 V transmitter, 48-LED receiver, no return conductor)

a. Mathematics consistent? No equations; no frequency, inductance, current, or power is given for L3/L4 (143 turns on a 12.5 mm Lucite rod, tapped at 37) [read]. Can't check; 2(E) shows an illustrative displacement-current figure. b. Consistent with experiment? Yes, in the ordinary sense. Two resonant coils with capacitive top hats joined by one wire are a classic Tesla-type single-wire link; the return path is displacement current through the top-hat capacitances to each other and to the surroundings. The inventor's own notes support that picture: one sharp tuning point, hand capacitance detunes C2, nearby metal "degrades the operational efficiency", and warnings that the fields cause headaches and damage nearby electronics [read: construction guide pp.8-10]. These are the behaviours of an electric-field-coupled tuned system, not a conduction circuit. Whether the field reaching the receiver is carried mostly by the wire or by free space and ground is the question that matters; the guide does not measure it, and a claim of "one-wire" transmission strictly needs shielding or distance tests the guide does not give. c. Tested? By the author (video y9aJohml130, now removed [web, Investigator]) and by replicators: Lidmotor's one-wire builds at 10 ft (2009-07-31) and at 25 ft around a wall [web]; both show LED brightness only. The investigator also found mainstream-adjacent patent literature on the same arrangement (Strebkov RU patents; Bank US9608441B2) [web, Investigator]. This supports that the effect is reproducible and that others independently work on it; it does not support an efficiency figure, because none exists. d. If true as stated: nothing would change. Single-wire capacitive coupling is accommodated by Maxwell's equations (displacement current closes the loop).

Verdict: Established, as capacitive/resonant coupling over a short distance at low power. The implied claim of high-efficiency power transfer: Never properly tested (no power, no efficiency, no measured loss).

Claim 5. Ainslie MOSFET heating circuit: heat in water exceeded measured electrical input (CEC 1.319); returned energy is insignificant

a. Mathematics consistent? The arithmetic closes: Q = 4.187 x 25 x 15.10 = 1580.6 J; CEC = 1580.593/1198.759 = 1.3185 [calc, confirms deep read]. Internal oddity: the scope readout in Fig 11 (19.0 us at 3.80%, which implies a 500 us period, versus the annotated 2.38 kHz) disagree [read]; trivial. The returned-energy argument (flyback diode D1 keeps the field-collapse current in the coil loop) is circuit-correct in outline: the current continues through the coil/resistor/diode loop and dissipates in the NiCr wire and D1 as heat. No current integral is shown. His assumption of 1 kV breakdown while measured spikes are 58-149 V is harmless. b. Consistent with experiment? A pulsed resistive heater must obey P = average of v i. A CEC of 1.32 is not consistent with that unless an error exists in the input integral, the temperature, the mass, or the calorimeter. 2(D) lists the candidates: the unstirred tube with a top-of-tube sensor (a stratification of about 4 K would erase the whole excess), the ratio's proportionality to the mass entered (20 g gives 1.055), a possible omission of the driver's supply, RF pickup on the LM335, and 1.49 MHz ringing on the supply that a once-per-second DC reading cannot capture (the footnote says AC components are "taken by signal integration at 40 MHz sampling", but it is not stated that Fig 15's input came from that integration rather than from the supply meter). In the cup arrangement the ratio fell and levelled near 0.83 above 42 C; he attributes that to loss [read]. A true free-energy effect would not be expected to vanish at 42 C with the same circuit; ordinary heat loss, which grows with temperature difference, fits. c. Tested? One run, one screenshot, by the originator of the experiment, no calibration run (no DC reference), no blank, no repeats, no errors; the page is unfinished (under-construction icon). Stiffler himself writes no sentence about the 1.319 result [read: deep read]. The Ainslie group's own paper (Dec 2009, COP above 4.0, submitted to IEEE per the document; Stiffler's name appears in the Google snippet only, role unconfirmed) is by interested parties and I have not read it; I could not retrieve any independent test or critique of the Ainslie circuit (web check failed, see section 0). Not replicated. d. If true: the first law would need an unmetered input or a hidden source. The honest reading is that the experiment never reaches the stage where that question arises, because the controls that would exclude error (DC-equivalent calibration, stirring, repeat, input waveform integration shown) are not there. Note also that Stiffler's own words in the 2011 forum post say CEC above 1 does not allow looping.

Verdict: Never properly tested (strong form: excess heat from the Ainslie circuit). The CEC 1.319 result is a single unverified measurement with plausible ordinary error sources in the temperature and mass terms; it is not a refutation and it is not support. "Returned energy insignificant": Plausible, untested (circuit reasoning sound; no integral shown).

Claim 6. Instruments must be battery powered, ungrounded and RF filtered, because grounding "will detune the AV Plug to where it no longer will interface with the Spatial Lattice"

a. Mathematics consistent? Prescriptions are not mathematical. The SEC Special Coil impedance tables use Z = sqrt(R + |X|) instead of sqrt(R^2 + X^2) (the deep read verified all 90 rows); every Z is wrong (about 9 times too small at |X| = 80 Ohm), though the minima fall at the right frequencies (8.90 MHz for 400 pF with 800 nH, etc.). "2.2 mH" in the header is 2.2 uH. The "amplification" frequencies of 12.0 MHz are numerically the +6 dBV edges of the secondary response, not separate peaks. "Amplification" has no stated units. b. Consistent with experiment? The conventional content is sound practice: a high-impedance RF node is loaded and detuned by a grid-powered meter's capacitance to ground, probe capacitance, and leads; floating battery instruments, ferrite beads and short leads reduce that; the paper's own advice to move the meter and see if the reading changes is a good test. Several of his points are the kind a competent RF engineer would give. Where it departs is the explanation: loss of interface with a "Spatial Lattice". The conventional explanation (stray capacitance and ground-loop current loading a high-Z node, with a common-mode path through the instrument) predicts the same observations. The consequence for testing is important: if the exciter works only when no conventional instrument touches it, then the right test is a black-box one that needs no probe inside the circuit: measure the DC supply input at a filtered point (the paper already does this) and measure the output as total heat in an enclosed calorimeter. The thermal-bottle test in the SEC Special Coil page is exactly that design (a 2 W resistor at matched DC power versus the circuit board, 4-minute readings for one hour) [read]. The page states no result. The quick LED/black-plate test showed no temperature rise [read]. A calorimetric result that is stated as obtained but not reported is an anomaly in the record; I have no basis to say whether it was run. c. Tested? Not independently. The papers concede that field pickup can give "meaningless" readings. d. If true: only the "Lattice" part would change physics; the instrument-handling advice changes nothing.

Verdict: Split. Instrument-handling advice (floating, filtered, short leads): Established, ordinary RF measurement practice. The reason given (grounding detunes the "interface with the Spatial Lattice"): Not testable as stated, because the Lattice has no measurable property apart from the effects it is invoked to explain. Because the same fields that corrupt instruments corrupt his own claimed gains, his measured numbers need a calorimetric or other black-box check; none is reported.

Claim 7. Stiffler's reports on testing: copying attempts failed because builders changed the design; Naudin's TEP "fell short"; the "acid test" of self-running is a diversion; quick tests showed no heating

a. Mathematics consistent? Not applicable (no equations). b. Consistent with experiment? The statement that CEC > 1 need not mean a self-running device is physically weak. If the gain were real and the output were usable electrical power at conversion efficiency above 1/COP, closing the loop (feeding output back to input) would remove the external supply and also eliminate every measurement-error explanation at once, since a system either keeps running or it does not. At COP 2.194 and a conversion efficiency above 46% the loop closes; ECAT's output is 120 V AC at a lamp, so rectification and regulation is routine engineering. That a self-run loop is "not possible" for a COP above 1 (his 2011 post) is not a conservation-of-energy argument if the energy comes from an outside source. I say this without deciding who is right: the cleanest experiment is one the inventor declined to treat as the test. On the other side of the ledger, his statement that his own quick tests found no temperature rise in the black-plate test is evidence against a gain and is kept in the record. c. Tested? His reports on others' attempts are unsourced (no attempts named); the Investigator found TEP material by Naudin live at JLN Labs with no Stiffler mention, and ctglabs.com empty [web, Investigator]. I have no independent source on whether the ECAT circuit was ever copied faithfully. Statements about suppression ("the acid test is a diversion by those who wish to suppress research") name no person or document. The Investigator's finds on the historical record do not show suppression: Stiffler's channel stayed online for years, his work was posted openly on forums, and the record stops with his death in 2018 [web, obituary note]. Google removed some search results after legal requests (Lumen notices 75850580, 95550170); who asked is unknown (page blocked) and this stays an open lead, not evidence in either direction. d. If true: nothing beyond Claim 1.

Verdict: Never properly tested (self-running / loop-closure test never performed; replications of ECAT not on record). Dismissed without testing: not shown; the record has no dismissals by mainstream bodies, only silence.

4. Investigator's findings I weighed (kept whatever the verdict)

• Inventor's own further work: forum posts 2009-2011 (CEC as conversion, ECAT start in 1990, notebooks from 1985, the "Urban Myth" post); the 2010 LED-array paper and the Panacea-BOCAF course (88 pp.) with "Patent Pending: 61254675" and the "negative resistance" and NILS claims; 163 YouTube videos including the 2017-2018 "Stiffler Loop" series; a CREE-board input figure of 630 mW. None gives an output power. These bear on claims 1-2; the "Is the Loop really doing anything?" title is a question the inventor himself asked and is worth reading.
• Support by others: Lidmotor (replications and one-wire at 10 and 25 ft), Raui, Inquorate, zeos71, mikrovolt (signal-generator-driven loop at 13.53 MHz, 24.4 V p-p), and the Ainslie group's 2009 paper. These show that the visible effects reproduce; none shows an energy balance. The two forum readings about the loop reducing to an Avramenko plug equivalent and about "negative" computed power are, in my reading, ordinary: the first says it is a diode-capacitor coupling, the second matches the phase-skew artefact in 2(A).
• Related work: Strebkov, Trubnikov, Bank, Egely, RSC Energia patents on single-wire resonant systems with Avramenko-type diode elements. The patents are real filings; patent grant is not evidence of efficacy and I have not read their claims.
• Historical record: death 2018-11-12 (full obituary not read; the identity rests on one notice plus a newsletter of 2019-01-24); provisional 61/254,675 with no published application (a lapsed provisional is common; USPTO Patent Center not yet searched); "Advanced Technology Group" of 1993-96 still unidentified; Stiffler Scientific has no corporate record found. Absence of evidence analysis: the post-2018 silence is explained by death; the pre-2018 absence of any published energy-balance paper after the 2010 announcement is not, and the competing explanations (no result that survived measurement; data kept for licensing; the author's health or other priorities) cannot be separated with the current record. The records that would decide: the archived stifflerscientific.com / drstiffler.com pages (Wayback was offline), the Panacea 2012 PDF, the "Series 18 Exciters Documentation" Rel 2.8, and the missing "Measurements for SEC Exciters" data.

5. Corrections to Step 1

1. claims.md item 5 says "the energy returned to the battery by the collapsing inductor is insignificant" as if established. It is Stiffler's assertion ("What I have confirmed ..."); no current integral for the battery side is shown. Restate as "asserted without a shown measurement".
2. claims.md item 1 says "arithmetic does not close: 12 W + 6 W is 18 W". That is a minor rounding point (0.11 W). The substantive defect is the accounting in 2(C): a 6 W transformer loss counted as "produced" power, which makes 18 W an addition of a load and a loss, not a measured total. facts.json should lead with that.
3. facts.json (ECAT) and claims.md say the quoted 90 mA conflicts with 8.25 W / 12 V = 0.69 A. The deep read records "quiescent current under load 90 mA" but the page's wording leaves open whether the two figures describe the same operating point. State it as "cannot be reconciled from the page", not as a proven contradiction.
4. claims.md item 3 says the exponent was "misprinted by a factor of 10". I confirmed this: correct values are 0.18225 J and 0.3645 J. facts.json should not repeat 1.8225E-2 J as right (it does not). No change needed there; noted as verified.
5. facts.json (ExHeatExp) is right that the digest missed Fig 15. Add that CEC as computed there depends proportionally on the mass entered and on a top-of-tube temperature in an unstirred tube (2(D)); facts.json mentions mass only.
6. The facts.json summary of SEC Special Coil says the Z column is "wrong" (verified). Add that the tabulated minima are in the right places, so the tables still locate series resonance (8.9 and 9.8 MHz for 800 nH with 400 and 330 pF) but cannot be used for Q or power.
7. Investigator's correction on ECAT dates (1990 versus 1993-96) accepted: the page's 1993-96 refers to circuits built with the Advanced Technology Group; his 2009 forum post says the first ECAT tries date to 1990 and notebooks to 1985. In his own post he also says the "math could not support the (conversion) concept", which should be quoted next to claim 1.
8. Investigator's lead says "Raui says he replicated 'some effects' ... and vouches there is 'something worth a further look'". Keep, but note that signal-generator drive plus FFT is not the exciter oscillator circuit, so it replicates the coil response, not the self-oscillating exciter.
9. claims.md "Source gap" lists the demonstration video as to be obtained: it has been removed from YouTube (Investigator); retain as a missing source, not a to-do.

6. Verdict summary