Primer: Ronald Stiffler's "Spatial Energy Coherence" exciters and the ECAT circuit

Companion files: arbiter.md (final verdicts, including its 2026-10-08 "QE revision", which this primer follows and does not change; this version applies the remediation in qe.md), physicist.md, mathematician.md, contrarian.md, sites.md and research/ (the Investigator's finds), and example.py, which reproduces every number in Section 5 and the supporting numbers in Sections 3 and 4:

.venv/bin/python review/Stiffler/example.py

Source gap (repeated from claims.md and arbiter.md). All nine documents in the folder are Ronald R. Stiffler's own saved web pages and white papers (c.2007-2009). He is an interested party: he promoted and sold SEC (Spatial Energy Coherence) boards, courses and licences through Stiffler Scientific [web: https://www.zpenergy.com/modules.php?name=News&file=article&sid=3221]. No document in the folder holds an independent measurement or replication. The figures the first pass left unviewed have now been read (QE revision, 2026-10-08) [read: deep reads of the ECAT and SEC Special Coil pages, images]. The ECAT page's schematic and scope photos give voltages only (TP-1 106 V, TP-2 172 V, TP-3 94.5 V, TP-5 0.31 V p-p (peak to peak) across an unvalued sense resistor R3, drain 205 V), show two sources (a pulse generator V1 driving the gate and the DC (direct current) supply V2), and include a bench-supply panel photo reading 25.0 V and 0.33 A, which is where the 8.25 W comes from; no figure shows a current trace or measures the lamp or the transformer loss. On the SEC Special Coil page, Figs 10, 11 and 13 (the "amplification" charts) plot frequency only, with no amplitude, gain or power; Fig 15 is a photograph of the thermal-bottle apparatus (meter digits unreadable) and Fig 16 a photograph of lit LEDs (light-emitting diodes), probably under the black plate. So "amplification" has no measured magnitude anywhere on that page, and the thermal-bottle result is absent from the images as well as the text. Still missing: the "Measurements for SEC Exciters" data, the "empirical formula" announced on SECExciter.htm, the demonstration video y9aJohml130 (now removed from YouTube), and any record of the "Advanced Technology Group" (1993-96). Rosemary Ainslie's own documentation of the heater circuit Stiffler tested (her 2002 Quantum article, patent and 2009 reprint) is not in this folder but is read in the sister review, review/Ainslie-Rosemary-Ann/, and is used below. Where a statement below rests on the Investigator's web finds (forum posts, video titles, Scribd snippets) rather than a document read in full, it says so.

Source tags: [read] = from the documents in the folder (via the Step 1 deep reads, or viewed directly for this primer); [web] = Internet source with URL; [memory] = recalled, with confidence; [calc] = arithmetic, reproduced in example.py.

---

1. The idea in one paragraph

Ronald Stiffler, a Texas experimenter who signed as "Dr. Ronald R. Stiffler, Senior Scientist, Stiffler Scientific", spent from about 1990 to 2018 building small electronic oscillators that drive coils at radio frequencies (RF; a few kilohertz in his early "ECAT" circuit, about 10 to 13 megahertz in his later "SEC" (Spatial Energy Coherence) exciters). His circuits light neon lamps and strings of LEDs (light-emitting diodes), produce well over a hundred volts at the free end of a coil fed with a few volts, and send power along a single wire with no return conductor. He believed these circuits gather energy that is already present in space (he called it the "Energy Lattice") and "cohere" it into the load, and he introduced a figure of merit, the CEC (Cohered Energy Coefficient), equal to output energy divided by input energy. He was careful to say this was not "free energy" or a perpetual motion machine: in 2009 he wrote "I do not believe in OU" (over-unity) and called CEC "a conversion of one energy form to another with a 'mathematical' gain. No creation of energy" [web: Energetic Forum thread 4067, post #1]. The reviews found that every effect he actually documents (voltage step-up, lit lamps, one-wire coupling) is what ordinary resonance, rectification and capacitive coupling produce, and that the "Energy Lattice" has no property anyone could measure. Only two of his documents put numbers on output and input together: the ECAT page (8.25 W in, read from a bench-supply panel; 18.11 W claimed out, with no stated output measurement) and one calorimetry run on a heater circuit (heat in water 1.32 times the electrical energy in). Neither was repeated, controlled or replicated. They are not refuted; they were never properly tested, and the second is a test a reader could repeat in a day.

---

2. What you need to know first

Energy, power and the first law. Energy E is measured in joules (J); power P is the rate of energy flow, in watts (W = J/s). Over a time interval, E = ∫P dt. The first law of thermodynamics says energy is conserved: in a circuit, the heat, light and radiation leaving must equal the energy delivered by the sources minus any increase in stored energy. A device can deliver more energy than you put in electrically if it draws on another source you are not counting (a solar cell draws on sunlight, a heat pump on outdoor air); what the first law demands is that the other source be identified and measured. Read: Schroeder, An Introduction to Thermal Physics, early chapters; Halliday, Resnick and Walker, Fundamentals of Physics, the chapters on energy and on the first law.

AC (alternating current) power and phase. For a voltage v(t) and current i(t), instantaneous power is p(t) = v(t)i(t), and average (real) power is the time average of that product. For sine waves with root-mean-square (RMS) values V and I and phase difference φ between them, P = V I cos φ. The product V I alone is "apparent power" (unit volt-ampere, VA). In a circuit that mostly stores and returns energy (a capacitor or inductor), φ is near 90° and cos φ is near zero, so a large V I can carry very little real power. Read: Horowitz and Hill, The Art of Electronics, chapter 1; any circuits text on phasors.

Resonance and Q. An inductor L (henry, H) and capacitor C (farad, F) together resonate at f₀ = 1/(2π√(LC)). Driven at resonance, the voltage across L or C can be Q times the drive voltage, where Q (the quality factor) is roughly the ratio of energy stored to energy lost per cycle (divided by 2π). A coil with Q = 30 driven with 4 V can show 120 V. This is voltage magnification, not power gain: the current is set by the source and the losses, and the power delivered is limited by what the source supplies. Read: Griffiths, Introduction to Electrodynamics, chapter 7 (inductance) and any circuits text on series and parallel resonance.

Rectifiers and voltage multipliers. A diode conducts in one direction only. Two diodes and a capacitor turn an AC voltage into a DC voltage; stacked diode-capacitor stages (a Cockcroft-Walton multiplier) give a multiple of the input peak voltage, with correspondingly less current. A capacitor charged to voltage V stores E = ½CV². An ideal multiplier delivers no more energy than its source supplies; it trades current for voltage. Read: Horowitz and Hill, the sections on rectifiers and voltage multipliers.

Displacement current and capacitive coupling. Maxwell added to Ampère's law a term ε₀ ∂E/∂t (ε₀ = 8.854 × 10⁻¹² F/m is the permittivity of free space, E the electric field): a changing electric field acts like a current. This "displacement current" is how AC passes through a capacitor, and it is how a circuit with one wire can still form a closed loop: the "return" path goes through the capacitance between the circuit's ends and their surroundings. An isolated metal sphere of radius r has capacitance C = 4πε₀r (2.8 pF for r = 2.5 cm). At frequency f, current through a capacitance C with voltage V across it is I = 2πfCV. Read: Griffiths, chapter 7, section on Maxwell's equations and displacement current.

Calorimetry. Heat absorbed by a mass m (grams) of water with specific heat c (4.18 J per gram per kelvin near room temperature [memory, high]) when its temperature rises by ΔT is Q = c m ΔT. A calorimeter measures an unknown power by the temperature rise it causes in an insulated mass. The standard way to make the measurement trustworthy is a calibration or blank run: put a known power (usually steady DC into the same heater) into the same cell for the same time, and see what ratio of heat to input your apparatus reports when the true ratio is known to be 1. Read: any introductory physics text on heat capacity; for practice, the cold-fusion calorimetry literature is an instructive (and contested) case study.

Switching circuits and the flyback diode. A transistor switch (here a MOSFET, metal-oxide-semiconductor field-effect transistor) turns current through an inductor on and off. An inductor carrying current I stores ½LI², and when the switch opens the coil drives a voltage spike ("back EMF", EMF = electromotive force) to keep the current flowing. A diode across the coil (a flyback or freewheel diode) gives that current a loop to circulate in, so the stored energy is dissipated in the coil and diode instead of making a destructive spike. The duty cycle D is the fraction of each cycle the switch is on. The Ainslie heater, which Stiffler tested, is such a circuit; the separate primer review/Ainslie-Rosemary-Ann/primer.md works through its circuit theory in detail.

Near field and far field. A small antenna or coil has a "near field" region (within about λ/2π of it, λ = wavelength) where stored electric and magnetic fields dominate and fall off quickly with distance, and a "far field" where radiation dominates. At 13.6 MHz, λ = 22 m and λ/2π = 3.5 m, so a bench experiment sits entirely in the near field. Instruments and hands placed there become part of the circuit. Read: any antenna text (Balanis, Antenna Theory, chapter on field regions); Stiffler's own Measurement Methods paper discusses this.

---

3. The textbook view

3.1 What Stiffler built [read: deep reads]

• ECAT ("Energy Conversion by Articulated Transfer", page revised 2007, circuits from 1993-96, first tries 1990 per his 2009 forum post [web: Energetic Forum thread 4067, post #369]). The page describes two set-ups, and does not say they differ [read: ECAT deep read 2, images]. (i) A schematic design: a 12 V supply, a single MOSFET pulsed at 4.3 kHz with 50% duty by a separate pulse generator (V1), a large air-core "power coil" with an inner "load coil", a "Polyphase transformer", and test-point photos that repeat at about 4.3 kHz and give voltages only. (ii) A lamp version: a saturating "Theta core" transformer, a 60 Hz matching transformer and a 120 V, 12 W lamp, whose drain photos repeat at about 1.1 kHz (205 V peak), and whose bench supply (a Protek 3006B) is photographed at 25.0 V and 0.33 A, i.e. 8.25 W [calc]. Stated: input 8.25 W, lamp 12 W, transformer loss about 6 W, "total power produced ... 18.11 watts", "power gain ... 2.194". No coil dimensions, part numbers, current trace or output-measurement method are given.
• SEC exciters (white papers 2007-2011). A single-transistor oscillator (for example the SEC15-3: supply 15-20 V DC, MPSA06-type transistor, 400 pF, 10-22 µH coils) drives one end of a coil. The other end feeds an "Avramenko plug": two diodes that rectify the RF into a load (neon lamp, LEDs, an electrolysis cell). "AV Plug" is his name for that output node. Frequencies are about 10-13.6 MHz.
• Single-wire coil excitation (white paper Rel 1-2009). A signal generator at 10 MHz, 4.32 V peak to peak (p-p), drives a 9-turn primary over a 686 µH ferrite-rod coil; the scope shows 4.32 V p-p at the input and 26.5 V p-p at point C (a ratio of 6.1); the free end of the secondary shows "in excess of 130 volts", later "as high as 168 volts at the 12MHz input frequency", and lights a neon lamp through the tester's body capacitance.
• One-wire transmission guide. A 24 V transmitter, a 143-turn coil on a 12.5 mm Lucite rod with a "top hat" (two stainless condiment cups), one 12 ft wire, an identical receiver coil and top hat, 48 LEDs lit. No frequency, current or power is given.
• ExHeatExp (calorimetry of Rosemary Ainslie's MOSFET heater circuit, page undated and marked unfinished; probably c.2009, range 2008-2010, from its references to the "Ainslie Documentation" and its 555 setting, which matches the Ainslie group's 2009 specification [read: ExHeatExp deep read]; probable, not proven). A nichrome-wire coil wound on a Pyrex test tube holding 20-25 mL of water, in a foam enclosure, pulsed by a 555 timer and MOSFET from a 20 V supply. One screenshot (Fig 15) shows the result: CEC 1.319. Section 5 works through it.

3.2 Resonance explains the voltages

In the single-wire paper the drive is 4.32 V p-p and point C reads 26.5 V p-p, a ratio of 26.5/4.32 = 6.1 [calc]. The free end reaches 130-168 V, 30 to 39 times the drive (the p-p versus RMS basis is not stated, so this is loose). A coil near its self-resonance with a Q of a few tens does exactly this. Notice that the listed parts do not resonate at 10-12 MHz: 400 pF with 8.8 µH gives 2.68 MHz, with 14.98 µH 2.06 MHz, and 686 µH with 23 pF 1.27 MHz [calc]. The observed peak must therefore be set by stray and winding capacitance and the diodes, which were not measured. That is a gap in the model, not a contradiction.

High voltage at a high-impedance node can carry very little power. The generator's input power into its 50 Ω load resistor is about (4.32/2√2)²/50 = 47 mW [calc]. A neon lamp glows at well under a milliwatt and a white LED glows visibly at about 1 mA [memory, high]. "168 V / 2.7 V per LED = 62 LEDs" [read] is a statement that the voltage could forward-bias 62 LEDs, not that it could drive them at any useful current.

Stiffler's own kit manual gives a neat textbook illustration. The SEC 15-3 is tuned for "maximum neon glow and minimum supply current" [read: kit manual p.6]. That is the signature of a parallel resonant tank (L and C side by side across the drive): at resonance the inductor's and capacitor's currents cancel, so the tank's impedance is at its maximum, the voltage across it (and the neon) is highest, and the current drawn from the supply is lowest. As an illustration with assumed values (10 µH, Q = 50, resonance 13.6 MHz), the tank's impedance is about Q × 2πf₀L = 43 kΩ at resonance and about 4.5 kΩ when detuned by 10%, so the supply current rises about tenfold off tune [calc, example.py B9]. The tuning rule supports the ordinary-resonance reading; it does not by itself test the gain claim. His one-line "Spatial Gate" challenge, more than 300 V from an 18 V input "without a step-up transformer or multiplier", by "pinging the Energy Lattice" [web, Investigator: Energetic Forum thread 4067, post #1], is likewise a voltage statement, which a resonant coil can meet; it says nothing about energy.

The "amplification points" on the SEC Special Coil page (about 12.0, 15.1-15.2 and 18.6 MHz) are shown only as frequencies, in the text and in charts whose only axis is "Hertz"; no amplitude, gain or power is given [read: SEC Special Coil deep read, Figs 10, 11, 13]. Its unlabelled "Measured (MHz)" table (3.9866 / 10.4230 / 12.3340) is very probably the same data as the single-wire paper's Fig 18 "Energy Peaks" (3.9066 / 10.42302 / 12.334 MHz): the last two agree, the first differs in one digit, and the table's derived cells favour 3.9866 (an inference; both values stay on the record) [read: deep reads]. A replicator, Inquorate, reported that a white-noise bump at the coil's self-resonance appeared only after the coil had been pulsed [web, Investigator: thread 4067 p.145]; a single forum report with no measurement attached, consistent with a changed state of the coil's diode plug and also with an unexplained effect. The arbiter did not rule between them.

3.3 Displacement current explains the one wire

Two resonant coils with capacitive top hats joined by one wire are a classic Tesla arrangement. The loop is closed by displacement current through the top hats' capacitance to each other, to the room and to ground. A top hat of 2.5 cm radius has C ≈ 4πε₀r = 2.8 pF; 3 pF at 100 V RMS carries I = 2πfCV = 19 mA at 10 MHz and 26 mA at 13.6 MHz [calc]. Forty-eight LEDs at 2.7 V and 1-3 mA need 0.13-0.39 W, and a 24 V supply at an assumed 50 mA gives 1.2 W [calc]. So the demonstration closes its energy budget with ordinary physics; these are order-of-magnitude illustrations, not measurements.

This is real engineering, not fringe: Strebkov, Avramenko and Nekrasov published on single-wire resonant power systems, and patents such as US6104107 (Avramenko, Uniline Ltd.) describe the same arrangement [web: https://patents.google.com/patent/US6104107]. The Investigator found further patents using the Avramenko diode plug: Strebkov's RU2340064C1 family (2007-2013), Trubnikov RU2473160C2, Bank US9608441B2, Egely US20140126679A1, and RSC Energia RU2355074C1 [web, sites.md]. A patent grant shows the filing exists, not that the device performs as claimed. Stiffler's own observations fit an electric-field-coupled system: one sharp tuning point, a hand near the top hat detunes it, metal near the wire "degrades the operational efficiency", and the fields cause headaches and disturb nearby electronics [read: construction guide pp.8-10].

3.4 Phase decides RF power, and timing errors fake it

At 13.56 MHz one cycle is 73.7 ns. A 1 ns timing difference between the voltage probe and the current probe is a phase error of 360° × 13.56 × 10⁶ × 10⁻⁹ = 4.88° [calc]. In a resonant load where the true phase is 87° (power factor cos 87° = 0.052), a 5° skew makes the computed power factor cos 82° = 0.137, 2.7 times too large, or cos 92° = −0.033, negative [calc]. So both an apparent gain and an apparent "negative power" can come from timing alone. A replicator reported in 2019 that at certain frequencies "the power calculation is always negative" [web: Energetic Forum thread 9344, post #316, via the Investigator]; that observation is what this artefact produces. Stiffler's own input formula on the SEC Special Coil page is "Pin = Vin × Iin" from RF meter readings with no phase [read], which gives apparent power; his Fig 14 diagram of the method (Iin = Vd/Rg, Va = Vg − Vin) uses magnitudes only, so Pin is an upper bound on real power, and no numbers are given [read: SEC Special Coil deep read, Fig 14]. A smaller reading point of the same kind: in the Measurement Methods paper's Fig 19 the 13.6 MHz label is not on the tallest peak on the screen [read: deep read p.17].

3.5 Probes and ground detune high-impedance nodes

A resonant node with total capacitance C₀ that gains an extra ΔC shifts its frequency by the factor √(C₀/(C₀+ΔC)). With C₀ ≈ 6 pF (an estimate for the one-wire coil), a 10 pF probe pulls the frequency to 61% of its value; a 3 pF stray to 82%; on a 400 pF node a 10 pF probe moves it by only 1.2% [calc]. So the advice "use floating, battery-powered instruments with RF filtering and short leads" is sound RF practice. Stiffler's own papers concede that field pickup gives "meaningless relative readings" and that if moving the meter changes the reading, the reading is wrong [read: Measurement Methods].

3.6 Capacitor energy and stacking

Stiffler's worked example of "Energy Doubling" puts 135 V on a 20 µF capacitor (E = ½ × 20 µF × 135² = 0.18225 J) and then 270 V on two 20 µF in series (10 µF, E = 0.3645 J) [calc]. The page prints both as 1.8225 × 10⁻² and 3.645 × 10⁻² J, a factor of 10 too small; the ratio of 2 is unaffected. Both cases hold the same charge, 2.7 mC. Two capacitors charged to full voltage store twice the energy of one, but an ideal doubler draws twice the energy from its source to do it. The example has no input figure, so it says nothing about "excess".

3.7 Open systems and the loop test

A heat pump moves 3-4 J of heat for each joule of electricity because it draws on outdoor air, a known and measured source. No one calls it a perpetual machine [memory, high]. Stiffler's "door in the wall" analogy on the ECAT page makes this open-system point [read]. The textbook consequence: if a device really delivers electrical output P_out = COP × P_in (COP = coefficient of performance, output over input) from an external source, feeding a fraction η (an efficiency between 0 and 1) of that output back to the input sustains it whenever η × COP ≥ 1. For ECAT's stated COP of 2.194, η ≥ 0.456; on the lamp-level figures alone (12 W back to an 8.25 W input), η ≥ 0.69 [calc]. A rectifier and regulator routinely exceed that. Closing the loop removes every input-measurement question at once, because the circuit either keeps running or it does not.

3.8 Ambient energy is small

If the extra energy came from radio waves already in the room, how much is there? A 50 kW broadcast station radiating equally in all directions gives P/(4πr²) = 4.0 mW/m² at 1 km [calc], so 1 W would need about 250 m² of perfect collecting area. At ECAT's few kilohertz the ambient level is lower still [memory, medium]. Whatever the "Energy Lattice" is, it is not the ambient electromagnetic field as mainstream physics knows it.

---

4. Where the inventor departs

The exact point of departure. Stiffler's departure is not in Maxwell's equations or in circuit theory; his circuits are ordinary and his measurement advice is often good. It is in the step from "the circuit shows high voltage, bright LEDs and odd ringing" to "energy is being gathered from space", without a measurement of input and output energy that excludes error. He knew this was the gap. He wrote that early ECAT work "was empirical ... little substantiated science was available" [read: ECAT page], that "the math could not support the (conversion) concept" so "(Conversion) was replaced with (Coherence)" "to avoid explaining something that seemed not to be explainable at the time" [web: post #369, 2009-05-01], and he proposed the right tests himself: a thermal-bottle comparison of a board against a resistor at equal input power (described, no result stated [read: SEC Special Coil]) and the ExHeatExp calorimeter.

What is to his credit, fairly stated. He disclaimed over-unity in plain words; he built kits that strangers could build and that did what he said they would visibly do (Lidmotor's SEC and one-wire builds at 10 and 25 ft, 2009-2011; Raui, Inquorate, zeos71 on Energetic Forum [web, Investigator; videos not watched by the reviewers]); he reported his own negative result (an LED/black-plate test showed no temperature rise [read: SEC Special Coil]) and the cup runs in which CEC fell to about 0.83; and in ExHeatExp he described an apparatus, a calibration statement ("LM335's that were calibrated before each run to a NIST traceable standard"; NIST = National Institute of Standards and Technology), a 40 MHz sampling method and low-inductance shunts [read]. Many claims in this collection of projects never get that far.

The ECAT accounting, read fairly. The page says the 60 Hz transformer "wastes some 6 watts of the total input power of 8.25 watts", then that the lamp gets 12 W and "the total power produced ... is 18.11 watts". Read literally, 6 W lost from 8.25 W cannot leave 12 W for the lamp. Read as a whole, 18 W is the power delivered into the transformer, of which 6 W is lost and 12 W reaches the lamp. The arbiter ruled the second reading the likely intent: a wording slip, not a proven inconsistency (12 + 6 = 18, leaving 0.11 W unexplained) [calc]. What stays damaging on any reading: no output measurement method is given, and no figure on the page shows a current trace; the 12 W is the lamp's 120 V, 60 Hz nameplate rating, not a measurement of a lamp driven by kilohertz pulses with "high voltage arcing at the base of the bulb".

The input side is clearer than the first reviews thought. The 8.25 W is the bench-supply panel reading in a photo on the page: 25.0 V × 0.33 A = 8.25 W exactly (8.1-8.4 W at the display's resolution) [read: image ecat009; calc]. (The reviews and the first version of this primer paired 8.25 W with the page's "12 V DC supply" sentence, giving 0.69 A and, from the 90 mA "quiescent current", 1.08 W and a gain of 16.8. The arbiter withdrew that step: the 12 V belongs to the 4.3 kHz schematic design, and the photo shows 25 V for the lamp run.) Three questions remain open. First, the separate gate-drive generator V1 is apparently not in the 8.25 W; an estimate with assumed MOSFET values puts its direct share under 1%, so it cannot explain a gain of 2.2, but it is an unstated omission. Second, the sense resistor R3 has no value, so input current cannot be derived from its 0.31 V trace. Third, the 90 mA "during a load condition" fits a curious coincidence: at 25.0 V it is 2.25 W, and 8.25 − 2.25 = 6.00 W, exactly the "some 6 watts" transformer loss. One reading is that the 6 W is the difference between two supply draws (0.33 A and 0.09 A in two configurations), not a measured dissipation; the competing reading is that 90 mA belongs to the 12 V design and the match is chance. The page does not decide, and the arbiter records both. (If 2.25 W were the true input the ratio would be 8.05; that figure only sizes the ambiguity and is not a claim.) [calc, example.py B1]

Elsewhere the page holds a sentence no reviewer first noticed: "Under normal operation the MOSFET and its heat sink is cool (not cold like the transformer) but cool" [read: ECAT page]. If "the transformer" is the 60 Hz matching transformer, "cold" is hard to square with its wasting "some 6 watts" (a small laminated transformer dissipating several watts runs warm to hot [memory, medium]), which would favour the reading that the 6 W is not a measured loss. If it is the polyphase transformer there is no conflict, and the 60 Hz transformer's temperature is not stated. "Cold" and "cool" are by touch; the page gives no temperature. The arbiter keeps this as an unresolved tension in the inventor's report.

A forum viewer, Kokomoj0, estimated from an ECAT video "an addictional 4 watts roughly", and said a proper measurement needs DC conversion and a load [web: Energetic Forum thread 9344 post #3, quote verified by the Investigator]. It is one estimate from a video, by a viewer with no stake, not a measurement; it is the only third-party figure on record for ECAT. Stiffler also reports that ferrite or metal-core versions lost the gain until the saturating Theta design, and that Internet copies "failed" because builders changed the design (no attempt named) [read].

---

5. Worked example: does the one calorimetry run hold up?

The question: Stiffler's Fig 15 screenshot (ExHeatExp.htm) reports more heat in the water than electrical energy in. Reproduce it, then ask what would have to be wrong for the true ratio to be 1.00, and which control would tell. This is the best-documented experiment in the folder and the easiest to repeat, so it is the right one to learn from. I viewed the screenshot (cec1.32.gif) myself for this primer; every number below is legible in it [read: Fig 15 image]. Run example.py to reproduce every step.

Step 1. Reproduce the screenshot

Fig 15 shows: specific heat 4.187 J/(g·K); mass 25 g; start temperature 26.15 °C; end 41.25 °C (also shown as 314.4 K); "Sum(VIT)" = 1198.759 J; "Q = cmdt" = 1580.593 J; CEC 1.319; 600 s; read interval 1 s; supply 20.00 V (min 16.97), 0.096 A (min 0.092, max 0.108).

• ΔT = 41.25 − 26.15 = 15.10 K. (314.4 − 273.15 = 41.25, consistent.)
• Q = c m ΔT = 4.187 × 25 × 15.10 = 1580.6 J. Matches.
• CEC = Q / E_in = 1580.6 / 1198.759 = 1.3185. Matches 1.319.
• Mean input power = 1198.759 / 600 = 2.00 W. Cross-check: the displayed 20.00 V × 0.096 A = 1.92 W. Close; the display is instantaneous, the sum accumulated once a second.
• Mean heating power into the water = 1580.6 / 600 = 2.63 W.
• Apparent excess: 381.8 J, 31.9% of input [calc].

So the program's arithmetic is right. The question is whether its inputs are.

Step 2. Two corrections whose sign we know

Q counts only the water. The glass tube and the coil also warmed up, and some heat leaked out of the foam box. Both are heat the circuit made that Q does not count.

• Glass: with an assumed 8 g tube and specific heat 0.84 J/(g·K) [memory, medium; assumed values, not on the page], Q_glass = 8 × 0.84 × 15.10 = 101.5 J. CEC becomes 1.40.
• Loss: with an assumed loss of 0.01 W per kelvin of temperature above ambient (a guess for a foam box), and a mean rise of about 15.10/2 = 7.55 K (the screenshot's chart lines are close to straight), loss ≈ 0.01 × 7.55 × 600 = 45 J. CEC becomes 1.44.

Both corrections push the ratio up. This is a key point the reviewers agreed on: "he forgot the losses" cannot explain the excess away. If there is an error, it must be in the temperature reading, the mass, or the input.

Step 3. Break-even: how large must each error be?

Set CEC = c m ΔT / E_in = 1 and solve for one variable at a time [calc, solved symbolically in example.py]:

The page text says "20-25 mL" while the program used 25 g. At 20 g the ratio is 1.055 [calc]. The ratio is proportional to the mass typed into the program, and no weighing is reported.

Step 4. What a 3.65 K error means for the sensor

The LM335 sensor outputs 10 mV per kelvin of absolute temperature [memory, high]: 2.993 V at the start (299.30 K) and 3.144 V at the end (314.40 K). A 3.65 K error is 36.5 mV, about 1.2% of the reading. Three kinds of error behave differently:

• (a) A constant calibration offset cancels in ΔT (it is in both readings). No effect.
• (b) A slope (gain) error would have to be 31.9%. A sensor calibrated against a traceable standard, as Stiffler says his were, is far better than that.
• (c) A voltage picked up only while the circuit runs, for instance rectified RF from the 1.49 MHz ringing and 149 V spikes Stiffler recorded on the same setup (Fig 14), adds in full: 36.5 mV would do it. Stiffler fitted a toroid filter on the sensor leads (Fig 18), which suggests he was aware of the risk (the purpose is an inference; the page does not say). Whether the sensor was checked with the circuit running is not stated.

Step 5. What a temperature gradient would have to look like

The page says no stirrer was used and the sensor sat "at the top in the last ~5mL", with "a slight shaking of the tube at the end of a test run" to mix [read]. Warm water rises. As a toy model, let temperature rise linearly with height from the bottom to the top of the water by an amount ΔT_v. The sensor in the top fifth reads the mean of that slice, Tb + 0.9 ΔT_v; the whole column averages Tb + 0.5 ΔT_v; the difference is 0.4 ΔT_v [calc]. For it to equal 3.65 K, the top would have to be about 9.1 K warmer than the bottom. Whether that happened, and whether the tube was shaken before the 41.25 °C reading, is not recorded. This is a size estimate for what a stirred run would have to rule out, not a finding.

Step 6. The returned (flyback) energy

People argued that the coil's collapsing field sends energy back to the battery and so hides part of the input. Stiffler's reply rests on his Fig 16 schematic: diode D1 sits across the coil (the same topology as Ainslie's own circuit, whose 2009 reprint says "A flyback diode, (D1) is placed in parallel with (RL)", a 1N4007 [read: Ainslie reprint p.5, via the ExHeatExp deep read]; it is not Stiffler's modification), so at turn-off the coil current circulates coil → D1 → coil and never reaches the battery; "the field collapse supplies a current that does indeed heat the coil (RL) and the diode (D1), but does not return energy to the battery" [read]. The circuit reasoning is correct for the schematic as drawn, and in Fig 17 D1 is epoxied to the tube, inside the calorimeter. How big is it? Fig 11 shows about 3.8% duty at 2.38 kHz [read]; with a mean current of 0.096 A the peak is about 0.096/0.038 = 2.53 A (consistent with 20 V across about 8 Ω). Stiffler gives no inductance for his coil, but Ainslie documents hers: L = 8.64 µH, R = 10 Ω [read: Ainslie reprint p.5]. Because the coil's time constant L/R (about 1 µs) is much shorter than the 16 µs on-time, the current reaches V/R = 2.0-2.5 A (R = 10 or 8 Ω) each pulse, and ½LI² is 17-27 µJ per cycle, or 41-64 mW, about 2-3% of the 2 W input (2-4% across the deep reader's cases) [calc, example.py Step 6]. A Wheeler-formula estimate for a 40-turn coil on a test tube (tube size assumed) gives about 5 µH, a similar figure; the first reviews' wider range of 2-19% assumed up to 50 µH, ten times what such a coil gives. Either way that energy becomes heat inside the cell, so it is part of Q, not additional to it. It cannot make the 31.9% excess. (Stiffler writes "I have confirmed" this but shows no waveform or integral, so the arbiter rates it Plausible, untested as a measured statement; the 2-4% is a calculation, not a measurement.)

The heater circuit itself has its own review. Stiffler's 1.49 MHz ringing is the same kind of MOSFET parasitic oscillation that the Ainslie review identifies as the source of her apparent gain (her documented states are 143-200 kHz in 2002 and "random oscillation" in 2009; 1.49 MHz is 7.5-10 times the 2002 figure, and "orders of magnitude higher than described" is true only against the 2.4 kHz setting) [read: ExHeatExp deep read; ../Ainslie-Rosemary-Ann/claims.md item 3]. A 1.49 MHz ring with L = 8.64 µH needs about 1.3 nF, mostly MOSFET output capacitance and wiring: ordinary tank behaviour [calc]. See review/Ainslie-Rosemary-Ann/primer.md for that circuit's theory and record.

One more input question stays open: the screenshot accumulates "Sum(VIT)" once a second from readings labelled "PSP Charge Supply", while the page's footnote says AC energy readings are "taken by signal integration at a 40MHz-sampling rate" [read]. Which method fed Fig 15 is not stated. Sampling a pulsed, ringing supply once per second can be wrong in either direction.

Step 7. The control that decides

Run the same tube, coil, water mass and sensor on steady DC at 2.00 W for 600 s, with the mass weighed and the water stirred. Then:

• If the DC blank also reads about 1.32, the excess is a calorimeter artefact (gradient, mass, sensor slope): the apparatus over-reads whatever heats it.
• If the DC blank reads about 1.0, the pulsed-run excess comes from something specific to pulsing: RF pickup on the sensor, the input sampling method, or a real effect. A shielded or non-electronic second thermometer and a fast-digitizer input integral separate those.

Stiffler's page recommends calibration ("if properly calibrated") but reports no blank, no repeat and no error bars, and the page ends unfinished after "A second inductor is added." [read]. In the cup enclosure, he reports that CEC could not be kept above 1 once the water reached 42 °C and then "leveled at around 83%", which he attributes to heat loss [read]. That is the result: one run, ratio 1.32, never repeated with a control. Not refuted, not supported; Never properly tested.

What the example teaches. An energy ratio is only as good as its least-controlled term. Here the arithmetic is perfect and the corrections we can name make the anomaly larger, which is exactly when a careful experimenter reaches for the blank run, because the remaining suspects (sensor pickup, a gradient, the mass, the input sampling) are each large enough on their own and none was excluded.

---

6. Evidence table

---

7. What a learner should take away

What is solid. - Resonant coils give large voltages from small drives; a few tens of volts to a neon lamp or an LED string needs only milliwatts. Stiffler's step-up ratios (6.1 at point C, 30-39 at the free end) are ordinary. - Single-wire transmission works, with the return path closed by displacement current through the terminal capacitances. It has a real engineering and patent literature (Tesla, Avramenko, Strebkov). - His instrument advice is mostly good RF practice: high-impedance RF nodes are loaded and detuned by probes and grounded meters, and V × I without phase is not power. - His own statements narrow his claim: no creation of energy, a CEC above 1 is a "conversion" in an open system, and looping is not expected.

What is wrong, and why the mistakes are instructive. - Voltage and brightness are not power. "168 V, so 62 LEDs" and "the LED glows intensely" say nothing about energy until current and phase are measured. This is the single most common error in coil-based energy claims. - An undefined figure of merit. CEC was defined (with joule-seconds, the unit of action, in place of joules) and never evaluated in the white papers; Eq 2 has an undefined symbol n. A figure of merit without a measurement protocol cannot carry a claim. - Arithmetic slips that do not change the story but show the checking was light: capacitor energies printed 10 times too small; impedance tables computed as √(R + |X|) instead of √(R² + X²) (minima at the right frequencies, every value wrong); Eq 9 with π where the wavelength λ belongs (the standard near-field boundary is 0.62√(D³/λ) [web: https://www.itecspec.com/3gpp/38.810/s/7.2.1.2]); "0.001 V across 1 Ω = 0.001 mA" (it is 1 mA). - The open-system defence needs a named source. A heat pump has COP 3-4 and nobody calls it perpetual because outdoor air is a measured source. ECAT's source is never named, and for an 18 W electrical output the loop test is cheap; the page does not argue why it could not be done. - An unusual result calls for a blank, not for an explanation. ExHeatExp is the instructive case: the right apparatus, a clean ratio, and no DC control. It tested a circuit whose apparent gain has its own record of critique and claimed replication; read it in review/Ainslie-Rosemary-Ann/ (primer and arbiter), where the parasitic-oscillation explanation and the 2.4 kHz / 3.7% specification Stiffler matched are set out.

What is genuinely open. - Does the ExHeatExp cell read about 1.0 or about 1.3 on a DC blank? No one on record has run it. - Does an SEC board on a battery in a thermal bottle heat the bottle more than a resistor at the same DC input? Stiffler described exactly this test and stated no result. - What did ECAT actually draw and deliver with the intact design (coils, Theta core and parts are not given in text)? Which set-up produced the 8.25 W and 18.11 W: the photos show the lamp version at about 1.1 kHz and 25 V, not the 4.3 kHz, 12 V design the text describes. Is the "6 watts" a measured loss or the difference between two supply draws, and was the 60 Hz transformer the one he called "cold"? - How do Stiffler's "Self Charging" NILS (Near Infinity Light System) units, mentioned as never released in the Panacea course [web: https://www.scribd.com/doc/35609446], square with his 2011 post that looping output to input "would require that we were manufacturing energy" [web: thread 4067 p.142]? Both come from the inventor or his promoter (an interested party); neither explains the other. - How efficient is resonant single-wire transfer when measured properly? Olkhovskiy, Muellerova and Martinek (2019, IEEE) address how to measure it [web: https://dspace.zcu.cz/items/ecbbaecc-629a-4d7d-8f19-0a21b92e631e/full]; no measured efficiency for Stiffler's one-wire guide exists. - Is the "Negative Resistance" claim in the 2010 Panacea course a small-signal statement (any active oscillator is a negative resistance in that sense [memory, high]) or a claim of net outward real power? It needs measured V, I and phase with matched probes. - What did Stiffler's 2017-2018 "Stiffler Loop" videos show, including the one titled "Is the Loop really doing anything?" (163 videos on @StifflerDr, unreviewed by anyone in this project) [web, Investigator]?

Where the record goes quiet (absence of evidence). The arbiter set out six places the record is thin; each is an anomaly to explain, not a neutral gap: - No energy-balance paper after 2010. ZPEnergy (2010-10-14) relayed a plan to publish a paper from replicators' data; none has been found, though white papers continued to at least 2011 (Series 18 documentation, Rel 2.8, September 2011; not read) and videos and circuits to 2018 (a replicator's 2018-04-30 post of a modified Stiffler circuit) [web, Investigator]. That leaves about eight years without an energy-balance paper; his death in late 2018 explains the end of the record, not those years. Competing readings: no result survived careful measurement (his own "Is the Loop really doing anything?" title, the "math could not support" remark and the negative black-plate test point this way, though he kept publishing and selling); data kept for licensing ("All licenses are available from Stiffler Scientific"; against: most circuit detail was posted openly); health or other priorities (no evidence either way). - No replicator calorimetry in about nine years, though the circuits are easy to build and the thermal-bottle test is easy to run. - The thermal-bottle result is not stated on the page that describes the test, and the page's photographs (now viewed) show the apparatus but no reading; it was not run, was null, or the page was never finished. The record cannot tell. - Patent. "Patent Pending 61254675" in the Panacea course (seen in Scribd document 35609446; the 2012 Panacea PDF was not read, so whether it repeats the words is unknown) has the form of a US provisional, probably filed in late October 2009 (inferred from a neighbouring provisional, 61/254,559, filed 2009-10-23 [web, arbiter's check: https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/8552592]; inference from neighbours, not confirmed). No published application by Stiffler was found. A lapsed provisional is common for individual inventors; USPTO Patent Center would decide. - Disappearing material. The demo video was removed; stifflerscientific.com and drstiffler.com are dead; Google removed some results after two legal requests (Lumen notices 75850580 and 95550170, requester unknown) [web, Investigator]. Lapsed hosting after his death fits the timeline, and his YouTube channel stays online, which cuts against active takedown; an unread notice is not evidence either way. - "Advanced Technology Group" (1993-96) and Stiffler Scientific have no corporate record found; possibly an informal team or an assumed-name business (Harris County index not searched).

A Ronald Stiffler, 75, died on 2018-11-12 (obituary notice [web: https://obits.postandcourier.com/us/obituaries/observergroup/name/ronald-stiffler-obituary?id=33303171]); a newsletter of 2019-01-24 says Dr. Stiffler "passed away last November" [web, Investigator]. That the obituary subject is the inventor is an inference from name, place and date. His education and the source of the "Dr." title are not established in the review files.

---

8. Check yourself

1. A coil driven at 4.32 V p-p shows 26.5 V p-p at one node. Is that a gain of 6.1 in power? What would you need to measure to find the power ratio?
2. In ExHeatExp, why can't omitted heat in the glass and heat lost through the foam explain the 1.32 ratio? Which three quantities are left as suspects, and how large must each error be?
3. A voltage probe and a current probe differ in delay by 1 ns at 13.56 MHz. The true phase between V and I is 87°. Estimate the range of computed power factors, and say why this matters for a claimed gain.
4. ECAT is said to deliver 18.11 W for 8.25 W input. What fraction of the output would a feedback converter need to return to the input for the circuit to run with no supply? Why is that experiment decisive where a meter reading is not?
5. Stiffler says a grounded scope detunes his AV Plug. Give a textbook explanation with a number, and describe a test of his exciter's energy balance that needs no probe on the RF circuit at all.

---

9. Further reading

Mainstream sources. - Griffiths, Introduction to Electrodynamics: chapter 7 (Faraday's law, inductance, displacement current) and chapter 8 (Poynting's theorem, energy conservation for fields). - Horowitz and Hill, The Art of Electronics: chapter 1 (AC circuits, phase, power) and the sections on rectifiers, voltage multipliers and probing. - Schroeder, An Introduction to Thermal Physics: the first law and heat capacity; heat pumps and their coefficient of performance. - Balanis, Antenna Theory: field regions (reactive near field, radiating near field, far field). - On calorimetry practice and its pitfalls, the debate over cold-fusion calorimetry (for example the papers collected around the 1989 Fleischmann-Pons experiments) is a good case study in why blanks, stirring and calibration decide arguments. - Single-wire power: D. S. Strebkov, S. V. Avramenko, A. I. Nekrasov, "Single-wire electric power system for renewable-based electric grid" (2000) [web, cited by the physicist; full text not read]; patent US6104107 [web: https://patents.google.com/patent/US6104107]. - On measuring single-wire power: Olkhovskiy, Muellerova and Martinek, a 2019 IEEE paper on measuring the efficiency of single-wire transmission, found by the contrarian (the repository page gave no efficiency numbers) [web: https://dspace.zcu.cz/items/ecbbaecc-629a-4d7d-8f19-0a21b92e631e/full; full text not read]. - The companion primer review/Ainslie-Rosemary-Ann/primer.md for the circuit theory of the heater Stiffler tested, and that project's arbiter.md for the record of critique and replication claims.

The inventor's own key works (in the folder unless noted). - "Energy Conversion by Articulated Transfer" (ECAT page, revised 2007). - "Alternative Energy Research and Spatial Energy Coherence with Single Wire Excitation", Rel 1-2009. - "SEC Exciter Measurement Methods" (white paper). - "Wireless or One-Wire Energy Transmission" construction guide; "Assemble Your SEC1520 Circuit Board Kit". - SECExciter.htm (exciter family and the "Energy Doubling" example); SEC Special Coil.htm (impedance tables, thermal-bottle and black-plate tests); ExHeatExp.htm (the calorimetry run); coils.htm. - Not in the folder [web, Investigator]: "An Efficient Method for Driving LED Arrays from Spatial Energy Coherence Exciters" (2010) https://www.scribd.com/document/27399091; the Panacea-BOCAF course "Spatial Energy Coherence" (88 pp.) https://www.scribd.com/doc/35609446; "Series 18 Exciters Documentation" Rel 2.8 (2011, studylib); Energetic Forum thread 4067 "High voltage from thin air" https://www.energeticforum.com/forum/energetic-forum-discussion/renewable-energy/4067-high-voltage-from-thin-air; YouTube channel @StifflerDr.

---

Answers

1. No. Voltage magnification at a resonant node is expected from a coil with a Q of a few tens; the current at that node is small and its phase relative to the voltage is near 90°, so the power there can be tiny. You need the real power at both ends: the time average of v(t)i(t) at the input and at the load (with matched probe delays), or the load's total heat in a calorimeter.

2. Both are heat the circuit produced that Q does not count, so adding them increases the true heat and raises the ratio (to about 1.40 with an 8 g tube, 1.44 with a modest loss estimate). The suspects left are the temperature rise (it must be over-read by 3.65 K, for example by RF pickup appearing only while the circuit runs, or by a gradient in the unstirred tube), the mass (18.96 g instead of 25 g; the page says 20-25 mL) and the input (under-read by 24.2%, for example by once-a-second sampling of a pulsed supply).

3. 1 ns is 4.88°, so the computed phase could be anywhere from about 82° to 92°: power factors from +0.137 (2.7 times the true 0.052) to −0.033 (negative). A timing error alone can produce an apparent gain or an apparent "negative power", so any RF V × I claim needs a probe-skew check (well under 0.1 ns at 13.56 MHz) and a known dummy load first.

4. η × COP ≥ 1 with COP = 18.11/8.25 = 2.194, so η ≥ 0.456 (or 0.69 if only the 12 W lamp power is fed back). It is decisive because it removes the input measurement altogether: a circuit with no external supply either keeps running or stops, whatever the meters say.

5. A grounded probe adds roughly 10-15 pF to a node; with about 6 pF already there, 10 pF lowers the resonant frequency to √(6/16) = 61% of its value, so a sharply tuned high-impedance circuit is knocked off resonance. The probe-free test is the one Stiffler proposed: run the board from a battery inside an insulated thermal bottle, measure only the DC input (which needs no RF probe), and compare the bottle's temperature rise with a resistor drawing the same DC power, until both reach steady state.