Follow-up Q16 (Morray-TH): the multi-tank cascade, the one-way valve, and whether the first stage "feels" the load

Date: 2026-10-08. Role: STEP F follow-up investigator. Web time: about 3 minutes (timer 177 s; Chrome, plus curl of pages found in Chrome). Stopped early: the question was answered from Moray's own text plus four standard references.

The question (owner's words, lightly polished)

Moray's multi-tank cascade (Sea of Energy 4th ed. p.116): oscillations in the first tank, reinforced by "harmonic coupling", "spill" over through a valve that blocks feedback into the next, lower-frequency stage. Is a gap that conducts only while energy passes, isolating the tanks otherwise, the key design point? How does it compare with Wien's quenched spark gap (1906) and impulse-excited tank chains? The tanks stay in phase, so when the load draws current the phase shifts slightly and the earlier stage should "feel" the resistance at the next impulse: is that right (injection locking, Adler's equation)? Does a collector have to feel the load to absorb at all (a matched load; an unloaded resonant antenna re-radiates)? And "the draft of the current is adapted to the load". [Moray p.139: the valve passes waves "in only one direction at a time ... without necessarily changing from H.F. or A.C. to D.C.": not a rectifier. Include a bench test: diode vs gas-tube threshold vs ferrite isolator as the valve.]

What was searched

Saved pages (evidence trail, not published): research/pages/wiki-spark-gap-transmitter.txt, nbs-circular-74.pdf/.txt, gustafsson-forward-scattering.pdf, razavi-2004-injection-locking.pdf, razavi-2004.txt, wiki-injection-locking.txt, wiki-pearson-anson.txt, wiki-isolator.txt, kelly-practical-guide-ch7.html.

What Moray wrote (the project's own text)

1. The cascade [read: Sea of Energy 4th ed. p.116, printed page 107]: "Oscillations are started in the first stage or circuit of the device by exciting it with an external energy source. The circuit is 'tuned' until the oscillations are sustained by harmonic coupling to the cosmic wave frequencies. The reinforcing action of the harmonic coupling increases the amplitude of the oscillations until the peak pulses 'spill' over into the next stage through a special detector or valve which prevents the return or feed-back of energy from succeeding circuits. These 'pulses' drive this stage which oscillates at a lower frequency and is again reinforced by harmonic coupling ..." Same page: "The oscillations are sustained as long as it remains properly tuned and the external circuit is completed through a suitable load." And: "The special tubes which appear to be the key to the success of this device are Ionic cold cathode tubes".
2. How the valve works, in Moray's own words: a threshold gas discharge, stage by stage [read: 4th ed. pp.55-56, printed pp.46-47; the same text in 5th ed. p.214]: the tubes "charge the accompanying capacitors or condensers at a more steady rate which, when a certain voltage is reached, ionization occurs in the gases of the discharged tube and causes the condensers of the valve circuit to discharge into other condensers of the oscillators ... When ionization in the preceding tubes is no longer possible because of the reduced voltage, the process starts all over again. The first valve passes vibration of energy into an oscillatory circuit, ionization sets in, a discharge occurs and energy passes through another valve into other oscillators. The process is repeated from the first stage on to the second on to the third and so on, much like a bucket bri[ga]de." He sums it up as energy "caus[ed] to oscillate through the secondary circuits through a blocked circuit of no return." The 5th edition prints "bucket brigade"; the 4th-edition scan reads "bucket bridge" (an OCR or print error).
3. Not a rectifier, but also used to rectify [read: 4th ed. p.139, printed p.131]: the stone "may well serve as a valve which allows energy waves to pass in only one direction at a time, that is to say, from the antenna capacitor into the various circuits or from the circuits into the antenna capacitor without necessarily changing from H.F. or A.C. to D.C." The same paragraph goes on to say the assembly, with the Figure 10 hookup, is "used for the rectification and amplification of radio or other high frequency signals". So Moray describes one device used two ways: as a one-way coupler between oscillating stages (the cascade) and, in a receiver hookup, as a rectifier. Kelly's ch.7 caption likewise says the valve "was capable of being set to either rectify a signal or to amplify a signal" [web: frienergi.alternativkanalen.com/Chapt7.html]; this is Kelly's reading of the patent-application notes, not a primary source.
4. "The draft of current is adapted to the load" [read: Radiant Energy pamphlet (folder file "1926") p.19: "It makes no difference whether one fifty-watt lamp is used or whether 50 are connected to the machine, the draft of current is adapted to the load"; Valley State College lecture 1962 p.9: the same sentence with "a 300 KW load"; Radiant Energy p.13: frequencies obtainable "by transposition just as the draft of the current with the R.E. device is adapted to the load"]. The 1962 lecture also says capture "obeys the laws of the conservation of energy" [read: Valley State College 1962 p.3]; the Sea of Energy passage above says "There is only a lowering of potential like water over the wheel" [read: 5th ed. p.214].

What the outside sources establish

5. Wien's quenched gap [web: Wikipedia, Spark-gap transmitter, section "Quenched-spark transmitters", citing Leggett 1921 pp.55-63]: in a plain coupled spark transmitter the energy sloshes back and forth between primary and antenna circuits ("Each time the energy returned to the primary, some was lost as heat in the spark") and two frequencies are radiated. A quenched gap (a stack of plates about 0.1-0.3 mm apart) "extinguished ('quenched') the spark at the first nodal point ... when the primary current momentarily went to zero after all the energy had been transferred to the secondary winding. Since without the spark no current could flow in the primary circuit, this effectively uncoupled the secondary from the primary circuit, allowing the secondary resonant circuit and antenna to oscillate completely free of the primary circuit after that (until the next spark)." Result: one frequency, decrements of only 0.08 to 0.25. NBS Circular 74 [web: nvlpubs.nist.gov/nistpubs/Legacy/circ/nbscircular74.pdf, about pp.230-233]: "It was found by M. Wien that if a series of short spark gaps be substituted for a single long gap ... the discharge path ... returns much more quickly after discharge to its initial condition of high resistance ... called the quenching action"; and, unlike an ordinary gap, the quenched gap "requires a fairly close coupling between the primary and secondary circuits". The same circular, p.93: "Free oscillations are obtained in the case of pure impulse excitation" (German Stossanregung).
6. Threshold gas tubes as one-way timed switches [web: Wikipedia, Pearson-Anson effect, citing Pearson and Anson, Proc. Phys. Soc. London 34, 1921-22]: a neon tube does not conduct until its striking voltage; then its resistance drops and it conducts "until the applied voltage drops below the extinction voltage". This is the behaviour Moray describes in item 2, and it was published as the neon relaxation oscillator in 1922.
7. Injection locking and Adler's equation [web: Razavi 2004, p.1417, read as page image]: for a tank of quality factor Q, natural frequency w0, oscillation current I_osc, injected current I_inj at frequency w_inj, the steady phase difference is sin(theta) = (2Q/w0)(I_osc/I_inj)(w0 - w_inj) (eq. 8), and the one-sided lock range is w_L = (w0/2Q)(I_inj/I_osc) (eq. 9). The phase reaches 90 degrees at the edge of the lock range. Razavi Fig. 5 analyses an injection-locked divide-by-2 stage, an oscillator locked by a drive near twice its own frequency. Adler, Proc. IRE 34 (1946) 351-357, is the original [web: Wikipedia, Injection locking, reference list].
8. A receiving antenna and its load [web: Gustafsson et al. 2012, lucris.lub.lu.se/ws/files/6090926/3031014.pdf]: "the scattering of small, single mode, and matched antennas scatter and absorb equal amounts of power" (Green's minimum-scattering result), and "the absorption efficiency cannot be greater than 1/2 for reciprocal linearly polarized lossless matched antennas with a symmetric radiation pattern." They measured loaded, short-circuit and open-circuit dipoles: loaded and short-circuit dipoles have equal integrated extinction, the open-circuit one less. With no load (or a short), a lossless antenna absorbs nothing: everything it intercepts is re-radiated.
9. Ferrite isolators [web: Wikipedia, Isolator (microwave)]: the common type is a circulator whose third port "is connected to a matched load. All the incoming signal is then absorbed". The reflected power is not stopped but dumped as heat in that load. Commercial ferrite circulators and isolators are offered from about 20-25 MHz upward (Google snippets: IMPO-RF "20 MHz-20 GHz"; Raditek coaxial "25MHz to 50GHz"; Ferrite Microwave Technologies "50 MHz to 50 GHz"); nothing found at the kHz-to-low-MHz frequencies of a lamp-lighting tank chain.
10. No independent engineering analysis of Moray's cascade was found; the free-energy pages found (Kelly ch.7; an Energetic Forum thread) repeat p.116 and the p.55 "bucket brigade" passage.

Reasoning

Is the gap that conducts only while energy passes the key design point?

Moray's own text says so twice. On p.116 the valve "prevents the return or feed-back of energy from succeeding circuits"; on pp.55-56 the mechanism is a gas tube that ionizes "when a certain voltage is reached", passes the charge on, and stops "when ionization ... is no longer possible because of the reduced voltage". That is a threshold switch that conducts only during the transfer and isolates the stages the rest of the time. It is the same idea as Wien's quenched gap (item 5): conduct during the transfer, then open so the next circuit rings freely and nothing flows back. It is also the bucket-brigade or charge-pump pattern of the neon relaxation oscillator (item 6). So the question's reading fits the source text, and the mechanism is standard. Whether the cosmic "harmonic coupling" supplies the energy is a separate question (claims 3 and 4) that this answer does not decide.

Two differences from Wien matter. - How the energy is transferred. Wien's gap moved energy through a closely coupled resonant transformer and quenched at a current zero, so little was lost. If a gap dumps one capacitor straight into another, about half the energy goes as heat (the two-capacitor problem). The worked example below shows 47-62 % lost in the gap. Moray's p.55 says the condensers discharge "into other condensers of the oscillators". Read literally, that is the lossy direct dump unless the "oscillators" include inductance that makes the transfer resonant. - Frequency. Wien's secondary rang at the same frequency as the primary. Moray's next stage "oscillates at a lower frequency". A tank fed one impulse every n cycles of the earlier stage rings at its own frequency, so stepping down works like an impulse-excited or injection-locked divider (Razavi Fig. 5, divide-by-2). Each stage's frequency must sit near a whole-number fraction of the stage before.

Does the earlier stage "feel" the load? Phase or amplitude?

Partly right, with one correction. - Through a threshold gap or diode: yes, at each impulse. How much charge passes depends on the voltage across the valve at the moment it fires, which is the earlier tank's voltage minus the next tank's voltage minus the valve drop. A heavier load drains the next tank, so its voltage is lower at the next firing, more charge passes, and the earlier stage gives up more energy. In the worked example it gives 3.2 mJ instead of 2.55 mJ. The earlier stage feels the load without any energy flowing backward. A diode charging a filter capacitor works the same way. This fits Moray's "draft of current is adapted to the load" and the p.116 condition that the load circuit must be completed: power is drawn as the load demands. That is ordinary for any source with low internal impedance; it does not show where the energy comes from. - Phase: only if a stage is detuned. By Adler's equation (item 7), the locked phase is sin(theta) = (2Q/w0)(I_osc/I_inj)(w0 - w_inj). With zero detuning the phase is zero whatever the load. A resistive load lowers Q and I_osc, so it changes the phase only when the stage runs off its natural frequency. In that case loading pulls the phase toward zero (example below). Mainly the load changes the amplitude, and so the timing and size of the next impulse. A reactive load would also retune the tank and shift its phase. - Through an ideal ferrite isolator: no. The earlier stage always sees the isolator's matched input. Whatever the load does not take is reflected and burned in the isolator's third-port resistor (item 9). The draft is not adapted to the load. Power not used is wasted rather than left in the earlier tank. This is the clearest difference between "one-way" in the isolator sense (linear and non-reciprocal) and Moray's "one direction at a time" (a timed threshold switch).

Must a collector feel the load to absorb at all?

Yes, by standard antenna theory (item 8). A lossless open-circuit or short-circuit antenna absorbs nothing and re-radiates all it intercepts. A matched load absorbs at most as much as is re-radiated (absorption efficiency at most 1/2 for the common case). For a high-Q tank the same holds over time: an unloaded lossless resonator driven by a field builds up until it re-radiates as fast as it receives. Only a load (or losses) takes energy out. So "the oscillations are sustained as long as ... the external circuit is completed through a suitable load" (p.116) fits receiver physics as written: without a load the collector would ring and re-radiate, not store energy without limit. The cascade's job is then to pass the collector's energy to the load as fast as it arrives, with the valve stopping it from sloshing back to the antenna (where it would be re-radiated). That is the same job Wien's quench did for a transmitter, in reverse.

Competing readings

• (a) Threshold-gap reading (Moray pp.55-56, the question's reading): a gas or cold-cathode gap that fires at a set voltage, like Wien's quench or a relaxation oscillator. The earlier stage feels the load through the conduction at each firing.
• (b) Diode or rectifier reading (p.139 second half; Kelly; the arbiter's "rectifier-type detector"): a contact rectifier passes current one way all the time. Between stages this would bias the next tank and add a direct-current part, which Moray's "without necessarily changing ... to D.C." rules out for the cascade use. It fits his receiver hookup.
• (c) Non-reciprocal reading (ferrite-isolator-like): no evidence Moray used magnetized ferrites, and isolators do not work at the cascade's likely frequencies (item 9). It is useful only as a bench control.

Worked example (numbers chosen for illustration) [calc]

Gap transfer between equal capacitors. C1 = C2 = 10 nF. Stage 1 is at 1000 V when the gap fires. The gap's maintaining voltage (the drop while it conducts) is Vm = 100 V. Charge flows until V1 - v2 = Vm, so each capacitor changes by dV = (V1 - v2 - Vm)/2. - Light load, next tank at v2 = 300 V: dV = 300 V, charge 3.0 uC; stage 1 gives 1/2 C (1000^2 - 700^2) = 2.55 mJ; stage 2 gains 1/2 C (600^2 - 300^2) = 1.35 mJ; gap heat 1.20 mJ (53 % transfer efficiency). - Heavy load, v2 = 100 V: dV = 400 V, charge 4.0 uC; stage 1 gives 3.20 mJ; stage 2 gains 1.20 mJ; gap heat 2.00 mJ (37.5 %). So the earlier stage gives more per impulse when the next stage is loaded: it feels the load. A direct capacitor-to-capacitor dump wastes about half. A resonant (inductive) transfer quenched at the current zero, as in Wien's gap, avoids most of that loss.

Adler phase. w0/2pi = 100 kHz, Q = 50, I_inj/I_osc = 0.1. One-sided lock range = (100 kHz / 100) x 0.1 = 100 Hz. With 20 Hz detuning, sin(theta) = (2 x 50 / 100,000 Hz) x 20 Hz / 0.1 = 0.2, so theta = 11.5 degrees. If loading drops Q to 40 (same current ratio), theta = 9.2 degrees. At zero detuning, theta = 0 for any load.

Bench test (diode vs gas-tube threshold vs ferrite isolator as the valve)

Set-up: two parallel LC tanks. Tank A, for example f_A = 200 kHz, is kept ringing by a signal generator through a weak coupling (standing in for the "harmonic coupling"). Tank B is at f_B = f_A/2 = 100 kHz, with a switchable resistive load R_L (open; 10 x and 1 x the tank's dynamic resistance). Measure, with tank A's drive fixed: (1) the power drawn from the generator (true power, by V and I in phase), (2) the power in R_L, (3) the voltage on tank A, (4) the phase of tank B against the generator's divided-down reference, and (5) signal leaking back from B to A, with B driven by a second generator at a slightly different frequency. - Valve 1, fast diode (e.g. 1N4148, or a germanium point contact to be closer to Moray's stone): expect conduction every cycle above about 0.3-0.7 V, a direct-current offset on tank B, and little stepping down unless B is set to f_A/2. Back-leakage through the diode's capacitance. - Valve 2, threshold gas tube (a small neon or gas-discharge-tube surge arrester, striking about 70-90 V, maintaining about 50-60 V): run tank A high enough to strike. Neon takes roughly tens of microseconds to deionize [memory: medium], so lower both tanks to about 10-20 kHz if the tube fails to quench between cycles. Expect firing once per n cycles, B ringing freely between firings, isolation between firings set by the tube's picofarad capacitance, and generator power that rises with load (the draft adapted to the load). Plot energy per firing against R_L and compare with the worked example. - Valve 3, ferrite isolator: impractical at 100-200 kHz (commercial parts start near 20-25 MHz). Either rescale the whole test to about 30-50 MHz with a coaxial isolator, or use an active isolator (a buffer amplifier) as the linear one-way control. Expect generator power independent of R_L, with unused power heating the isolator's termination. - Deciding observation: with the gap valve, does the power taken from the drive follow the load (it should), and does any configuration deliver more power to R_L than is drawn from the generator, inside the measurement error? The second is the question claims 3 and 4 turn on; no such test is on record.

Open

• No schematic of the cascade survives in readable form: the p.115 drawing ("See schematic drawing of R E device") is garbled in OCR. A page-image read of 4th ed. p.115 (and the 5th-edition equivalent) would show how many stages, what frequencies, and whether the stages are joined by coils (resonant transfer) or capacitors (lossy dump).
• The stage frequencies and the valves' firing voltages are not given in the pages read.
• No test is on record of energy drawn per impulse against load for Moray's tubes, and no measurement of reverse isolation for them.
• Whether the "harmonic coupling" adds energy at all is the subject of claims 3 and 4, not this question.

Corrections to earlier documents

• Claim 4 in claims.md and arbiter.md calls the valve a "rectifier-type detector". Moray's text supports a two-use description: in the cascade it is a threshold gas-discharge coupler that passes energy "in only one direction at a time ... without necessarily changing ... to D.C." (p.139) and fires at a set voltage (pp.55-56). In his receiver hookup the same assembly is used "for the rectification and amplification of radio ... signals" (p.139). Suggested wording: "a one-way coupler (threshold gas-discharge valve) between resonant stages, also used as a rectifier in receiver hookups". This sharpens the correction already noted in aaa-progress.md. It does not remove the rectifier reading, which p.139 itself supports.