RankingsPrimerArbiterPhysicistMathematicianContrarianQEClaimsSources

Farnsworth-PT: Arbiter (Step 3)

Document under review: US Patent 2,172,152, "Radio Frequency Multipactor Amplifier", Philo T. Farnsworth, filed 21 Feb 1938, granted 5 Sep 1939 [read; confirmed [web] https://patents.google.com/patent/US2172152A/en]. I read the three reviews, facts.json and claims.md. I did not re-read the patent PDF; I rely on the reviewers' three independent, mutually consistent readings of it [read, via reviewers].

Source gap note (repeated)

The folder holds exactly one document, the patent, and its text layer is readable (OCR noisy, e.g. the drawing sheet shows "1958" for 1938). There is no source gap for the patent itself. There are NO other Farnsworth works in the folder. Everything about the multipactor tube's performance, history, fate and Farnsworth's wider ideas rests on [web] or [memory] only, and is labelled so below. The patent contains no equations and no measurements, so no verdict below rests on data from the document.

Background in brief (for the reader)

Secondary emission: an electron striking a surface releases on average delta secondary electrons; delta can exceed 1 for suitable surfaces and impact energies. Multipactor ("AC electron multiplier"): in an RF field, electrons shuttle between surfaces; if transit time is an odd multiple of half an RF period and delta > 1, the electron count grows exponentially until space charge or phase slip limits it [web: https://en.wikipedia.org/wiki/Multipactor_effect]. Farnsworth studied this from 1934 as an amplifier/oscillator [web, same URL]. The effect is today mostly a nuisance in satellites, accelerators and waveguides [web, same URL].

Reconciling the reviewers' disputes

1. Numbers that differed or needed recomputation

  • Mathematician: parallel-plate half-period resonance gives 100 MHz, 1 cm: 62.5 V peak, 32 eV impact; 100 MHz, 0.5 cm: 15.6 V, 7.9 eV; 10 MHz, 1 cm: 0.62 V, 0.32 eV. I re-solved this model (electron leaves at rest at RF phase 0.3 rad, field E = (V0/d) sin(wt+phi), transit = half period). Result: 62.49 V / 31.75 eV; 15.62 V / 7.94 eV; 0.625 V / 0.318 eV [my calculation, numpy]. Confirmed. Caveat: this is a crude parallel-plate analogue of a helix-and-gauze tube with an unknown DC bias, so it shows only that RF amplitude alone gives impact energies far below the roughly 300 eV where silver surfaces peak, hence the DC bias must matter [SUPERSEDED by QE revision 2]. It is an illustration, not a model of the tube.
  • Contrarian: 100 V gives about 5.9e6 m/s and "crossing 1 cm takes about 1.7 ns, which is half an RF period at about 300 MHz". I recomputed: v = 5.93e6 m/s is correct (it is the final speed). 1.7 ns is the time at constant final speed. An electron accelerated from rest through 100 V across 1 cm takes 2d/v = 3.4 ns, so half-period matching is at about 148 MHz, not 300 MHz [my calculation]. The contrarian's number is off by a factor of 2 (it overstated the frequency). The qualitative conclusion (at 10 MHz a 100 V, 1 cm gap is far faster than the half-period, so higher-order or longer-path resonance, or lower voltage, is needed) stands and is in fact strengthened. No verdict depends on this.
  • Mathematician: delta^10 for delta = 2, 3, 5 gives 1.0e3, 5.9e4, 9.8e6. Confirmed: 1024, 59049, 9765625.
  • The "one in four" gauze transmission. All three flag that the patent equates transmission of secondaries with open-area ratio ("very nearly ... one to four") [read, per reviewers]. [SUPERSEDED by QE revision 1: the patent applies it to electrons arriving from the helix, not secondaries; the mathematician's reading was correct.] Either way it is asserted, not derived, and secondaries leave a wire surface with a broad angular spread, so a geometric-area rule is only a rough approximation [memory, medium: plausible geometry, no verdict depends on it].

2. Performance evidence (the one real disagreement of substance)

The physicist found no evidence a tube was built; the mathematician found none for replication or commercial use. The contrarian found a 1934 trade-press article. I fetched it: Arthur H. Halloran, RADIO magazine, Oct 1934, reproduced by Borderland Sciences. It says "An undistorted output of 25 watts was obtained from an input of 25 watts", that one small tube gave "100 watts of undistorted energy before the cathodes were destroyed by heat", oscillation "from 200 kilocycles to 60 megacycles", and that the "tube's theoretical output is twice that of an equivalent hot-cathode tube operated as a Class A amplifier" [web: https://borderlandsciences.org/journal/vol/44/n05/Farnsworth_Multipactor_Tube.html]. Assessment: the contrarian was right that such a source exists, and the physicist's "no evidence it was built" is therefore too strong for the multipactor tube family. But: (a) it is a 1934 promotional-era magazine piece, four years before this patent's filing, and nothing says it describes this screen-cathode/helix geometry; (b) the phrase "25 watts out from 25 watts in" is ambiguous. The contrarian read it as a power gain near 1. It could equally mean input DC or drive power; I cannot tell from the excerpt, and the article elsewhere speaks of cathode-heating limits. I therefore do not treat it as evidence of low gain nor of high gain. It is inventor-side, uncontrolled, and unreplicated by anyone I can cite. It does not change any verdict; it moves the multipactor family from "no record of being built" to "reported built, with uncontrolled inventor-side figures".

3. Fate of the tube: "tested and failed", "never tested" or "dismissed"?

No reviewer or I found a source showing a neutral, controlled test (positive or negative), a source showing a documented failure on test, or a source showing suppression or blocked funding. The only documented limit is cathode heating (Halloran, above). Why the tube vanished from mainstream use is not documented in anything I read. The patent's citing list shows only four later patents, from 1960, 1963, 2023 and 2024 [web: Google Patents URL above]; two of the four are by one modern inventor in an energy-harvesting context, which I flag only as a caution that this patent is now being cited for much larger claims than it makes itself. I did not read those patents. So the correct label for Farnsworth's amplifier claims is "never properly tested (by a neutral party, on record)", and NOT "tested, not supported" and NOT "dismissed without testing". The physicist's "overtaken by triodes and pentodes" is [memory, medium] and unverified; I could not source it, so I do not use it. The label "Historically superseded but instructive" is therefore not assigned, because "superseded" needs a source on why and how.

4. Memory facts checked (budget 8; 2 fetches used)

  • Filing and grant dates; cited prior art 2,071,515 "Electron multiplying device" and 2,071,516 "Oscillation generator": [web] Google Patents URL above (upgraded from [memory]/[read]-only).
  • Farnsworth studied the effect from 1934 as an amplifier: [web] Wikipedia URL above. The contrarian's "Gutton 1924" detail stays single-source and I leave it as unconfirmed.
  • Halloran article numbers: [web], see above.
  • Silver and Cs-O yields (mathematician cited ScienceDirect pages: air-exposed silver delta max about 2.6 at about 300 eV; Cs oxide about 10 at low energy). I did not re-fetch these; they stay [web] as cited by the mathematician, unverified by me. They are illustrative and no verdict depends on them.
  • Multipactor resonance condition (odd half-period transit, delta > 1): [web] Wikipedia URL above. The arXiv and Fermilab URLs the contrarian cited were seen only as search snippets by that reviewer and not verified by me.
  • Remaining [memory] items (photoemission linearity below saturation, photomultiplier linearity, Child-Langmuir space-charge saturation; high confidence, textbook) are retained as [memory, high]. Secondary-emission shot-noise being significant [memory, medium] and "multipactor seeds from stray electrons or cosmic rays" [memory, medium] were not web-checked; no verdict depends on either.

Claim-by-claim arbitration

Claim 1: Permeable secondary-emitting screen cathode between helical accelerating anode and collecting anode can multiply an electron current under RF drive

  • Agreement: all three: no equations, qualitatively consistent, no conflict with experiment, no change to mainstream physics, no energy-conservation problem (extra power comes from DC supplies; RF only gates the process).
  • Disagreement: none on the verdict. Differences only on whether the tube was ever built (see 2 above).
  • Stronger argument: all agree and the physics is textbook. Mechanism (secondary-emission multiplication, multipactor) is Established [web, memory high]. A reservation the contrarian adds, which I accept: electrons passing straight through the open gauze are not multiplied, so per-pass efficiency is geometrically limited.
  • Experiment/calculation that would settle it: build the patent geometry and measure gain, third-order distortion and noise versus frequency and bias, with and without a seed source (the contrarian's proposal, adopted).
  • Final verdict (split): the multiplication mechanism: Established. This particular tube as a working amplifier: Never properly tested (revised in QE revision 4 from "Plausible, untested"; no neutral test on record; an inventor-side 1934 report for the tube family exists, geometry not confirmed to match).

Claim 2: Repeated trips give a greatly amplified output current

  • Agreement: no quantified gain anywhere in the patent; the patent itself writes "as is now well known in the art" and defers to prior art for the amplification; multipactor growth saturates (space charge, phase slip) so it is an oscillator or limiter unless held below threshold; the patent does not say which regime it operates in.
  • Disagreement: the physicist split it into mechanism Established and amplifier Never properly tested. The mathematician said Plausible, untested. The contrarian said Never properly tested, with "greatly" Not testable as stated.
  • Stronger argument: the physicist and contrarian. The scale reserves "Plausible, untested" for claims with no real test, while here there is a documented inventor-side report (Halloran) and still no neutral controlled test; "Never properly tested" is the label that fits and matches the instruction to say whether it was dismissed. I add the stability point in the mathematician's favour: a round-trip yield above 1 gives geometric growth (delta = 3 over 10 impacts gives 59,049x), so a linear amplifier needs either operation below threshold or an explicit limiter; the patent is silent. "Greatly" with no number cannot be refuted.
  • Experiment that would settle it: measure small-signal power gain and distortion of the patent tube across bias, stating whether it runs below threshold; report stability margin.
  • Final verdict (split): the electron-multiplication mechanism: Established. "Greatly amplified" linear amplification in this tube: Never properly tested (not shown dismissed without testing; not shown tested and failed). The word "greatly" without a figure: Not testable as stated.

Claim 3: Output linear in electron supply per cycle, so a gun-control signal modulates the carrier

  • Agreement: all three note the patent makes the linearity statement only in the photo case, "in its maximum value ... up to the point of space charge saturation", and extends it to the gun by reference; below saturation a fixed multiplication factor M gives linearity; at saturation output is independent of the seed; M depends on RF amplitude and bias so amplitude-modulation distortion and noise coupling are expected near threshold; no data.
  • Disagreement: none on the verdict.
  • Stronger argument: the shared position. A further point from the mathematician that I keep: baseband modulation of a DC beam does not automatically become carrier-amplitude modulation unless multiplication is bunched at the carrier frequency; plausible, not shown.
  • Experiment that would settle it: sweep gun-control current, plot output carrier amplitude, and measure intermodulation distortion.
  • Final verdict: Plausible, untested, valid only in the unsaturated small-signal regime (which is also the patent's own stated limit).

Claim 4: Above about 10 MHz a seed of electrons is needed; photo-initiated output is linear in light up to space-charge saturation

  • Agreement: qualitatively sound (build-up needs a fixed number of RF cycles, so a seed matters when each cycle is short); photoemission linear in intensity below saturation is textbook [memory, high]; "10 MHz" is empirical and given without data. (The N/f build-up remark here is SUPERSEDED by QE revision 3.)
  • Disagreement: the physicist split off the 10 MHz threshold as Never properly tested; the others kept one verdict.
  • Stronger argument: the physicist's split is cleaner: the general need for seed electrons is mainstream multipactor knowledge (stray electrons or cosmic rays normally supply them [memory, medium]), while the specific number has no support. The contrarian's 1.7 ns/300 MHz arithmetic is corrected above (148 MHz), and without dimensions and voltages the 10 MHz figure cannot be checked from the patent at all.
  • Experiment that would settle it: measure multipactor start-up time against frequency with and without a light or gun seed in the patent geometry.
  • Final verdict (split): need for a seed electron source and photo-linearity: Plausible, untested for this tube (the underlying photoelectric linearity is Established). The specific ~10 MHz threshold: Never properly tested.

Claim 5: No focusing fields are needed; simple combined amplifier and modulator

  • Agreement: this is an object of the invention in the patent, not a result; "simple" is a design opinion; no beam-optics argument given; many tubes of the era worked without magnetic focusing [memory, medium]; the patent does include electron optics (ring control electrode, gun anode), so "no focusing" is shorthand for "no magnetic or extra focusing field".
  • Disagreement: the physicist and contrarian give a split; the mathematician gives the same split. No real disagreement.
  • Stronger argument: shared.
  • Experiment that would settle it: trace beam current to the screen and to the collector through the helix, with and without a magnetic field.
  • Final verdict (split): "simple" and "no focusing fields necessary" as design remarks: Not testable as stated. The narrow technical form ("works without magnetic focusing"): Never properly tested.

Corrected claim list

  1. A vacuum tube with a permeable, secondarily emissive screen cathode between a helical accelerating anode and a collecting anode can multiply an electron current when driven by an RF carrier. Mechanism Established; this tube as a working amplifier Never properly tested (QE revision 4).
  2. The patent asserts, by reference to prior art ("as is now well known in the art"), that proper potentials produce repeated trips and a "greatly amplified" output; it gives no figures. Mechanism Established; practical linear gain Never properly tested; "greatly" Not testable as stated.
  3. The patent states that photo-initiated current is linear in illumination "in its maximum value ... up to the point of space charge saturation" and extends linearity to the gun-supplied electron supply by reference, so a gun control signal modulates the carrier. Plausible, untested, small-signal only.
  4. The patent says it is "found necessary" above about ten megacycles to supply starting electrons by light or the gun. Need for a seed Plausible, untested; the 10 MHz figure Never properly tested.
  5. The patent lists as an object that no focusing fields are needed and the device is a simple modulator-amplifier. Not testable as stated; narrow form Never properly tested. Also corrected (see QE revision): the "one in four" figure is the patent's assertion for electrons arriving at the screen from the helix (not secondaries), equating transmission with open-area ratio; the "1958" in the OCR is the year 1938; the patent's first-page date check [web] confirms 1938.

What would settle it

One experiment answers claims 1 to 5 together: build the patent geometry (silver gauze about 25 percent open, oxidised and caesiated; helix inside; cylindrical collector; electron gun with ring control electrode) and record power gain, third-order intermodulation, noise figure, start-up time and output versus gun current, across 1 to 100 MHz and a sweep of helix DC bias, with and without a light or gun seed. A neutral lab should do it, publishing the raw data. A quick calculation that would settle the "impact energy" question: a dimensioned transit-time model (helix pitch, screen radius, DC bias) showing where the half-period resonance lies; the patent gives no dimensions, so the numbers would have to come from the 1934-38 Farnsworth patents (2,071,515, 2,071,516, not in the folder) or a rebuild.

Open questions

  • What gain did these tubes actually deliver, and was Halloran's "25 W out from 25 W in" a power gain, a drive figure or a DC input figure? Source: the full Halloran/RADIO (Oct 1934) article and Farnsworth Television & Radio Corp. test records.
  • Did the screen-cathode/helix geometry of this 1938 patent ever get built, and by whom? No source found.
  • Why did the multipactor amplifier disappear from practice? No source on this; cathode heating is the only documented limit. Suppression or blocked funding: no evidence found either way, and none is asserted.
  • How linear can an avalanche device be made, below threshold or with a limiter? Not addressed in any source I read.
  • What do the two 2023/2024 Robert M. Lyden patents (US11588421B1, US12136824B2; [web] https://patents.google.com/patent/US2172152A/en) that cite this one claim, and do they rely on Farnsworth's tube doing something the 1939 patent does not claim? Not read; flagged only.
  • Not in the folder (cited, not obtained): the earlier Farnsworth multipactor patents (e.g. 2,071,515, 2,071,516, 2,091,439) that this patent defers to for the gain.

Overall

A conventional, honest vacuum-tube patent. No extraordinary claim, no over-unity or energy-from-nothing claim, no conflict with mainstream physics, and nothing mainstream would have to change. The weakness is evidential, not theoretical: no data in the document, no neutral test on record. For a learner it is a good example of how a patent differs from a measurement, and of how an avalanche mechanism (multipactor) that is a hazard today was once pursued as an amplifier. Where the conclusions rest on [memory] alone (tube fate versus triodes, shot-noise, seed sources), they are marked as such above.

Source tags summary

QE revision (after qe.md, verdict MINOR)

New evidence gathered: a Sonnet/Opus deep read now exists (deep/US2172152-...-b4b016.md, text checked against a page image of PDF p.3), and I re-checked the disputed passage against the extracted text myself [read: p.3]. One web check: Google Patents citing list [web: https://patents.google.com/patent/US2172152A/en] shows four citing patents: US2925528A (1960, Hartnell-Beavis), US3096490A (1963, Raytheon), US11588421B1 (2023) and US12136824B2 (2024), both by Robert M. Lyden. I did not read them.

  1. "One to four" (corrects Reconciling section 1). The passage reads: free electrons pass the helix and go toward the screen, "the greater number will strike, producing secondaries. Some will pass through, however, the ratio being very nearly that of the apertured area to the total area, or one to four" [read: p.3]. The screen spec repeats it as geometry: "ratio of apertured area to total area is of the order of one to four" [read: p.2]. So about one in four electrons ARRIVING from the helix go through the holes to the collector; the rest strike and make secondaries. The mathematician's reading was right and the contrarian's "secondaries" correction was wrong; my earlier statement siding with the contrarian was an error made because I did not re-read the text. Open-area transmission is a fair first approximation for roughly radial arrivals, modified by field penetration at the holes; it remains an assertion with no measurement. A gain model following the patent: per arrival t is collected, (1-t) strike and yield delta(1-t) secondaries of which a fraction r return, so round-trip gain g = (1-t)deltar and G = t/(1-(1-t)deltar). Runaway threshold: r = 1/((1-t)delta) = 0.513 for delta = 2.6, t = 0.25 [my calculation, not run in code this round]. Qualitative conclusion unchanged: high gain needs g near 1 and is then very sensitive to delta.
  2. Impact energy (corrects Reconciling section 1, first bullet). I had endorsed "the DC bias must matter". In the patent's screen-to-helix-to-screen geometry the screen sits at DC ground and the helix has a DC bias; an electron leaving the screen and returning gains no net energy from the DC part, since net gain is e times the difference in helix potential at launch and impact (K_imp - K_launch = e(V_h,in - V_h,out)), at most 2eV_rf. So the DC bias sets timing and the RF swing supplies the impact energy [primer Step 2 and example.py Part 2, per QE]. The two-plate numbers (62.5 V, 32 eV etc.) stand as stated but are still not a model of this tube.
  3. Claim 4 reasoning. N/f build-up gets faster at high frequency, so it does not explain why a seed is needed above 10 MHz. The better argument: stray seed electrons arrive at a rate R per second, so seeds per RF cycle is R/f, which falls as f rises; build-up needs a seed within the first few cycles. The patent states only "found necessary ... in the time available" [read: p.3]. Also recorded from the deep read: the patent lists three starting mechanisms (ionised particles, photoemission, electron gun) and says it "preferably" uses the gun "since it possesses advantages in regard to modulation" [read: p.3]; light source and optics are "not shown". Claims 1-6 all require the gun; photosensitivity appears only in claims 2 and 4. This sharpens claim 3 (modulation is via the gun control electrode) and does not change the claim 4 verdict.
  4. Claim 1 / claim 2 label consistency. Halloran (1934) concerns gain and power, not the bare multiplication mechanism. For claim 1, the mechanism is Established; "this tube works as an amplifier" is then the same question as claim 2's practical gain, and for consistency it takes the same label: Never properly tested (an inventor-side report for the tube family exists; no neutral test; geometry not confirmed). This replaces "Plausible, untested" for claim 1's tube-level part.
  5. Source tags fixed above: Lyden citation now [web]; "Unread in the folder" reworded to "Not in the folder (cited, not obtained)". The contrarian's "all 3 text pages" is trivial (PDF has 4 pages, p.1 is the drawing sheet).

Changed verdicts: only the label on claim 1's tube-level part (Plausible, untested -> Never properly tested), for consistency with claim 2. Claims 2 to 5 are unchanged. No substantive verdict changes: the one-in-four, impact-energy and N/f corrections do not touch any verdict.