Changelog
A dated record of what changed on this site: content additions, data updates, and substantive corrections, newest first. How the content itself is produced is documented in the editorial methodology. This log begins with the site’s 4 October 2026 edition; records of earlier revisions are retained internally.
- — Two of the hosted Notes on the Mullins Mark I Cyclotron are revised: the operator's manual now writes down the hydrogen-line flush procedure as a startup step (jumpering the mass-flow controller's override lead opens its valve fully so the pumps can evacuate the line to the regulators), closing two of its listed errata, and the RF, dee and ion-source note carries the matching override polarity.
- — Corrections from an outside review of the site's consistency. On the amateur builds census: the Houghton row now pairs the field of its highest-energy run with that run's frequency rather than the magnet's later maximum; the Rutgers 12-inch row pairs its deuteron field with the deuteron frequency; the Cyclotron Project's frequency is marked as bench work; the reference machine's energy is described as computed, like the others, rather than measured; COLUMBUS's program dates are aligned with its entry; and the $125,000 estimate is attributed to Fred Niell, for parts and services, as Symmetry reported it. The comparison table now checks every unmarked field and frequency pair against the resonance condition at build time. Three Design Guide rules on iron saturation cross-reference each other with a note on why their figures differ by steel grade, and the rule drawn from the 2006 Houghton thesis names the machine it describes, Livingston's first cyclotron. The calculators page no longer promises a builder that is already live.
- — Seven of the eight hosted Notes on the Mullins Mark I Cyclotron are revised, each with a dated revision note on its title page. The chamber and vacuum note corrects its wall and bore arithmetic, gives the lids as circular, withdraws its earlier reading that the clean chamber was compression-limited after a molecular-flow model of the pumping path found the port and hose set the speed, and records the ion gauge's one-sided trip gate and the valve positions. The performance record now says the low-frequency species sweep would inform rather than settle the ion identity, since two candidate species coincide at every frequency. The RF, instrumentation, operator's manual, magnet and simulation-models notes carry smaller corrections. The bill of materials and the four model packages are unchanged.
- — A Current Openings page under Careers lists cyclotron, radiochemistry and accelerator jobs in the United States: title, employer, location and pay as each listing states them, with a link to the listing itself. The list is drawn from one standing job-board search, filtered to roles in the field, so it is not exhaustive and endorses no employer. Each listing shows the date it was posted and is removed thirty days later, whether or not the position has been filled. The News page's job-board directory remains for academic and laboratory posts.
- — Twenty-nine new Design Guide rules on gas flow through tubes, slits and orifices between molecular and viscous flow: the rarefaction parameter that places a line in its regime, tabulated flow coefficients across the Knudsen minimum, the effect of wall accommodation, orifice and short-tube factors, slip coefficients, and the thermomolecular correction for a gauge at a different temperature from the chamber. All come from Sharipov and Seleznev's 1998 reference-data review, which is cited and linked rather than hosted, with quotations held to twenty-five words. Each rule says which of the review's tables carry over to hydrogen and which the review limits to monatomic gases. The rules were externally audited before publication and every finding dispositioned. The library gains that review and the hydrogen viscosity reference that accompanies it. The Design Guide now holds 2,199 rules.
- — Every page now has a plain Markdown rendering for AI assistants and text tools, at the page's address plus index.md, generated from the published page at build. The renderings keep what a generic conversion loses: on a Design Guide rule the source's verbatim quote, this site's paraphrase and this site's editorial note are labelled separately; FAQ short answers stay apart from their detail; figures and calculator controls are described rather than dropped. The HTML pages remain canonical and the renderings are not offered to search engines. robots.txt also now states the site's content signals: search, reading by assistants and AI training are all permitted.
- — The Data page becomes Data and Downloads: five new exports join the design rules, bibliography and census snapshot — the Construction Classics index, the amateur builds census, the builder FAQ, the US legal survey by jurisdiction and the university programs table, all CC BY 4.0 — with a section on the hosted documents and every count on the page read from the file it describes.
- — A Tools page at /tools/ lists everything on the site that is operated rather than read: the hosted Virtual Cyclotron Laboratory and Cyclotron Builder, the browser calculators, and the live model, maps, filters and exports inside the reference pages, each with its grade and where it computes. It takes the Calculators slot in the main navigation; the calculators page is unchanged and is its first section.
- — The all-in-one Design Guide filter view is no longer offered to search engines; the twenty-five subsystem pages and the five level pages are the indexed surface. The filter view itself is unchanged and still linked from every subsystem page for readers who want to combine filters.
- — Search titles and descriptions revised on eleven pages so that a result answers the question the searcher typed: the directory, careers, the energy and lid-deflection calculators, legal, how a cyclotron works, the library, should you build one, history, the amateur builds census and the twenty-five Design Guide subsystem pages. Where a count belongs in a title it is computed from the dataset at build time. Page content is unchanged.
- — The homepage now opens with six entry points by what a visitor wants to do — understand how a cyclotron works, explore the world's cyclotrons, decide whether to build one, design a machine, research the engineering, study or work in the field — each with one sentence on what the visitor gets and the pages that deliver it. The worldwide directory, the Cyclotron Builder and careers, which had no card before, each have one; the laboratory sits under understanding. Prompted by an outside review of the site's discoverability. No page moved and nothing was removed.
- — The Cyclotron Builder opens at build.cyclotroninfo.com, hosted on its own subdomain and linked from the Learn section and the homepage. It is a planning-grade design tool: Simple mode asks six questions about purpose, budget, space, power, skills and parts already owned, searches its catalogs of magnets, coils, chambers, pumps and RF systems, keeps the designs that fit and shows the one with the most energy, with a costed bill of materials, a build-time estimate and a report whose safety section comes first; Full mode adds experiments the machine must support, energy floors, a pin on every subsystem, a compare screen and the cloud of every design that fit, and can verify a design's orbits. Costs are estimates with a range, energies are computed from the field at the usable radius, the beam current rests on one measured anchor, and nothing past 1 MeV is detailed; it produces no drawings. Each brief is computed on a backend from the brief and nothing else: no cookies, no accounts, no state kept beyond the browser (see About, Privacy).
- — Builder FAQ, from an outside editorial review: the "Who this is for" statement on the index is split into two paragraphs for scanning, the picture of the machine first and the units and vocabulary second, with no change of content; and the space-charge answer's closing sentence now matches its own opening, stating the expectation rather than a proof.
- — Builder FAQ: three more figures. Where the hydrogen ions can resonate (candidate positions for H+, H2+ and H3+ at two RF frequencies, and why the proton and the H3+ third harmonic coincide), the three dee arrangements from above with the gap in section (two dees, one dee against the chamber, one dee against a dummy dee), and what sets what in a small cyclotron's beam physics (the dependency chain from purpose to pressure target, with the constraints that push back). Each went through an external physics check before publication, with every finding adopted; that completes the six illustrations planned for the FAQ.
- — Builder FAQ: three figures. The first orbits seen from above (where the ions are born and why the first half-circle matters), where the dee's current flows and why it needs a coil across it (the tank, drawn with the Rutgers 12-inch's numbers), and what a resonance sweep looks like with its three controls (gas off, RF off, frequency moved). Each is an illustration of the idealised picture and says so; each went through an external physics check before publication, with every finding adopted.
- — A builder FAQ at /learn/faq/: fifty-two answers to the questions people ask when they are thinking about building a small cyclotron or have one half built, in eight groups across three parts (deciding and designing; building it safely; running it and proving a beam). Each answer stands on its own in its first paragraph and opens into the detail below it; every claim cites the Design Guide rule or the documented build behind it, and each answer assumes only a stated baseline, defining everything else where it is used. The index carries a "start here if…" list from symptom to answer. The grounding entry carries a reference-topology schematic that went through an independent electrical review before publication, and is labelled for what it does not show. Every answer went through external review rounds before publication, with every finding dispositioned. Readers can send feedback on any answer, or a question the FAQ should answer, from the pages themselves and from the Contact page; questions that fit get answered and posted as new entries, recorded here.
- — Six guided exercises for the Virtual Cyclotron Laboratory, with a series index at /learn/exercises/. Each page asks for a written prediction, opens the laboratory in a prepared state with one control to move, says what to expect and what a different result means, and cites the Design Guide rule behind every claim: find the resonance, watch an ion fall out of step, more voltage for the same energy, where the ions are born, what the leftover gas costs, and weigh the ion. Nothing is scored or stored. The laboratory itself gains settings links, a Copy-link button and pressure as a sweep axis. Externally refereed; every finding dispositioned in the series' notes.
- — The magnet-design page's permanent-magnet section is rewritten and expanded from the Design Guide's new permanent-magnet rules. It now follows the one built permanent-magnet cyclotron magnet from material model to field tuners, and adds what accelerator laboratories measured on permanent-magnet dipoles since 1980: how a fixed field is set without a current knob, how temperature drift is compensated and proven, how such magnets age, and how shims are computed from a measured field map. Eight sub-headings, each rule cited inline; externally refereed, with every finding dispositioned in the page's notes.
- — The homepage now carries a photograph: the Crocker 60-inch cyclotron at the Berkeley Radiation Laboratory in March 1939, from the US National Archives (record 39147342, marked Unrestricted), with its rights basis stated in the caption. The two Berkeley photographs on the history page are captioned the same way.
- — Thirteen more public-domain documents are hosted in the Construction Classics, on three new shelves. X-ray output below 100 kV holds a 1970 federal survey of what classroom discharge tubes emit, the 1978 tabulation of tube output and spectra at 20–60 kVp, and the 1961 and 1972 handbooks behind the barrier equation. Electrons in gas holds Johnson's 1922 gas-focused oscillograph, Brode's 1925 absorption coefficients, the Berger–Seltzer range tables and a Langley study of beam glow. Hot-filament arc sources holds two 1925 Physical Review papers on hydrogen ion species, two Oak Ridge filament-life reports and the 1954 Ohio State cyclotron final report. Each shelf states every document's rights basis and names the companions that did not pass the test, with the reason.
- — The Virtual Cyclotron Laboratory gains a Feedback link in its footer: it opens an email to the site's address with the page, the revision id and the card identifier filled in, so a report can be reproduced exactly on the backend. No form and nothing stored. The laboratory also now links from the homepage as a fourth route beside Understand, Assess and Design.
- — The Virtual Cyclotron Laboratory opens at lab.cyclotroninfo.com: an interactive simulator of an illustrative ~150 keV classical proton cyclotron, hosted on its own subdomain and linked from the Learn section. Set the field, RF frequency, dee voltage and source, watch a weighted ensemble of ions accelerate, inspect single paths crossing the gap, sweep one control across a range and export the curve. It is illustrative — shaped like the reference machine, not a calibrated replica of any real cyclotron — and every reading on the page comes from one evaluated ensemble named by its revision. Each run is computed on a backend from the settings in the request and nothing else: no cookies, no accounts, no state between visits (see About, Privacy).
- — Design Guide waves 22 and 23: 113 rules on two subjects. Fifty cover hot-filament ion-source practice — a 1954 Ohio State procedure for positioning the filament under beam and the source's construction, Oak Ridge's 1953–54 studies of calutron filament life (the three ways filaments fail, low-voltage "diode-like" operation, erosion against arc power, pyrometry and weighing as wear measurements), the two 1925 Physical Review measurements of how H+, H2+ and H3+ form in hydrogen, filaments in 1–3 T fields at two superconducting cyclotrons, and a patent on reversing the heater polarity. Sixty-three cover permanent-magnet accelerator magnets — a 1992 permanent-magnet cyclotron mass spectrometer, a variable-field permanent-magnet dipole tuned by a moving iron shunt, Halbach's design principles for rare-earth magnets and hybrids, passive temperature compensation of NdFeB, the Fermilab Recycler's ageing, compensation and shimming record, a 1 T NdFeB ring dipole, and low-cost Halbach magnets corrected with iron wires. The library gains seventeen entries, all linked to their holding repositories. Every rule went through the external referee before publication; its findings were dispositioned — thirty-seven by extending the verbatim quote to the adjacent source text, the rest by tightening editorial notes. Two corrections are worth naming: the notes on the 1925 hydrogen-ion papers now carry the modern channel thresholds from NIST beside the historical interpretation, and a centre-region note now distinguishes the first half-turn's reach (about twice the curvature radius) from the radius itself.
- — Design Guide wave 21: 61 rules on driving a high-Q dee with solid-state RF — LDMOS module and combiner design, what a module does when it faces a resonator's reflection (and what the tests do and do not show), coupling, Q and shunt-impedance characterisation of a cavity before power is applied, cavity conditioning and arc protection, digital low-level RF (fast/slow partitioning, sequenced start-up, spark recovery, tuner control), a retrofit case history, and absolute dee-voltage calibration by the X-ray endpoint method (1975 and 2018). Nine sources, all linked to JACoW, arXiv or the CERN-hosted 1975 proceedings. Every rule went through the external referee before publication; its findings were dispositioned — twenty-one by extending the verbatim quote to the adjacent text of these openly licensed papers, the rest by rewriting editorial notes so that a protection arrangement reads as one mitigation among several, never as a guarantee.
- — Design Guide wave 20: 90 rules closing three gaps — the physics of beam loss in residual hydrogen (evaluated cross-section compilations, how to look up molecular ions and integrate loss along the spiral, ion-gauge corrections for hydrogen); X-ray output from kilovolt electrodes and its shielding (a 1970 Bureau of Radiological Health study of X-rays from demonstration discharge tubes, measured tube outputs, and the NBS protection handbooks' barrier methods, housing limits and survey practice); and electrons in gas for a visible-beam demonstration (gas focusing from a 1922 oscillograph, slow-electron attenuation, electron ranges, beam-induced fluorescence, and RF and DC breakdown in coaxial lines and hydrogen). The library gains twelve entries, all linked to their holding repositories. Every rule went through the external referee before publication; its findings were dispositioned — thirteen by extending the verbatim quote to the adjacent public-domain source text, the rest by tightening the editorial notes so that tube-specific measurements read as comparisons, not bounds, and one arithmetic slip in a note was corrected in place.
- — Design Guide wave 19: 28 rules from two compact-cyclotron design papers — Karamyshev's three-sector, sixth-harmonic 15 MeV PET-cyclotron concept (IPAC 2022) and the VECC 10 MeV high-current compact cyclotron's magnet design (APAC 2007) — covering wide-sector magnets, low-ampere-turn coil design, high-harmonic RF and its central-region consequences, hill/valley geometry, sector-width isochronisation and equilibrium-orbit phase budgets. A new subsystem page, space charge, opens with its first rule. The library gains four entries (the two papers plus two cross-section compilations for hydrogen ions in H2: Barnett's ORNL-6086 and Phelps 1990), all linked to their holding repositories. Every new rule went through the external referee; its findings were dispositioned before publication, eight by extending the verbatim quote to the adjacent source text and the rest by tightening the editorial notes.
- — The news page groups several reports of one story: coverage of the same announcement within about three weeks now appears once, under the most recent headline, with the other outlets named and linked beneath it. The grouping is mechanical — shared names and phrases in the headlines — and the page's no-editorial-selection rule is unchanged. Prompted by an outside review that counted six headlines for one Calgary cyclotron announcement.
- — Methodology page correction: its description of the Design Guide's external rule audit still said a random sample of the back catalogue is audited monthly. That sample ran until every published rule had been through the audit once (complete in September 2026) and was then retired in favour of re-auditing edited cards whenever a batch of post-publication corrections accumulates; the page now says so.
- — The Design Guide gains five static pages, one per applicability level (/design-guide/level/1/ to /level/5/), each carrying only that level's rules with the same dg-NNNN anchors. Until now a first reading at level 1 opened the all-in-one page, which downloads every rule and filters in the browser; the level-1 page delivers the forty-odd rules it is about. The all-in-one guide is unchanged and remains the view for combining subsystems, levels, and search terms; "start at level 1" links on the homepage and the guide's front page now open the level page.
- — Corrections from an outside editorial review. The start-to-first-beam spread quoted on "Should You Build a Cyclotron?" had fallen behind the builds census (it still read "two to four years" for a record that now runs one to six); both it and the builds page's own summary line now compute from the census. The homepage no longer describes the collection as "verified" — it is source-checked, with each part's review status stated on the methodology page — and leads with the three reader routes ahead of the collection figures; its news card says plainly that the jobs section is a directory of job boards. Library topic chips are limited to subjects with three or more documents (one-off indexing phrases remain on their entries and in search), and the news feed's market-research exclusion now catches "CAGR" and "forecast period" headlines.
- — Four of the hosted "Notes on the Mullins Mark I Cyclotron" are revised to their builder's September 2026 editions, each opening with a dated revision note: the RF/dee/ion-source note gains the builder's direct RF-system measurements (terminal capacitance, loaded Q, and the component table behind them — with dee voltage still stated as unmeasured), and the simulation-models note withdraws its earlier stem and feedthrough capacitance-by-subtraction statements. The instrumentation and performance notes carry the same revision box; page counts on the collection page update with the files. The series remains CC BY 4.0 under its author's name.
- — The worldwide cyclotron directory is live: 1,564 machines in 92 countries on a filterable map and list — identity, location at stated precision, category, current status, technical class, verification date, and public-tour availability with each facility's own tour link. The directory is published as a dated quarterly edition (this one: 4 October 2026), and every census-derived figure on the site computes from it, so the whole site shares one as-of. The data is downloadable as JSON or CSV under CC BY-NC 4.0; the Design Guide and bibliography exports are CC BY 4.0. Every record is verified against public sources, which are retained and available to researchers on request.
Spotted an error, a dead link, or an out-of-date claim? Corrections are welcome at info@cyclotroninfo.com — fixes are made in place and noted here.