This is a census of documented amateur cyclotrons: machines designed and built by
individuals, high-school teams, or small undergraduate groups outside professional
accelerator programs. An entry qualifies if the build itself is documented — a project
site, thesis, conference paper, or credible press account — not merely rumored. Design
studies count when they left a public record; unverifiable claims are noted as such.
Few of these projects reach beam, and the reasons repeat across seven decades: a
cyclotron demands a large and carefully shaped magnet, high vacuum, a kilovolt-level RF
resonator, a working ion source, and diagnostics — every subsystem at near-professional
standard before anything at all happens. Builders at the 2010 amateur cyclotron
conference estimated two to three years of work and roughly $125,000 in parts if bought
new (Symmetry, Aug 2010), which is why nearly every successful machine leans on donated
magnets, surplus hardware, and patient scrounging. The entries below are presented in a
uniform format, grouped by what the record shows each machine did — operating today,
ran and stopped, still building, stalled, or never documented past the build — and
chronological within each group, with every claim traceable to the linked sources.
27 documented projects · 8 with a demonstrated or routinely
operated beam.
Plotted as a timeline, the census tells its story at a glance: a cluster of school
builds in the postwar decades, then a silence of more than twenty years before the
renaissance that began in the 1990s and has not slowed since.
operated beam demonstrated in progress stalled outcome unknown
Documented amateur and school cyclotron projects, from start year to first beam or last
documented activity (dot = single documented year; arrow = ongoing). Bar colour marks each
project's documented outcome, keyed above — cool for the machines that reached beam,
warm for those that have not, grey where the record does not say; spans for entries the
census dates with “c.” are approximate. Hover a row for details, or select it to jump to
the full entry below.
What the census teaches
27 projects is a small evidence base, censored in a known direction: a
build that was never written up never entered, so failure is undercounted here. Four
patterns survive that caveat.
First beam takes one to six years. 7 of the
8 machines that reached beam date both the start of work and the
first beam, and the gap runs from 1 year to 6. The short end belongs to
the home builds — both Niell machines and the Mullins
Cyclotron went from first work to observed beam in about a year — while the
Rutgers 9-inch took four years and
Houghton six. The two-to-three-year estimate from the
2010 amateur conference (Symmetry, Aug 2010) sits inside that spread, not at its edge.
The magnet's provenance does not pick the winners. 4 of
the 8 machines that reached beam wound their own magnets from raw
steel and wire; the rest ran on donated, surplus, laboratory, or repurposed historic
iron, and the stalled projects divide roughly the same way. On this record, scavenging
a magnet saves money and months; it does not decide whether beam happens.
Achieved energy is measured in keV. Most documented beams sit between
about 70 keV (Niell II, computed from its observed resonance)
and about 184 keV (the Rutgers 9-inch, computed from a run's
field and radius); the Mullins machine's ≈164 keV is the
best-documented measured figure in that cluster. Two university-lab machines sit
above it: the Rutgers 12-inch at 800 keV, reached eleven
years after the project began, and the
1957 Iowa State undergraduate cyclotron, whose
1.5 MeV is its paper's stated maximum with the method of determination unstated.
The famous bigger numbers are softer than
they look — El Cerrito's “million-electron-volt” machine is a 1947 press claim with no
surviving measurement, and the MIT build's 2 MeV design energy has no documented beam
behind it. Choosing your machine draws the
design lesson: pick the purpose first and let it set the energy.
The commonest ending is silence. 11 of the 27
entries close at “outcome unknown” — the record simply stops before it says what the
machine did. Four are postwar school builds known mainly through period press, and the
problem did not end with print: the 2007 MIT machine survives as a thesis and two
videos, and the 2019 Takanashi build through news coverage of its construction. That
decay is why an entry here requires a documented record, and why the sources cited
below carry archived copies wherever a snapshot is verified.
The census in one table
Every machine on one grid, so the pattern shows without reading 27 entries:
pole diameters from 150 to 483 mm, fields from 0.19 to 1.7 T on the
machines that ran, and energies that are, in nearly every source here, the energy at
the radius where beam was detected, computed from the field rather than measured. Two machines have driven
nuclear reactions with the beam, both on internal targets: the Rutgers 12-inch makes
D–D fusion neutrons at about 500,000 per second on a deuterated target, and Houghton
has detected the same reaction. No machine in the census documents an extracted
beam. Six state outright that the beam
stays inside the chamber — the Coburg team give keeping the beam inside as their reason,
which limits what escapes but is not the same as a machine producing none — and the rest are silent, which is not the same as no. The nearest thing is
the Rutgers 12-inch’s electrostatic deflection channel: real extraction hardware,
a 30 kV septum and electrode built in 2009, but pointed at an internal phosphor screen
and used as a velocity filter to measure beam energy rather than to bring beam out of
the chamber. Extraction hardware is designed more often than it is tested: Knox built an
extractor around a deflection plate on a machine that never reached beam, one dee on
The Cyclotron Project carries an unused
extraction slit, and Coburg — having given keeping the beam in as the reason not to —
began building an extraction system with IBA in 2021. The Maryland 19-inch is the only
machine here whose published design goal is a beam out of the chamber — joined in intent
by the IUAC tabletop, whose director's stated milestone is "We first need to demonstrate
beam extraction" — and neither has run yet; beam extraction explains why that is the hard
step.
Procured to IUAC's design: H-frame water-cooled DC electromagnet, 1.2 T median plane, 305 mm poles, 51 mm nominal gap, 2 × 162 turns at 200 A (tender 09/GOR/2024–25)
Field and RF are not always a matched pair. Where one is a design figure and the
other a bench measurement, the two will not satisfy f = qB/2πm —
that is the sources' state, not an error here, and the markers say which is which. Hover either
value for the specific run or measurement it comes from.
27 machines; 8 produced beam. Energy and current appear only for those
8, and only as the sources report them: c marks an energy
computed from the measured field and the radius at which beam was detected — the practice in
most sources here — r an energy the builders report without stating
the method, and an unmarked energy has an independent check. Field:
d design value, e the builders’ estimate,
b a bench measurement of a subsystem rather than a running machine,
unmarked measured or operating. Ø is pole diameter unless marked (dee) or (chamber). A dash
means the sources do not state it; a blank energy or current means no beam result is documented
for that machine — which, for the rows whose outcome is undocumented, is not the same as saying
it never produced one. Hover a value for its source page; every row links to the entry and its sources. Click a
column heading to sort.
The programs
Seven of the 27 entries below are not standalone builds but machines belonging
to four sustained programs — and the census's clearest pattern is that this is where beam
happens. Four of those seven reached beam, against four of the twenty entries that stand
alone. The programs also fail differently from individual builds: an individual runs out of
money, patience or garage, while a program runs out of students. Each of the four
has a different answer to that, and the answers are more transferable than any of their
specifications. The counter-example is instructive too — the Cyclotron Kids built their
machine at a national laboratory with staff support, which is program-shaped in every
respect except that nothing followed it.
One thread runs through the census without ever getting a table row of its own: Fermilab.
Two of the Fort Wayne 1951 students went on to work there, and
Symmetry's
2010 survey names two more employees — Ron Walker and Chris Olsen — who "each built
cyclotrons while working there (Olsen actually built three)." Olsen's own account:
"I'd worked at the campus particle accelerator lab all four years of college and had built
a couple particle accelerators at home … I brought my own cyclotron to my interview
here" (Fermilab,
2015) — and he assisted the Knox build below. None of these personal machines has a
recoverable date, place, size or outcome, so they stay here as builders' history rather
than machine entries; if a record of any of them surfaces, it earns a row.
The Koeth lineage — Rutgers, then Maryland
1995 – present · three machines
Begun in 1995 as a personal project by two Rutgers undergraduates, this is the longest
continuous amateur-origin cyclotron effort on record and the one that has produced the
most physics. A 9-inch prototype built around a surplus NMR magnet demonstrated proton
resonance in 1999; a 12-inch machine followed, built around a 4,600 lb magnet recovered
from a 1950s laboratory where it had analyzed the extracted beam of a 60-inch cyclotron.
From 2001 the 12-inch became the centerpiece of the Rutgers modern physics teaching lab,
reliably producing 800 keV protons by 2006 and, after a PIG source upgrade in 2012, tens
of microamperes over 30-hour runs. In 2016 it moved to the University of Maryland, where
it anchors a capstone design class and a 19-inch successor is being built — the first
amateur-origin machine designed to extract its beam.
Its continuity mechanism is the annual student project, and the record reads as a list of
them: orbit calculations, weak-focusing pole tips, a two-dimensional field mapper, an RF
auto-tuner, an electrostatic deflection channel, spiral-sector azimuthally-varying-field
pole tips, a central-region simulation, a measurement of bunch length within a single RF
cycle, and most recently fluorine internal targets with a field-insensitive gamma
detector. The machine now drives a deuterated target for D–D fusion neutrons at about
500,000 per second, and has been pulsed down to a 10 µs beam-on-target window to measure
neutron diffusion time in a moderator. Its builders report that more than a quarter of
the students who passed through went on into accelerator physics — which is the output
the program is actually optimizing for.
The lineage also propagates in writing. The German COLUMBUS school cyclotron's builders
record that their ion source "was built after the pattern of Tim Koeth", citing the
project's ion-source memo; India's IUAC tabletop cyclotron is being designed with the
12-inch's builder advising remotely; and the builder's own 2026 project history reports
that students elsewhere used the documentation as a model for their own accelerator
projects. Two decades of that documentation — construction memos, conference papers, and
the neutron-program reports — are collected, with archived fallback links, in the
library's Rutgers/Maryland corpus reading guide.
2001 – present · one machine, twenty years of theses
A deliberately small machine used as a continuing undergraduate research program since
2001. It first accelerated hydrogen ions in 2007 and has since run protons, molecular
hydrogen ions, helium and — in 2022 — deuterons, whose run produced a detected
D(d,n)³He neutron signal. Beam currents are nanoamp-scale with a peak near 100 nA. It
hosted the first amateur cyclotron conference in 2010, which is how several builders in
this census first met each other.
Its answer to generational turnover is documentation: roughly twenty theses and
conference talks sit in the college repository, each student generation recording what it
changed and why, so the next one inherits a written machine rather than a folkloric one.
That is also why this entry is among the best-sourced in the census — the comparison
table's Houghton row is built almost entirely from student theses. The trade-off the
program lives with is that hardware failures have repeatedly interrupted operation, and
the thesis record is candid about it.
2010 – present · school project turned university instrument
Started as a school project at the Gymnasium Ernestinum in Coburg, supported from 2012 by
the Institute for Nuclear Physics at Forschungszentrum Jülich — which supplied both
expertise and a 430 kg magnet — along with Hochschule Coburg and German vacuum industry.
It was the first cyclotron built for school and teaching purposes in Europe, and its
design was presented at the 2013 international cyclotron conference. The team chose deliberately not to extract the beam,
giving as their reason that no harmful radiation should escape during experiments — a
design decision worth reading beside the Rutgers entry, where an internal target produces
neutrons with no beam leaving the chamber either.
Its continuity mechanism is institutional migration: the machine now lives in the applied
vacuum technology laboratory at Hochschule Coburg, where it is used for undergraduate
measurements in the technical physics degree and for two or three school workshops a
year. Its builders published the design as a short book in 2020 — the only machine in
this census with a published teaching text — and a Japanese translation is in
preparation with the AxeLatoon collaboration. A school project that became a teaching
instrument, then a reference design for other people's programs.
A program in which students at Japan's technical colleges (高専 / KOSEN) design and build
small cyclotrons themselves, mentored by accelerator scientists from KEK, RIKEN and
several universities. It chose the cyclotron for two reasons its founders state plainly:
the machine is understandable within high-school physics, and one member of the
collaboration had already built a small cyclotron at home — so a build procedure was
already circulating among the staff. That member's apartment-built machine is an entry in
this census, which makes AxeLatoon the clearest case anywhere of a hobby build seeding an
institutional program.
Around forty students are active across the sites. Oyama's machine has been operating
since January 2023, its detector registering particles as a current while the team works
toward an unambiguous first ion-beam observation; Oyama also wrote the reflected-power
tuning procedure the other sites now follow. Nagano has its electromagnet and both pumps in
hand and waits on an RF source, a chamber and gauges. Toyota, working with vacuum hardware
borrowed from KEK, reaches about 1.9 × 10⁻² Pa in an hour of pumping; its
recruiting demonstration puts a tungsten filament, an accelerating electrode and an LED
inside a stainless cooking pot, sealed to the vacuum line with bathroom caulk because
welding was out of reach, and got to 2.4 × 10⁻² Pa. The team work the number the
way the physics asks: that pressure is a mean free path of about 27 cm, so the filament and
the LED need only sit closer together than that. Tsuyama
runs accelerator building as a credited course, though its machine is an electrostatic
linac and so falls outside this census. Ibaraki's build stopped when its core members
reached job-hunting year, and the collaboration names generational turnover as the
structural problem for student projects rather than glossing it. Stated ambitions are a
nationwide roll-out and a national inter-college accelerator contest. Site-by-site
progress is reported in the collaboration's 2025 review in the Journal of Plasma and
Fusion Research, linked from the Oyama entry below.
Machines that reached beam and whose public record runs to the present. What "operating" means varies — a teaching-lab machine on a run schedule, or a program its builders still document as active.
Rutgers 12-Inch Cyclotron
Tim Koeth, Stu Hanebuth, and successive Rutgers undergraduates1995–present
operatedUSA800 keV
Pole diameter
305 mm (12 in)
Field
1.2 T (12 kG) design; 0.96 T average measured for the deuteron runs with spiral AVF pole tips
RF frequency
≈15.1 MHz measured tank resonance; 7.15 MHz for deuterons; 18.3 MHz design at 1.2 T. Note the deuteron pair 7.15 MHz / 0.96 T is ~2% off mutual consistency with f = qB/2πm (7.15 MHz ↔ ≈0.94 T) — both are recorded settings from the 2020 draft, not a validated matched pair
Particle
protons and deuterons (H₂⁺ also observed at reduced field)
Energy
800 keV, 200 nA demonstrated by 2006; 1 MeV capability; tens of µA after 2012 PIG source upgrade
Extraction
none — the 2009 electrostatic deflection channel (30 kV electrode, stainless septum) turns the beam onto an internal phosphor screen and is used as a velocity filter to measure beam energy
Neutron production
D–D fusion neutrons on a deuterated internal target, ≈5 × 10⁵ n/s during a 5 s continuous run; routine operation is long-pulsed at ≈10% duty factor
The best-documented amateur-origin cyclotron: begun as a personal project by two undergraduates in 1995, it became the centerpiece of the Rutgers modern physics teaching lab from 2001, reliably producing 800 keV proton beams by 2006 and, after a PIG ion source upgrade in 2012, tens of microamperes over 30-hour runs. It is also the census's most productive machine as an instrument: tuned for deuterons at 7.15 MHz and 0.96 T with spiral azimuthally-varying-field pole tips, it drives a deuterated internal target to produce D–D fusion neutrons at about 500,000 per second, calibrated against a NIST-traceable ²⁵²Cf source, and has been pulsed down to a 10 µs beam-on-target window to measure neutron diffusion time in a polyethylene moderator. Students have also attempted the ¹⁹F(p,αγ)¹⁶O reaction with fluorine internal targets and a field-insensitive gamma detector. Nineteen undergraduates gained hands-on accelerator experience on the machine, several going on to accelerator-physics careers. In 2016 it moved to the University of Maryland, where it anchors a capstone design class.
Mark Yuly and successive undergraduate students, Houghton College, New York2001–present
beam demonstratedUSA≈157 keV
Chamber
170 mm (6.7 in) inner-diameter aluminium chamber (a post-2008 replacement for the original 149 mm brass ring)
Field
≈1.1 T (11 kG)
Particle
protons, deuterons, helium ions
Energy
≈157 keV computed at the highest-frequency run (12.13 MHz, 796 mT, 2013); nanoamp-scale internal beams routine, ≈100 nA peak
Vacuum
diffusion-pumped high-vacuum system
Beam current
nanoamp-scale routine; ≈100 nA peak (2013); ≈20 nA deuterons (2022)
A deliberately small cyclotron developed as a continuing undergraduate research program begun in 2001, with each student generation documenting improvements in theses and conference talks — the college repository holds roughly twenty of them. It first accelerated hydrogen ions in 2007 and has since run protons, H₂⁺, helium and, in 2022, deuterons; the deuteron run produced a detected D(d,n)³He neutron signal. Beam currents are nanoamp-scale, peaking near 100 nA, and hardware failures have repeatedly interrupted operation. It hosted the first amateur cyclotron conference in 2010 and remains a teaching and research tool.
Ch. Wolf, M. Frank, E. Held and students, Gymnasium Ernestinum, Coburg2012–present
operatedGermany
Pole diameter
150 mm (6 in), Bruker BE-15 laboratory magnet
Field
0.187 T (1.87 kG) at the documented proton run; the 2022 table gives 185 mT for protons and 370 mT for H₂⁺; 0.38 T design point, up to 0.7 T available
RF frequency
2.82 MHz at 1000 V dee amplitude in the documented run; 5.63 MHz design. Since the matchbox was improved, two frequencies are selectable, 2.82 and 5.64 MHz (the 2022 paper's table gives the lower one as 2.85 MHz, its own text as 2.82)
Particle
protons and H₂⁺, identified by specific charge
Energy
Three published figures, and they are not the same quantity: 24–48 keV design (2013 paper), 8–16 keV as built (builders' 2020 book), and ~4.1 keV for protons / 7.5 keV for H₂⁺ in the 2022 technical-data table. Beam energy has never been measured — particles are identified by specific charge
Vacuum
10⁻⁶ mbar base and 10⁻⁵ mbar with hydrogen flowing in the 2022 technical data; the 2013 design paper's table gave 10⁻⁵ and 10⁻⁴ mbar
Chamber
200 mm (7.9 in) diameter × 75 mm high vacuum chamber, built free of charge by VACOM
Ion source
thermionic filament in a Macor chimney with puller electrodes on the dee; 7 A heating, 120 V anode, 2 mA emission, 0.10–0.15 ml/min hydrogen
Detector
home-built Faraday cup on a semi-rigid cable used as a radial guide rod; it reads a few pA to nA, but the experiment evaluates only whether a signal is present, not its magnitude
Built as a school project at the Gymnasium Ernestinum in Coburg with a 430 kg magnet and expertise from Forschungszentrum Jülich, which describes it as the first cyclotron built for school and teaching purposes in Europe. The builders date the project to 2012 and first beam to April 2014. Its 2013 conference paper sets out the design: a donated Bruker laboratory magnet, a 120 W marine-band transceiver as the RF source, a Koeth-pattern filament ion source, and a deliberate decision not to extract the beam so that no radiation can leave the experiment — a position the team has since revisited: in August 2021, with help from IBA in Belgium, they began building an extraction system to deflect the beam through a Wien filter and measure the ions' speed and energy against calculation. The machine now lives in the applied vacuum technology laboratory at Hochschule Coburg, where it is described as fully functional and is used for undergraduate measurements and two to three school workshops a year — an amateur-origin machine that became a teaching instrument. Its builders published a short Springer book on the design in 2020, which documents the machine as operated rather than as designed: sweeping the magnet slowly while the Faraday cup sits at a fixed radius records an I(B) spectrum whose first peak, at 187 mT for 2.82 MHz, gives a specific charge within 1% of the proton value, with a second peak identifying H₂⁺. The signal's strength is deliberately not evaluated — the experiment asks which particles are circulating, not how much beam there is, so the machine has no measured beam energy or current. A 2022 conference paper reports the machine as it now stands: a 200 mm × 75 mm chamber, an improved matchbox offering two selectable frequencies, and a base pressure an order of magnitude below the 2013 design figure, alongside the workshop and internship programme built around it.
Simon and Chris Mullins (father and son, home-built)2017–present
operatedUSA≈164 keV
Field
580–590 mT typical (582 mT at the ≈164 keV run)
RF frequency
8.9–9 MHz (8.87 MHz at the ≈164 keV run)
Particle
protons
Energy
≈150 keV typical operation; ≈164 keV best demonstrated
Beam current
3.2 nA best (at the ≈164 keV run); 1–2 nA typical
Vacuum
≈6 × 10⁻⁶ torr while operating
Ion source
hydrogen filament source; H₂ feed through a used 1 sccm mass-flow controller (the one significant upgrade to date)
A father-and-son basement build documented on a dedicated project site, carried from magnet design through working machine between 2017 and 2019. Operating at 582 mT with an 8.87 MHz RF system, it demonstrated acceleration of protons to approximately 164 keV. The build log covers the magnet, vacuum system, RF chain, and beam measurements. Status as of August 2026, per the build log: still operational, with over 170 logged runs and the most recent in 2025; typical operation is 1–2 nA of beam at 580–590 mT and 8.9–9 MHz, about 150 keV protons, at roughly 6 × 10⁻⁶ torr — the one significant upgrade to date is a hydrogen delivery system built around a used 1 sccm mass-flow controller.
Machines that reached beam and whose public record then ends — finished projects, graduated students, dismantled benches.
Iowa State University Undergraduate Cyclotron
Undergraduate Physics Club, Iowa State University, Ames — original group led by Clark M. Varnum, advised by D. E. Hudson; later student generations through at least 19631954 – at least 1963
A 2.5-ton, 10-inch cyclotron built by Iowa State undergraduates, begun in fall 1954 as a joint project of the Undergraduate Physics Club and the Physics Department laboratory program and first run in spring 1957. It accelerated protons to 1.5 MeV at 2 µA and served a sustained teaching program — undergraduate laboratories, NSF Undergraduate Research Participation, two master's theses, and nuclear work including N-13 production from carbon for half-life measurement. Four Iowa Academy of Science papers document it (McGuire on the machine, Burns on the experimental program, Mueller and Moses on orbit calculations, 1961–1963). McGuire's own scoped claim: 'the first operating “small” machine known to the author which was large enough for serious work in nuclear physics and which at the same time was built and manned essentially by undergraduates.' The machine's fate after 1963 is undocumented.
Fred Niell (high-school student, home-built)c. 1993–1994
beam demonstratedUSA
RF frequency
resonance observed at 2.2–2.3 MHz — reported by the builder as roughly a tenth-order harmonic of the N₂ fundamental, not the fundamental itself
Particle
residual-air ions (mostly nitrogen)
Vacuum
small diffusion pump backed by a rotary pump
A cyclotron designed and built at home by a high-school student, set up in the room next to his bedroom, with every component — magnet, RF amplifier, vacuum gauges — of his own construction. Running on residual air rather than an isolated gas species, it demonstrated cyclotron mass resonance with a clear output peak at 2.2–2.3 MHz consistent with nitrogen ions and their harmonics. The associated science-fair project won the grand prize at the 1994 International Science and Engineering Fair.
Fred Niell (high-school student, home-built)1994–1995
beam demonstratedUSA≈70 keV
Field
≈0.67 T (6.7 kG), calculated from turns, current and gap; not measured
RF frequency
0.5–16 MHz variable; H⁺ resonance observed at 10 MHz
Particle
protons, helium ions
Energy
≈70 keV computed for H⁺ at the observed resonance (r = 5.7 cm); not measured directly
Vacuum
diffusion pump with molecular-sieve trap
A second, more research-oriented machine built the following school year, fixing the shortcomings of the first: a variable-frequency RF system (0.5–16 MHz) driven by a tube push-pull amplifier allowed resonance mapping of more than one ion species. Measured resonance peaks for hydrogen and helium matched prediction within about 2% once the magnet coil count was corrected, corresponding to roughly 70 keV protons. The machine was left at the school when the builder went to college and was later dismantled for parts.
Tim Koeth and Stu Hanebuth, undergraduates at Rutgers University1995–1999
beam demonstratedUSA
Pole diameter
229 mm (9 in), Varian V-3400 NMR magnet
Field
0.889 T (8.89 kG) operating; 1.2 T maximum
RF frequency
13.56 MHz
Particle
protons
Energy
≈184 keV computed at 7.0 cm radius (run 91699C); up to ≈50 nA proton current at the chamber periphery
The feasibility prototype for the later 12-inch machine, built around a repurposed Varian NMR magnet turned on its side, with a fixed 13.56 MHz industrial-band RF system setting the 0.889 T operating field. Successful resonance runs in 1999 showed a strong proton current at the predicted field, with beam detected at up to about 50 nA at the periphery. Its results justified moving to a 12-inch magnet with a 1 MeV design goal.
Undergraduate design classes led by Timothy Koeth, University of Marylandc. 2016–present
in progressUSA
Pole diameter
483 mm (19 in), restored historic magnet
Particle
protons (planned)
Energy
5 MeV design target per the builder's 2026 project history; no beam yet
A successor machine under construction by successive undergraduate cohorts at the University of Maryland, built around a restored 19-inch magnet alongside the relocated Rutgers 12-inch cyclotron. Unlike the 12-inch machine, whose beam stays inside the chamber, the 19-inch design is intended to extract beam for experiments; Terp Magazine reports that on completion it would be the largest cyclotron built by undergraduates.
Reed Michael Upson and Ray Anchordoquy (backyard workshop build)2018–present
in progressUSA
Dee diameter
356 mm (14 in)
Field
≈0.3 T (3 kG) median-plane
Particle
protons
Energy
60 keV target (revised down from 1.1 MeV, the builders citing incidental radiation)
Vacuum
10⁻⁵ torr target
A two-person build housed in a purpose-built, Faraday-caged garden shed, under way since 2018 and documented on a project site. The original 1.1 MeV goal was deliberately reduced to about 60 keV to avoid incidental radiation hazards. As of the site's last update in late 2024, subsystems including vacuum and the ion source were being validated individually and no beam had yet been produced.
Oyama KOSEN students (Athena club) with the AxeLatoon collaboration — KEK, RIKEN and university staff2021–present
in progressJapan
Design
CST Studio 3D electromagnetic field simulation and particle orbit calculation, done by the students before the chamber was designed
Size
palm-sized (手のひらサイズ)
Particle
protons (intended)
Dee
dee electrode and filament originally fixed by 3D-printed mesh plates, later redesigned once the accelerating field distribution was accounted for
Status
operating since January 2023; the internal detector registers a current the team attributes to particles, with first unambiguous ion-beam observation still pending
The first of the AxeLatoon student cyclotrons to reach operation. Twelve first-year students across mechanical, electrical and materials departments began design work in July 2021 with staff from the collaboration, simulating the field distribution and particle orbits in CST Studio before designing the chamber. Chamber fabrication began in spring 2022, with the filament and its drive circuit, the hydrogen system and the other peripherals prepared in parallel; pumps and the RF oscillator were connected around July 2022, and the machine began operating in January 2023. Repeated running has the internal detector registering a current the team attributes to particles, but the team is explicit that the first unambiguous ion-beam observation still needs fine tuning and time. The group also wrote the detailed RF tuning procedure — nulling reflected power — that the other AxeLatoon sites now follow. A deliberately-kept teaching moment: energising the filament before the chamber was evacuated burned it out in about a second and a half, in a very bright flash.
Inter-University Accelerator Centre staff, New Delhi — from a proposal by P. C. Deshmukh; begun under G. Rodrigues, led by R. Mehta after his retirement; T. Koeth (University of Maryland) advising remotely2024 – present
in progressIndia
Pole diameter
305 mm (design)
Field
1.2 T median plane (tender specification)
Pole gap
51 mm nominal (design)
RF frequency
18.2 MHz; bench-tested to 200 W, 2 kW CW amplifier in development
The census's first national-laboratory build: a teaching-scale proton cyclotron under development at India's Inter-University Accelerator Centre since 2024, from a proposal by P. C. Deshmukh that grew out of a 2023 conversation with IUAC's director — with Timothy Koeth of the Maryland 19-inch advising remotely, a documented design lineage from the Rutgers/Maryland educational machines. The paper trail is procurement-grade: a magnet tender specifying 1.2 T on 305 mm poles across a 51 mm gap, an annual report recording two fabricated chambers and a full Dee–dummy-dee RF system, and a conference paper reporting the RF bench-tested to 200 W at 18.2 MHz. Two design energies circulate (1 MeV in the Shaastra feature, 1.2 MeV in the CAMOST report) and neither is a measured beam: no accelerated or extracted beam is documented, and the director's own stated milestone is 'We first need to demonstrate beam extraction.' The feature reports medical-isotope and proton-therapy hopes and an exploratory industry partnership as future directions.
Projects the record shows stopping before, or after, a beam result.
Knox College Cyclotron
Jeff Smith, undergraduate at Knox College, Illinoisc. 1999–2001
stalledUSA
Magnet
the college NMR lab's nuclear magnetic resonance magnet, ≈20 cm diameter pole faces, maximum field 2 T
Design energy
1.5 MeV protons — the design point stated by the department, not a result
Dees
two copper dees on blocks of insulating dielectric, mounted so each could be moved independently
Chamber
0.48 cm thick brass; the top plate was held on by external air pressure alone, in the pattern Niell used
RF system
the NMR lab's radiofrequency sources and power amplifiers; dees operated at 3,750 V through a manually adjustable resonating circuit. The dees' capacitance was never measured — the circuit was tuned by trial and error
Ion source
coiled coated tungsten filament at the centre of the chamber, its electrons ionising low-pressure hydrogen; built to be completely removable because the builder judged it the design's weak point
Vacuum
15.2 cm (6 in) diameter diffusion pump; no pressure is published
Feedthroughs
nylon plugs, O-rings and brass screws held in place with Plumber's Goop; rubber stoppers drilled along their axes sealed the gas line and the Faraday collector
Extraction
an extractor system was designed around a negatively charged deflection plate; the machine also carried an insertable Faraday collector plate
Outcome
never successfully tested — the magnetic field moved the wires powering the ion source until they shorted the dees
A fixed-frequency cyclotron built as a senior honors project at Knox College, begun the summer before the builder's junior year on a Ford Fellowship and the Richter Fund, after he encountered cyclotrons at a Fermilab public lecture; Fermilab cyclotron builder Chris Olsen assisted. The department states the design point plainly: 1.5 MeV protons, using the 2 T magnet in the college's NMR lab along with that lab's RF sources and power amplifiers. Two independently movable copper dees sat on dielectric blocks inside a brass chamber whose top plate was held down by atmospheric pressure alone; the resonant circuit was tuned by trial and error because the dees' capacitance was never measured; a coiled tungsten filament at the centre ionised low-pressure hydrogen; and both a deflection-plate extractor and an insertable Faraday collector were built. It did not reach beam. A later undergraduate builder at Houghton College, surveying the amateur machines before starting his own, records the failure mode: the magnetic field moved the wires powering the ion source until they shorted the dees. The department's page says he “came very close to seeing beam before graduation rudely cut his project short” and that he left the machine ready for another student to finish; a 2010 account reports it still in the college basement, with one attempt at revival abandoned. The builder went on to a Cornell physics doctorate and to SLAC.
Heidi Baumgartner and Peter Heuer (high-school students), with German Diagama; built at Jefferson Labc. 2006–2013
stalledUSA
Dee diameter
305 mm (12 in) single dee with grounded dummy dee
Field
1.6 T (16 kG) design target; the magnet was never power-tested
RF frequency
24 MHz design, 3 kW tube amplifier (a PlasmaTherm unit retuned from 13.56 MHz)
Particle
protons
Energy
2 MeV (design)
Vacuum
10⁻⁷ torr (design)
Two high-school students conceived a 2 MeV proton cyclotron in 2006 and, after a funding email reached an associate director of Jefferson Lab, were invited to build it at the national lab; construction began in 2009 and ran over several summers. All components — a 1.6 T H-magnet, welded stainless chamber, 24 MHz RF system, and thermionic ion source — were completed, but lab safety rules for minors prevented power testing, and no beam was ever run. The machine was transferred to Old Dominion University to await further student work; both builders went on to physics careers.
James Kwolik, St. Cloud, Minnesotaby 2015 (construction years undocumented)
stalledUSA
An unfinished build documented by the builder's own 2015 photo tour, preserved in this site's library: a 9.25-inch-OD machine with a 1,080 lb magnet carrying commercially machined low-carbon-steel pole tips chamfered for horizontal focusing at 1.2 T, an oil-cooled high-power RF matching network targeting 10–18 kV peak-to-peak on the dees, and a rotary-vane-plus-diffusion-pump stack with an LN₂ cryotrap the account claims reached 10⁻⁹ torr. The machine still needed its 3 kW RF supply and was in storage — which is where the documented record ends. Indexed 2011 construction posts exist on social platforms but could not be retrieved for verification, so the construction years stay undocumented. The serious-hardware, no-RF ending is one of the census's recurring shapes: the magnet is the romance, the RF chain is the wall.
Student-led team of roughly ten CEGEP and undergraduate students (“Engineering a Mini Cyclotron Crew” / BICEPs), Dawson College, Montreal2023–2025
stalledCanada
Project
BICEPs — “Bi-Institutional Cyclotron for Experimental Physics”; the college foundation listed it as the “Engineering a Mini Cyclotron Crew” (EMC²). The idea originated as a CERN Beamline for Schools proposal under DawsonHEP and was revived in fall 2023
Magnet
recovered laboratory electromagnet; now sitting unused in a McGill University physics lab
Design dee diameter
300 mm
Design field
1 T
Particles (design)
protons and H₂⁺, from ionised hydrogen
Research output
a betatron-motion and radiation-shielding study, presented at the Canadian Undergraduate Physics Conference 2024 and published in DrJES vol. 15 (Winter 2025)
Outcome
wound down in 2025 as members moved on to university; no construction milestones were published
A student initiative at a Montreal CEGEP — BICEPs, the “Bi-Institutional Cyclotron for Experimental Physics”, known to the college foundation as the “Engineering a Mini Cyclotron Crew” (EMC²) — that set out in fall 2023 to build a compact cyclotron and wound down in 2025 as its members moved on to university. The design called for 300 mm dees in a 1 T field, accelerating protons and H₂⁺. The team presented a study of betatron motion and radiation shielding at the Canadian Undergraduate Physics Conference in 2024 and published it in the college research journal in 2025; the recovered laboratory electromagnet now sits unused in a McGill University physics lab. Project history and status per correspondence with a project co-author, August 2026.
Projects whose public record stops before any beam result is stated. They may have succeeded, failed, or never been finished; the sources do not say.
El Cerrito High School Cyclotron
Four students with physics teacher Ben Siegel, El Cerrito High School, California1947
outcome unknownUSA
Magnet
≈0.5 t welded soft-steel double yoke; ≈1 t machine
RF frequency
20 MHz (reported), ≈1.5 kW
Particle
protons
Energy
“million-electron-volt” claimed in period press; unverified
The earliest well-documented school-built cyclotron, constructed in about three months for roughly $500 and featured in Popular Science in 1947 (and, per a later account, in Physics Today). The students operated the machine by remote control from a plywood panel in a closet-sized room, and the Research Corporation waived royalties on the Lawrence cyclotron patents it infringed. Contemporary press reported the machine running at 20 MHz, but no beam measurements survive, so the million-electron-volt figure should be read as a period claim.
W. H. Bancroft, Jr. and Howard College (now Samford University), Birmingham, Alabama — a faculty-led in-house build1948 – completed by 1959; later scrapped
outcome unknownUSA
Pole diameter
406 mm (16 in), 4 in centre gap
Field
15–16 kG (≈1.5–1.6 T) design maximum, air-cooled coils
RF frequency
not stated; four 304TL tubes, 3 kW maximum output (design)
Particle
protons (intended)
Energy
4 MeV design target; no beam on record
Ion source
Oak Ridge type (design)
A faculty-led college build begun in 1948 at Howard College in Birmingham, Alabama — the institution that became Samford University — and still listed as under construction when its builder filled in the June 1958 world-census data sheet, ten years on. Every number on the sheet is a design figure: a 16-inch, 12-ton magnet (10 tons iron, 2 of copper) at 15–16 kilogauss with air-cooled coils, a 21-inch dee driven by four 304TL tubes at up to 50 kV dee-to-dee, an Oak Ridge-type ion source, iron-ore blocks for shielding, and a 4 MeV proton target, at an estimated total cost of $120,000 with most materials donated by local industry. The machine was completed: in 1959 the college paper's science editor described it as operational — the first cyclotron built in Alabama — and after the college's 1957 move it was relocated to the Homewood campus into its own building, where per a Samford physicist it was eventually scrapped; a 1955 photograph survives in the university archives [record extended 2026-09-12 from the Samford campus-history account]. No performance record of any kind has been found — the 1958 sheet's performance page is blank and the 1959 account is a student paper's description, not a measurement — so whether the machine ever produced beam is undocumented and its outcome is recorded as unknown.
Three high-school students, Fort Wayne, Indiana1951
outcome unknownUSA
In 1951 three students remodeled their high-school basement into a cyclotron laboratory and were already at work on the accelerator by the time the principal found out. The group kept careful engineering notes that later amateur builders cite as an early blueprint, and two of the students went on to work at Fermilab. No surviving documentation confirms whether the machine ultimately accelerated ions.
A press photograph is the whole surviving record: on 10 March 1955 a Bettmann photographer shot 15-year-old Carl Svoboda 'performing an experiment with a cyclotron (center) he built himself' for about $300 — 'a far cry from the $9,500,000 betatron at the University of California, but an atom-smasher just the same,' the caption says. Contemporary photographic documentation of a completed apparatus, and nothing more: no species, no beam result, no later history. A reader's comment on a 2020 accelerator-history article points to a surviving personal archive — a lead, not a record. Distinct from the El Cerrito High School build across the bay eight years earlier.
Successive student teams (L. Zuckerman, N. Ostroff, J. Horowitz and others; later S. Goldwasser and others), Central High School, Philadelphia1958 – c. 1970
outcome unknownUSA
Pole diameter
178 mm (7 in)
Field
≈2 T (20 kG), a builder’s later estimate; never measured
RF frequency
≈7 MHz initially, later ≈20 MHz
Particle
protons (intended)
Energy
≈1 MeV theoretical; no measured beam
A cyclotron built and rebuilt by roughly three generations of Philadelphia high-school students between 1958 and about 1970, documented in a detailed retrospective by one of the later participants. The 7-inch machine was theoretically capable of about 1 MeV, but the account itself doubts a real proton beam was ever achieved; the best evidence of operation was a marginal exposure of dental X-ray film.
Reinier Beeuwkes III, 17, Newton High School, Massachusetts1958
outcome unknownUSA
The machine that won the 1958 Westinghouse Science Talent Search — and the census's one documented electron cyclotron. The official winners booklet lists 'Design and Construction of an Electron Cyclotron'; Popular Electronics ran the photograph that July: 'He designed and built this cyclotron for accelerating electrons to high speeds. It cost less than $150, and is only as complex “as a radio,” he says, but it netted the 17-year-old Newton, Mass., lad a $7500 science scholarship.' TIME's account gives $121. No operating figures, achieved energy, or technical description survives in the located record — a classical cyclotron geometry accelerating electrons stays nonrelativistic only to a few keV, and what the machine actually demonstrated is exactly what the missing technical account would have said. Status is therefore unknown, on a machine good enough to win the country's top school science prize.
Graduate students and faculty, SNU physics department — the official account credits Kwon Young-dae and his students1959–60
outcome unknownSouth Korea
Korea's first domestically built accelerator, by the official account: the Korean Academy of Science and Technology's merit biography of physicist Kwon Young-dae states that in 1959 'Korea attempted for the first time to independently manufacture a cyclotron, and a proton beam was successfully extracted.' That extraction claim is a retrospective biography sentence, not a technical record — no measurement, collection point, or commissioning account survives in located sources, so the claim is reported here as the biography's and the status stays unknown. A 2025 builder memoir (mirrored online, not independently verified) describes a 20 cm chamber and a 2.5-ton magnet; a 1961 paper title reportedly gives the machine's designation as 1.5 MeV, and a 1962 companion paper on its vacuum chamber says 1 MeV — designations, not measurements. A funded graduate project rather than a hobby build; it enters this census the way the IUAC machine does, as a documented student-built machine with its scope stated plainly. The biography's own verdict: the machine was small and did not lead to sustained research, but showed a generation of students that experimental physics could be done in Korea.
Student Cyclotron Committee, Stuyvesant High School, New York Cityearly 1960s
outcome unknownUSA
Stuyvesant students organized a Cyclotron Committee with specialized teams for the electromagnet, vacuum chamber, and RF system, collecting donations that included a half-ton of steel and five miles of copper wire at a project cost of roughly $10,000. A 1962 school publication suggested the machine was completed, but a retrospective in the school's own newspaper found no evidence the project actually reached completion.
Albert Swank Jr., 17, with his father, Anchorage, Alaskac. 1967
outcome unknownUSA
The teenage build behind the census-adjacent story everyone knows. Albert Swank Jr. reached national news in December 2005 for arranging to move a 20-ton Johns Hopkins medical cyclotron into his Anchorage home — and NPR's report notes, almost in passing, that the family building 'is the same building where he built a smaller cyclotron when he was 17,' his father working 'side by side with me while I was building my first cyclotron.' Swank was 55 at the time of the report, which puts the build around 1967 — an inference from the reported ages, not a documented date. Nothing else survives in the located record: no specifications, no beam result. The two machines are recorded separately — this entry is the 1960s build; the 2005 relocation project has its own record on the world map.
An undergraduate thesis project documenting the design and construction of a cyclotron intended to accelerate protons to 2 MeV for low-energy beam experiments in MIT's nuclear engineering department. The thesis and two videos are the surviving public record; no demonstrated beam is documented, and the project appears not to have been completed.
permanent magnet beneath the chamber, ≈¥150,000 — half the budget
Chamber
self-designed, fabricated to order (≈¥100,000); the first design used a non-standard vacuum flange spec, found only after fabrication
Ion source
hot tungsten filament ionising hydrogen admitted to the evacuated chamber; the filament is threaded through a 0.45 mm hole hand-drilled in 0.8 mm copper wire under a watchmaker’s loupe
Dee
single hollow semicircular brass dee
RF system
secondhand, sourced from online auctions and eBay; final hardware suggested by an amateur-radio friend
Detection
metal rod intended to collect accelerated protons and read them as a current; no reading has been published
Particle
protons
Cost
≈¥300,000 total (roughly US$2,000–2,800 at 2019 rates)
A tabletop classical cyclotron built in an apartment living room by a theoretical particle physicist with no experimental or engineering training, who bought a lathe and a milling machine and worked by trial and error — reading papers and manufacturer catalogues to choose hardware, and buying most of it secondhand. Machine shops initially refused the vacuum chamber job, suspicious that no individual would build such a thing, before one agreed; the finished chamber then turned out to use a non-standard flange spec. The builder is explicit that the machine is not for experiments: building it was the point, and he chose a cyclotron over a fusor because fusor instructions are already all over the internet, leaving no trial and error. He exhibited it at Maker Faire Tokyo in 2019, where a RIKEN colleague explained the physics to the press, and states that he designed, built and got the machine running. No beam energy, current, field or frequency has been published, so what it achieved is undocumented. He went on to build a Wideröe-type linear accelerator in the same room — the machine he exhibited at Maker Faire Tokyo in 2022 — and his machine became the seed for a national student cyclotron programme (see AxeLatoon, below). He is a researcher at RIKEN, and the programme's own record has him introducing the home cyclotron to first-year students at Oyama KOSEN in November 2021 and the home linac at Nagano KOSEN a month later. Both posts describe him presenting the machines and taking questions, not running them; no measurement from either occasion is published.