Construction Classics

The 22-inch test cyclotron

Most reports in this collection are a snapshot: one machine, one document, written when the work was finished. These seven are something rarer — consecutive progress reports that follow a single small cyclotron over three and a quarter years, from resolving its first proton orbits to being rebuilt, renamed, reassembled and proposed for conversion again. Nothing was tidied afterwards, because nobody knew yet how it would turn out.

Why these documents can be hosted

All seven are Oak Ridge National Laboratory division progress reports prepared by Carbide and Carbon Chemicals Company, a division of Union Carbide and Carbon Corporation, under AEC contract W-7405-eng-26. AEC contractor reports are treated as US Government works and carry no copyright; pre-1978 distribution without a copyright notice is an independent public-domain basis. No copyright notice, publisher imprint or journal-reprint marking appears in any of the seven.

Three checks worth recording. The covers carry distribution controls, not classification — ORNL-1531 states “This document consists of 32 pages, Copy T of 114 copies, Series A”, which is an AEC distribution count of the same kind already noted on ORNL-1339. ORNL-1795 is explicitly headed “Part I — Unclassified Work”; only that part is offered here, and any Part II is a separate document that this collection does not hold. And the reports’ own reference sections list division papers “submitted to” journals, which are citations to other work rather than markings on these reports. Verified by reading the scans; determinations made August 2026.

What a test cyclotron was for

Oak Ridge in these years ran the 86-inch and the 63-inch as production and research machines. The 22-inch existed to answer questions on their behalf — cheaply, and where breaking something did not cost a research programme its beam time. ORNL-1884 states the purpose plainly: it is used in “a continuation of the study of problems associated with the acceleration of large ion currents.”

That is why this seam reads unlike a construction report. A test machine gets rebuilt whenever the question changes, so the documents record decisions, revisions and reversals that a finished-machine report would compress into a specification. It is also why the machine has three names: it was a 22-inch through 1952, became the 44-inch when its magnet was reworked, and by March 1955 there was a proposal to convert it again into a 48-inch heavy-ion machine.

The rename, and why it matters for searching

ORNL-1531 hangs a numbered footnote on the phrase “The 44-in. cyclotron”: Formerly referred to as the 22-in cyclotron. One line, and it is the key to the whole seam — without it the 1951–52 reports and the 1953–55 reports look like two different machines. Anyone searching this literature by machine size will split one continuous record in half. The lesson generalises: in a laboratory’s own progress reports, machines are named by their current pole diameter, so a rebuild renames the machine.

The four things worth stealing

Trade energy for current with a spacer. The rebuilt machine was designed to run at two radii: a removable 14½-inch spacer between the vacuum tank and the faceplate moves the dees back from the field centre, so the same cyclotron runs at about 1.5 MeV for high-current test work or about 5 MeV with the dees in close (ORNL-1663, ORNL-1670). One mechanical part, two machines. A high-current target can sit at the dee edge at either the 11-inch or the 20-inch radius.

Grind the poles flat before you shim. ORNL-1670 has the crew going over the magnet pole faces with a portable grinder, aiming at ±0.01% field uniformity — and states the reason: to “provide a standard base for the various magnetic shim designs that may be tested.” You cannot evaluate a shim against a baseline that is itself uneven. The order of operations is the rule, not the tolerance.

And the seam tells you how that went. Two reports later, after a further half-year of grinding and shimming the tank walls, ORNL-1795 reports the field uniform to within 0.05% — five times the ±0.01% they had set out for. Both numbers are worth carrying: the aspiration, and the fact that a national laboratory with a machine shop missed it by 5× and kept working. A single report would have printed only one of them.

Measure the resonator before designing what feeds it. ORNL-1663 reports every RF component designed except the filament-coupling circuit, because that circuit “depends upon the electrical characteristics of the resonant dee system” and “cannot be designed until these characteristics are determined.” Naming the one thing you are not yet able to design, and why, is better engineering practice than guessing at it.

Choose what to contract out, then expect it to bite. ORNL-1670 records the split: liner, dees and dee-stem housing to an outside contractor; faceplates, dee stems, ion source, target probe and vacuum system made in-house. ORNL-1884 records the consequence — the dee-stem housing liner was late because of brazing errors made by the contractor. Both halves of that story are in the collection because progress reports print the second half; a final report rarely does.

Read them for the admissions

The seam’s real value is tone. ORNL-1884 says the electrical work “progressed as rapidly as the availability of funds and the status of other work in the Division would permit” — a sentence no brochure contains. Percentage-complete figures appear and then have to be revised. A machine is assembled, vacuum-tested, and still has “some work remaining on the ion source, the handling equipment, the oscillator auxiliary equipment, and magnetic field shimming.” For a builder judging whether their own project is going badly, a national laboratory’s quarter-by-quarter account of a small machine taking three years is more useful than any finished specification.

About these copies

All seven carry usable OCR text layers. One caution that applies to the whole seam: these reports are set in two columns, and the text layer interleaves them. Extracted prose will splice a sentence about the 44-inch to an unrelated sentence about the 86-inch a few inches to its right. Read the page images, or extract with layout preserved; a quotation that reads oddly is probably two quotations.

ORNL-1795 is served from files.cyclotroninfo.com rather than from the site, because at 29.7 MB it exceeds the 25 MB per-file limit of the site’s host. Its size sits in seven halftone pages — the dee-stem standing-wave plots, the RF joint, target assemblies. Lossless re-compression was measured and reached only 27.0 MB, so it is hosted whole and unrecompressed, as ORNL-3540 and TID-454 are.

Design rules extracted from these documents

39 Design Guide rules come from these reports. Representative examples:

• dg-926 — resolve individual turns with a 0.020-inch tantalum wire probe scanned from 1.2 to 11.5 inches
• dg-927 — expect a spurious rise in wire-probe current with radius — the wire heats and emits thermionically
• dg-950 — flatten the base field before testing shims
• dg-955 — expect achieved flatness to land short of the grinding aspiration
• dg-943 — sequence RF design around measurement
• dg-951 — outsource castings and weldments at your peril; keep leak-integrity parts in-house
• dg-964 — a major rebuild of even a small, staffed machine runs about two years

Continues directly from the commissioning practice cluster, whose two 1952 quarterlies sit inside this run. See also magnet design for shimming, and ion sources for the DC-injection work the 22-inch was built to test.