Construction Classics

RF power engineering

Three reports on the same question from three directions: how do you actually drive the dees? One rebuilds a working amplifier around a then-new ceramic tetrode, one argues the accelerating waveform should not be a sine at all, and one drives three dees from three-phase RF. Together they cover the tube, the waveform, and the topology.

Why these documents can be hosted

All three are AEC contractor reports — Oak Ridge under W-7405-eng-26, Berkeley and Colorado under their own AEC contracts — and AEC contractor reports are treated as US Government works, which carry no copyright. Pre-1978 distribution without a copyright notice is an independent public-domain basis. Each scan carries the US Government sponsorship disclaimer, ORNL-2403 is marked Unclassified with its AEC distribution list bound in, and none of the three carries a copyright notice, publisher imprint, or journal-reprint marking. Verified against the scans and the OSTI records; determinations made August 2026.

The tube: rebuilding a power amplifier

COO-535-543 is the most directly useful of the three for anyone contemplating an amplifier upgrade, because it is a retrofit report rather than a clean-sheet design. The University of Colorado 52-inch cyclotron already had a working RF system; this describes replacing the power amplifier with a then-modern ceramic-and-metal tetrode and reports the operating experience afterwards.

The engineering argument is simplification, and it is measured in operator actions rather than decibels. After the rebuild, excluding the dee resonant circuit, exactly one circuit in the RF system still required remote tuning. Once the operator sets the frequency and the dee-voltage control, the dee voltage is turned on and off with two push buttons and needs no further attention; a frequency change takes one to ten minutes, depending only on how far the shorting bar must travel. The dee-voltage control includes current limiting specifically to stop inexperienced operators from destroying the tubes — a design goal worth copying in any machine that students will run. Table 1 lists the system parameters.

Scan quality note: this document carries an explicit “portions may be illegible” disclaimer, and its cover pages are badly degraded with a date stamp overlaying the title. The body text and its OCR layer are sound from the summary onward.

The waveform: why not a sine?

ORNL-2403 is the outlier of the collection and the most interesting to a modern reader. Goodman’s claim is that the conventional sinusoidal accelerating voltage is simply the waveform that resonant circuits produce most easily, not the one the beam wants — and that replacing it with a square wave would reduce beam loss to electric defocusing. In AVF machines he argues it would additionally buy greater tolerance on average field accuracy and possibly better beam quality: a waveform change substituting for magnet precision.

The practical obstacle is that high-Q systems resist producing anything but a sine, and the report devotes a brief analysis to that problem before presenting two systems that approximate a square wave by adding first and third harmonics, with parameter tables for each and measured waveforms from one of them. Read beside the tetrode report, it reads remarkably like a 1957 anticipation of switching-amplifier thinking, arrived at from beam dynamics rather than from power electronics.

The topology: three dees, three phases

UCRL-3153 documents three cloverleaf cyclotrons built with three dees excited by three-phase RF. Smith covers the phase generator that produced the three-phase drive, the servomechanisms and phase-control equipment that kept the dees in tune, and — the part worth the reading — the relationship between the resonator’s electrical characteristics and the control problem. Multi-dee machines are not a tabletop proposition, but the coupling between resonator design and control stability is the same problem a single-dee builder meets when the amplifier and the tank fight each other. The systems were built in 3 kW and 37 kW sizes, and the report notes the approach extends to megawatts.

Design rules extracted from these documents

36 Design Guide rules come from these three reports — COO-535-543, ORNL-2403 and UCRL-3153. They were chosen to correct a balance — plenty of resonator craft in the collection, little on amplifier topology and protection — and most of what they yielded is exactly that. Representative examples:

• dg-720 — the dee must be a high-Q energy-storage resonator, never a switched load
• dg-718 — squaring the dee waveform with a 1/3-amplitude third harmonic attacks weak-focusing beam loss
• dg-711 — let the current-limit reference track plate voltage so dissipation, not current, is held constant
• dg-730 — unneutralized inter-electrode capacitance couples the control loops
• dg-717 — be wary of vacuum capacitors in the high-power PA plate circuit
• dg-716 — for sliding RF contacts use heavy contact fingers — twice normal finger-stock thickness

Used by the dee RF coupling deep dive. The 20-inch machine built on UCRL-3153’s topology is hosted as UCRL-3187.