Introduction
The IPNS accelerator system consists of an H- ion source, a Cockcroft-Walton preaccelerator, a 50 MeV Alvarez linac, a 450 MeV Rapid Cycling Synchrotron (RCS), transport lines and ancillary subsystems (controls, diagnostics, services). Figure 1 shows the layout of the IPNS accelerator, including linac, RCS and spallation target. The accelerator normally operates at an average beam current of 14 to 15 uA, delivering pulses of approximately 3 x 1012 protons at 450 MeV to the target, 30 times per second. The preaccelerator and linac entered service in 1961 and served as the injector to the 12.5 GeV Zero Gradient Synchrotron (ZGS) high-energy physics accelerator until it was shut down in 1979. The RCS was developed and constructed in the mid 1970's as a proposed booster for the ZGS. However, with shutdown of the ZGS imminent, these plans were dropped and instead the RCS was used initially (1977-1980) to provide beam to the ZING experimental target and in 1981, began providing beam to the present IPNS target.
The Ion Source and Preaccelerator
The H- ion source and associated equipment are housed in the terminal of a 750 kV Cockcroft-Walton preaccelerator. The H- ion source is a magnetron type in which negative ions are extracted directly from the hydrogen plasma on the surface of the source cathode. The extractor electrode and magnet poles are at terminal ground potential and the source itself, including the pulsed arc supply, pulsed hydrogen gas supply, and cesium supply (cesium greatly increases the H- generation) are pulsed to a negative 20 kV potential. The H- beam is extracted, bent 90o (to remove electrons from the H- beam) by a magnetic dipole, focused by a set of three quadrupole magnets and injected into the high-voltage column of the preaccelerator. The preaccelerator produces approximately 30 mA, 750 keV, 70 microsecond pulses at a repetition rate of up to 30 Hz.
The Linac
The linac is a copper-clad-steel structure 0.94 m in diameter and 33.5 m long. The linac was constructed in eleven sections which are bolted together. It contains 124 drift tubes, each with a dc quadrupole magnet. The magnets are divided into 12 series groups and powered by 12 dc power supplies located on the service floor. Transistorized shunts are attached to each of the first 58 magnets, allowing remote control of individual magnets. Nominal vacuum level in the linac is 2-3 x 10-7 torr, maintained by 5 ion, 1 turbo-molecular and 4 cryo pumps. The linac is water-cooled with a closed-loop system which is temperature regulated to within 0.2oF to keep the cavity tuned during normal operation to 200.07 MHz ±1 kHz. The 200 MHz pulsed rf power is obtained from a 4-stage amplifier; the output stage is a 7835 triode with a normal operating level of 3 MW and a peak-power rating of 5 MW. The 50 MeV beam exiting from the linac is about 1 cm in diameter; the pulsed current is about 10 mA; the 70 microsecond pulses can be delivered at a repetition rate of up to 30 Hz.
The Rapid Cycling Synchrotron (RCS)
The RCS is a strong-focusing, combined-function synchrotron that accelerates the beam from an input energy of 50 MeV to a final energy of 450 MeV. It is a six-period machine with a magnet structure of DOOFDFO and a circumference of 42.95 m. The ring magnets, part of a biased 30 Hz resonant circuit driven from twin solid-state power supplies, generate a magnetic field from 0.28 to 1.0 Tesla so that the beam orbital radius remains constant during the acceleration. Two pairs of sextupole and one pair of quadrupole magnets, powered by 30 Hz programable power supplies, provide betatron tune correction and manipulation.
H- stripping injection, pioneered on the Booster I experiment for the ZGS, is accomplished on the RCS with a carbon stripper foil located on the inside radius of a long straight section (L-1) outside the limit of the circulating proton beam. The equilibrium orbit is deformed in the injection region into the foil by a series of three small, pulsed "bumper" magnets. The H- beam is injected through a singlet ring magnet so that at the stripper foil, its path matches the deformed orbit. During injection, the bumper magnet current decays at a controlled exponential rate, moving the closed orbit away from the stripper foil and uniformly filling the horizontal aperture.
The proton beam is accelerated by two ferrite-loaded coaxial cavities. Moving at 2.2x108 m/s (almost 75% of the speed of light), the 450 MeV protons circle the ring in just under 200 ns (2x10-7s) just prior to extraction.. The beam is bunched such that it fills just over a third of the circumference, giving a bunch about 70-80 ns long with about a 120 ns gap between head and tail. The accelerated beam is extracted in a single turn by two ferrite-loaded kicker magnets and two septum magnets, one pulsed and one dc. The extracted beam, 450 MeV, 70-80 ns pulse, peak current ~ 12 A, is then transported to the neutron-generating target. The high-energy neutron pulse is correspondingly short, but the moderation process spreads it out a bit, resulting in the few microsecond pulse that is seen at the instruments.
Accelerator Operations
Three performance measures for the IPNS accelerator are operating hours/year, system availability, and current delivered to target. Operating hours on IPNS have historically been funding-limited rather than machine-limited. Figure 2 shows the operating record since the facility began operation in 1981. Government funding for user facilities saw a step increase with the Scientific Facilities Initiative in FY96, allowing IPNS operation to increase over 50% from its low in 1982-84. Funding permitting, IPNS could operate as much as 30 weeks (over 4500 hours of beam-on-target) per year. Although beam current is primarily a machine-limited parameter, it is the engineers' and technicians' fixing of faults and improvments to the hardware, and the skill of the operators in "tuning", that has allowed what was initially a 10-12 uA machine to consistently achieve average currents of 14-15 uA.
System reliability (delivering beam when scheduled) is a very important performance measure from the user's perspective. At the conclusion of each "run" (2-4 week period, during which beam is scheduled to be available to the users 24 hours a day), reliability is calculated as the ratio of hourss-available to hours-scheduled. Accelerator reliability and average current since 1981 is plotted in Figure 3, one point for each scheduled run. For the last ten years, rarely has reliability for a single run dropped below 80%, and yearly averages are close to (or exceeding) 95%. During this same period, IPNS target-system reliability has hovered around 99.5% except for approximately once every five years when the target reaches end-of-life and must be replaced. Figure 4 shows the accelerator and target-system reliability yearly-averages since 1982, and the combined reliability which is the important parameter for users to IPNS.
Prior to 1983, current was limited by the ion source. Since then, it has been ultimately limited by "beam loss", the fact that as more beam is injected, the fraction of the beam that gets "spilled" in the accelerator reaches some limit (determined usually by radiation levels or local heating of components) that it is not prudent to exceed. A can be seen in Figure 3, the maximum current has been relatively constant for a number of years, as expected because the basic machine parameters have been constant over that period. With the exception of one run in 1991 (where low current was required for target studies) and much of 1994 (where a damaged septum magnet limited operation to about 10 uA), yearly average current has exceeded 14 uA with some runs as high as 15 uA.
V. Stipp, F. Brumwell and G. McMichael, "The ANL 50 MeV Injector - 35 Year Anniversary", Proceedings of the 1996 Linear Accelerator Conference, Geneva, Switzerland, Aug. 26-30, 1996.
J.C. Dooling, F.R. Brumwell and G.E. McMichael, "The IPNS Accelerator 50 MeV and 500 MeV Transport Lines", Proceedings of the 1997 Particle Accelerator Conference, Vancouver, Canada, May 12-16, 1997.
J.C. Dooling, F.R. Brumwell, M.K. Lein and G.E. McMichael, "A Real-Time Energy Monitor System for the IPNS Linac", Proc. 2000 Linear Accelerator Conference., Monterey, CA, August 20-26 2000 (to be published).
L.I. Donley, V.F. Stipp, F.R. Brumwell, G.E. McMichael, "Reliability History and Improvements to the ANL 50 MeV H- Accelerator", Proc. 2000 Linear Accelerator Conference., Monterey, CA, August 20-26 2000 (to be published).