Several design measures were taken to minimize the effects of these conditions. The detector was mounted inside a canister packed with an annulus of foam cushioning (polyurethane). The detector rested on a 1/4-inch, foam-rubber-cushioned, metal support ring. The ring itself was spring suspended inside the canister. The top of the detector dewar was tightly packed using foam cushioning against the top lid of the canister. The collimator cone was suspended from the bottom of the dewar, so that the collimator and dewar (with protruding detector) moved as a unit. The eanister was hung from a yoke, hinged to allow the canister to remain vertical as the boom was pushed out to its 20-degrees-from-vertical position. A latch-plate locking pin arrangement was used to secure the canister to the boom at all times other than while the canister was elevated to take a measurement. The IMP data acquisition portion of the cab was air conditioned, using a roof-mounted unit. The air conditioner was operated at full cool continuously. The IMP electronics, computer and printer were mounted in a standard instrument rack which was shock-mounted inside the IMP, The scheme was successful in allowing use of laboratory equipment for the field operation. Maintenance experience was similar for the IMP computers and the laboratory computers. The pulse height analyzer (PHA) was designed for field use and had a satisfactory maintenance record. Rain stormsinitially caused problems because of water penetration at cable connectors. These were solved by providing a flexible rubber boot over the cables at the canister entrance and at the feedthrough in the rear of the cab. In addition, the feedthrough was recessed inside the cab wall and partially protected by a door. It was also necessary to put thermal insulation around the cab side of the feedthrough plate to avoid condensation problems. The dew point at Enewetak is usually greater than 80°F, so a cold connector(i.e., less than 80°F) caused condensation. The Onan electrical generator was modified to increase its reliability. The fuel pump was changed to an electrical fuel pump. An oil bath air cleaner was installed. A water separator and an improved filter were installed in the gasoline feed line. One of the Onans operated for about 4,400 hours, which is a factor of two longer than the normallifetime. Rust prevention was also a design concern. The inner and outer surfaces of the sheet metal forming the IMP body were coated with zinc chromate primer. Outer surfaces then received a coat of white acrylic enamel. Inside surfaces were coated either with Glyptal varnish or commercial undercoating material, depending on the location. These initial measures were combined with a maintenance "grind and repaint" program. There was considerably less deterioration on the IMPs than on other equipment on-atoll that did not receive this sort of attention. 3.4.2 Maintenance Scheme Critical spare parts and replacement components were maintained on-atoll These included an Onan electrical generator, air conditioners, air compressors, the extendable mast, and spare parts for the IMP engine and Onan engine. Electrical spares included the cable harness and detector power supplies. Mechanical and electronic maintenance was performed by the two-man IMP technician erew, assisted by the two Air Force mechanic-drivers assigned to the IMP. Occasionally the base operating contractor's vehicle maintenance shop provided a special service, such as welding. The required spare parts and components were selected based on general experience, modified by on-atoll history. Replacements were ordered as parts were used from the spares inventory, or as failure required a part. A regular maintenance schedule was established and usually adhered to. week was usually devoted to maintenance for each operating IMP. One day per six day work This was modified based on urgency of survey schedules, and further modified depending upon transportation needs; that is, if an IMP were working a remote island not served by a military work boat, and required one or two days to complete the survey, the survey would be completed and maintenance delayed until the third day. On the average, two IMPs were available about 80 percent of the time. available about 95 percent of the time. 105 At least one IMP was