Bowin 



It is undesirable to subject electronic components to ambient 

 temperatures in excess of about 90°F. Because temperatures this 

 high, and higher, are relatively common in many parts of the world 

 throughout much of the year, it is necessary to furnish air- 

 conditioning for most computer systems. At the time of installation 

 of our original system, the CHAIN was not air-conditioned. It was 

 therefore necessary to build a room around the system in the main 

 laboratory of the ship. Three water-cooled air-conditioners were 

 used for cooling the room. This proved completely satisfactory. 

 Even though difficulties with one or more of the air-conditioners 

 was undesirably common, at least one of them was always operating. 

 We were thereby able to maintain the computer at its required 

 temperatures. At the same time as the system was expanded in 

 1963, air-conditioning was supplied to the laboratories, living 

 quarters and messrooms of the CHAIN. Following that time, we 

 dispensed with the room in the main laboratory that had previously 

 sheltered the computer. Although we were no longer directly 

 responsible for air-conditioning, it nevertheless required the 

 suspension of operations whenever the ship's main compressor unit 

 failed. This experience suggests that some additional independent 

 air-conditioning capability should be provided for the computer 

 system if the ship itself is generally air-conditioned, or that m.ore 

 than one air-conditioning unit should be available in any event. 

 High hiimidity has not been identified as a difficulty in our past 

 systems. The only problem encountered occurred when water 

 actually dripped on the components of the system from either an 

 overhead vent or from condensation upon the cooling ducts in the 

 laboratory. Both these troubles have been corrected; one by 

 sealing off the overhead vent and the other by coating the ducts with 

 an insulating material. 



Vibration is another concern of most people who are contem- 

 plating utilizing a computer aboard a ship. Again this has not been 

 a serious problem in our installation. We have attributed only one 

 failure directly to vibration when it was believed that a low voltage 

 protection relay may have been erroneously actuated by vibration. 



The central processor of the IBM 1710 computer system proved 

 extremely reliable. In the years of operation we had only three 

 failures in the central processor, and two of these were caused 

 when the IBM maintenance man accidentally shorted printed circuit 

 cards. 



One failure of a printed circuit card in the analog to digital 

 converter demonstrated a weakness of equipment diagnostic 



257 



