Editing UNIVAC I (section)

===Reliability===
Eckert and Mauchly were uncertain about the reliability of digital logic circuits—little was known about them at the time.  The UNIVAC had been designed with parallel computation circuits and a [[statistics|statistical]] comparison of the results.  In practice, however, only failing ''components,'' i.e., the vacuum tubes, yielded comparison faults, as the circuit designs as such proved very reliable. A regimen was established to ensure the reliability of the fragile vacuum tubes, the [[choke point]] of the entire operation. Prior to use large lots of the predominant tube type [[25L6]] were burned in and thoroughly tested. (Often half of any given production lot would be thrown away.) Technicians would then install a tested and burned-in tube in an easily diagnosed location such as the memory recirculate amplifiers. Then, when further proven aged and proven reliable, this "golden" tube was sent to stock to be pulled out for difficult-to-diagnose logic positions.

Furthermore, it took approximately 30 minutes to turn on the computer—all [[cathode heater]] power was stepped up gradually in order to reduce the in-rush current the concominant thermal stress on the tubes. As a result of these measures, uptimes ([[Mean time between failures|MTBF]]) of many days to weeks were eventually obtained on the processor. (The UNISERVO did not have vacuum columns but rather springs and strings to buffer the tape from the reels to the [[Capstan (tape recorder)|capstan]]. These mechanical components then became the most frequent source of failures.){{citation needed|date=July 2022}}