The determination of stray losses from input-output tests

The paper discusses the precision that can be expected in determining efficiency of certain types of direct-connected sets by careful input-output tests. The method of dealing with the results obtained is given in some detail and the conclusion is reached that such efficiency tests can be made to within about 1 part in 500. This degree of precision is closer than the limits within which the efficiency can be relied upon to remain constant. The efficiency is sometimes indefinite to as much as the general order of ½ per cent, because the no-load losses do not remain constant. Stray losses determined from input-output tests may therefore vary considerably more than the efficiency measures, but large stray losses can be detected with certainty and the small losses usually found can be found with sufficient degree of exactness to serve as a reliable check on other and more suitable direct methods. Without such check, direct methods are usually not quite conclusive, because of failure to include all possible sources of loss or to realize completely operating conditions during tests. Accurate input-output tests under commercial conditions require too much time and involve too much labor to allow of their regular use for general testing, although they are useful as a basis of final reference.