TM No. 377 



The strip chart record (figure Ii>7) for either the CMDUM or LIMDUM wave 

 meters was edited and labeled as to location, date, time sequence, depth, and 

 recorder chart speed., The strip chart tapes were then taken to the NUWS com<= 

 puter laboratory, where the tapes were read on a Telerecordex film reader 

 (manufactured by Data Instruments, Division of Telecomputer Corp., North 

 Hollywood;, Calif. )» This device allows scanning of the strip chart data*. 

 The conversion step is shown in schematic form in figure III-2. The strip 

 chart contains the train of voltage pulses. The frequency of a pulse is 

 directly proportional to the current speedy and the sense (plus or minus) 

 designates the direction of flow, When the vertical hairline of the film 

 reader is adjusted in line with the peak of the pulse and a foot pedal is 

 pressed, the pulse time is automatically placed upon punch cards. At a 

 point in the sequence when a reversal in the sign of the pulse indicates a 

 zero crossing or null (see figure III-2), a 'zero crossing card" is intro- 

 duced by pushing the appropriate button on the film reader. This card has 

 no time associated with it. Its function is to indicate to the computer 

 that the time points occurring before and after the zero crossing should 

 not be used as a pair to compute a velocity, since the computation would 

 have no meaning n 



The result thus far is a deck of punch cards listing the time of occurrence 

 of each voltage pulse, which delineates the time intervals ^J^ fcfz A/C..-»^17^ 

 (figure III-2) indexed by the proper sign. 'The output of the 'reader Constitutes 

 a proper card format for further conversion into velocity data. 



Analog V elocity Conversion Program ■=- The cards produced from the film 

 reader were placed as input data in an IBM l620 or a CDC 3200 digital computer, 

 whereupon the velocity data were generated. The calibration programs (figure 

 III-l) discussed in chapter II were programmed into the memory cells of the 

 computer so that the appropriate program could be selected. The computer 

 scanned the sequence of time values of the pulses and allocated a velocity 

 value to each time difference between successive cards (except those separated 

 by a zero crossing card). Thus.- for each pair of cards representing a time 

 interval in milliseconds, a unique va?.ue of velocity was produced. This value 

 was indicated as occurring at an instant of time halfway between the time 

 interval defined by two consecutive cards (see figure III-2). Note the need 

 to "interpolate" a speed value. This value is, in a sense, the average 

 velocity unique to the indicated instantaneous rate of impeller rotation; i.e., 

 the rate of passage of two consecutive impeller magnets. For nominal flow 

 speeds encountered in wind waves of 15 cm sec'l and 60 cm sec" 1 , this interval 

 of time was of the order of 2^0 milliseconds and 50 milliseconds, respectively. 

 The faster the flow speed, the more data points produced per unit of time. 

 Hence, the faster the flow speed, the less the error in the estimate of the 

 moving average of the velocity. The velocity conversion program for the 

 OMDUM III meter data output is given in appendix E. 



Linear Interpolation Program — At this point in the data processing, the 

 velocity -time data (i.e./v* at timeT* in figure 111=2) have been printed 

 out and usually plotted, or at least visually edited. A close scrutiny has 

 been made for bad points, which occurred (as was pointed out in chapter II) 

 mostly at the peak velocity regions. The graphing of the data points, when 



h5 



