REPRODUCTION OF MAGNETICALLY RECORDED SIGNALS 1159 



whereas there is reason to believe that the effects of self-demagnetization 

 cannot possibly account for more than something like ten or fifteen db loss 

 and hence could not follow the equation given above. 



In view of this it seems reasonable to try as a first guess the assumption 

 that all the unexplained loss is due to spacing. 



If this assumption properly accounts for the shape of the measured re- 

 sponse curve there will be at least some reason to suppose it may be correct; 

 particularly so if the required amount of effective spacing seems reasonable. 



The lowest solid curve, No. 7 of Fig, 9, has been computed on this basis. 

 That is, a spacing loss corresponding to 0.81 mil effective spacing has been 

 subtracted from curve 4. It is seen that this computed response curve fits 

 reasonably well with the measured points. Furthermore, 0.81 mil effective 

 spacing corresponds to quite reasonably good magnetic contact. 



If this interpretation of the measured data is correct then it is obvious 

 that the high-frequency response could be improved a great deal if more 

 intimate magnetic contact between the reproducing head and the recording 

 medium could be achieved. To this end an attempt was made to lap the 

 surface of the head in such a way as to remove material very gently and 

 slowly. After lapping, the response was appreciably improved as indicated 

 by the set of measured points around curve 6. This curve was computed 

 assuming an effective spacing of 0.36 mil. Note that the computed curve 

 now fits the measured points very well indeed. 



After still more lapping,^ the measured response points around curve 5 

 were obtained. In this case it is necessary to assume only 0.23 mil effective 

 spacing in order to account for the measured curve. Further lapping failed 

 to give further improvement in response but a defect in the head which may 

 account for this has since been found and it is believed that with great 

 care one might actually measure something very close to curve 4. 



To summarize, this is what seems to have been found. It is possible to 

 compute a response curve taking into account gap loss, eddy current losses, 

 and thickness loss. If this curve is compared with the final measured re- 

 sponse curve it is found that the measured curve gives less high-frequency 

 response than was computed. The difference between the two curves is 

 just the right sort of function of frequency and of just the right magnitude 

 to be accounted for by an effective spacing of 0.00023 inch between the 

 reproducing head and the recording medium. It seems probable that the 

 effective spacing could not have been much smaller than this value and 

 therefore it may be correct to assume that practically all the unexplained 



' After each lapping it was found that smaller values of bias current sufficed to give 

 maximum reproduced voltage. This is presumably because the improved magnetic con- 

 tact made the bias current more effective. 



