DYNAMIC MEASUREMENTS ON ELECTliOMAGNKTIC DEVICES 1 I 13 



well controlled. The two systems therefore operate in substantially a 

 synchronized fashion when convenient cycle times are chosen, such as 10 

 cycles per second. The overhead fluorescent lamps fluctuate in light 

 intensity at 120 cycles per second, so that the contacts in the fluxmeter 

 operate in synchronism with the fluctuations in light intensity. Any 

 measurements made with overhead illumination getting into the photo- 

 cell will therefore have a superimposed 120-cycle ripple. For the same 

 reason extreme care has to be taken to avoid any 60-cycle pickup in the 

 apparatus. The lamp and the heaters on the first amplifier tul)es are 

 supplied with well filtered dc and cannot be operated by ac. Noise not 

 in synchronism with the contacts is of little importance, as the averaging 

 of the meter cancels it. A comment on the linearity of the photocell 

 should be made at this point. This system operates on a variable width, 

 rather than a variable density basis. Consequently the linearity of the 

 system depends upon the uniformity of emission of the photocell surface. 

 This can be verified by moving a vane with a micrometer and record- 

 ing the output of the amplifier with a precision dc voltmeter. If the rela- 

 tionship is not linear other photocells can be substituted until sufficient 

 linearity is achieved. 



Amplifier System 



General Fig. 4 showed the three dc amplifiers in block diagram form. 

 These amplifiers have each been designed to operate with full internal 

 gain from dc to 10,000 cycles. The external transfer characteristic of 

 each is controlled by its input and feedback networks. The design of 

 these networks proiddes ideal transfer characteristics from dc up to 

 10,000 cycles. At 10 kc, the frequency response deviates by less than 3 

 db from the ideal. This same frequency corresponds to a time constant of 

 16 microseconds. On a small signal basis, the fidelity of measurement 

 therefore will extend to events occurring in times of the order of 16 

 microseconds. 



A more serious limitation to the accuracy is the finite plate voltage 

 available for the output amplifier which results in amplifier over loading 

 on large peaks. As will be shown, this can delay the response of the meter 

 to sudden velocity discontinuities. 



The amplifiers have been designed to operate from two voltage sup- 

 plies, plus and minus 250 volts. The —250 volts is a series regulated throe 

 stage circuit with an output impedance of less than 0.8 ohm at all fre- 

 quencies. It also ser\^es as the reference ^'oltage for a three stage shunt 

 regulated +250-volt supply. Keeping the magnitudes equal minimizes 

 errors due to power supply voltage variations. 



