1444 THE BELL SYSTEM TECHNICAL JOURNAL, NOVEMBER 1953 



The photocell amplifier converts the current from the high impedance 

 photocell into a proportional voltage, having an internal impedance of 

 10 ohms. 



The following amplifier-differentiator has an internal gain of 80 db 

 from dc to 10 kc. When used as a differentiator the rising external gain 

 characteristic reaches 67 db at 10 kc. 



The output amplifier also has an internal gain of 80 db from dc to 10 

 kc. The feedback network includes an equalizer to produce current, and 

 hence flux, in the air core output transformer proportional to input volt- 

 age from dc to 10 kc. The inductance of the transformer, with constant 

 applied ac voltage, would cause a 6 db per octave decrease in current 

 above the frequency where its Q is unity. This is counteracted by design- 

 ing the external amplifier gain to increase at 6 db per octave, starting at 

 the same frequency. The output amplifier thus has the same characteris- 

 tics as a differentiator at high frequencies and its external gain is 64 db 

 at 10 kc. 



The overall gain of the system, then, is 131 db at 10 kc. Stability has 

 been obtained, even with this much overall amplification without resort 

 to compartmentation. Each tube stage utilizes shielded turret construc- 

 tion, exposed interstage leads are very short and only the input grid 

 leads are shielded because they connect to controls on the front panel. 

 A box shield surrounds the gain selector and the displacement-velocity 

 switch of the middle amplifier, the lowest level point on the main panel. 



This system uses an electronic differentiator. Ordinarily in analogue 

 computers, these are avoided because of the high-frequency noise intro- 

 duced as a result of the attendant large amplification. In the present 

 system, the averaging of the succession of impulses by the dc instrument 

 minimizes this effect. With each cycle a discontinuous section of the 

 noise is sampled, containing a dc component, but as these are random 

 in sign, their average gives rise to no error. 



The system is dc coupled up to the output transformer. Slow drifts, 

 due to grid currents or other reasons, do not result in instrument current 

 errors because of the differentiating action of the transformer. No actual 

 dc source appears in the meter circuit itself. The only concern here is to 

 maintain the amplifiers somewhere near their best operating point. Actu- 

 ally the main source of dc drift is the temperature coefficient of the photo- 

 cell and the voltage stability of the lamp power supply. 



Calibration. The calibration of the system starts with a static measure- 

 ment of the total displac^ement of the reciprocating motion to be studied. 

 This can be done using thickness gauges or a tool maker's microscope. 

 Then the device is brought into alignment with the light beam and cycled 



