88 BELL SYSTEM TECHNICAL JOURNAL 



The caliljration of these attenuators is based on the measurement 

 of the individual resistances. Of course, sufficient measurements are 

 made to determine that any capacitances which enter do not affect 

 appreciably the accuracy of the attenuator at the maximum frequency 

 used, which may be as high as LSO kc. 



By modifying the circuit of Fig. 11 A, we may use it to measure gain 

 as shown in Fig. IIB. In this arrangement, the lower branch contains 

 an impedance Z4 that is adjusted to introduce a loss equal to that of 

 the matching impedance Zo in the upper branch. In other words, 

 with the amplifier under test out of the circuit and the standard 

 attenuator set at zero, the detector will read the same for either position 

 of the output switch. Then when the amplifier is introduced into 

 the circuit, the attenuator is adjusted until the detector reads the 

 same for either switch position, which means that the gain of the 

 amplifier is just neutralized by the attenuator and the setting of the 

 latter is read as gain. 



This circuit is used principally for the measurement of gain of audio 

 frequency amplifiers, and is capable of measuring gain as high as 

 120 T.U. corresponding to a power output of 1,000,000,000,000 times 

 the power input. 



Cross-Talk. When there is an appreciable amount of coupling 

 between two telephone circuits, any mutual interference which results 

 is known as cross-talk. It is measured in cross-talk units, a cross-talk 

 unit being defined as the relation existing between the two circuits 

 when the current in the disturbed circuit is one millionth of the current 

 in the disturbing circuit, the impedances of the two circuits being the 

 same. Under these conditions, one cross-talk unit may be assumed 

 the same as 120 T.U. An interesting form of cross-talk is that due to 

 loading coils and is of a complex type, produced by a combination of 

 capacitance, inductance and resistance unbalances in the windings. 

 Since the actual cross-talk caused by an unbalance in the coil is depend- 

 ent upon all of the conditions of the circuit, it is necessary that any 

 measurement of cross-talk made on the individual coils be made in a 

 circuit as nearly as possible the equivalent of the line in which the coil is 

 to be used. Consequently, all cross-talk circuits for the measurement of 

 loading coil cross-talk consist of networks simulating the impedance of 

 an ideal line of the type for which the loading coil is designed. The 

 principle of the method is to apply to the disturbing circuit a definite 

 input of a single frequency, usually 900 cycles, and to measure the 

 cross-talk in the disturbed circuit at the desired point in it by com- 

 paring the tone heard in the telephone receiver connected at this point 

 with the tone obtained from a cross-talk meter which is simply a device 



