Sec. 6.5] PASSIVE INTERCONNECTION OF AMPLIFIERS 



55 



reduced network cannot be less than the lowest positive eigenvalue in the 

 original lot. Accordingly, the optimum noise measure Me,opt of the re- 

 duced network cannot be less than that of the best of the original 

 amplifiers. 



Fig. 6.7. Lossless interconnection of several amplifiers. 



We shall now extend our proof to cover passive interconnections of 

 amplifiers. We again consider n amplifiers but shall now imbed them 

 into a dissipative 4;z-terminal-pair network in the configuration of Fig. 6.7. 

 Suppose, for a moment, that we disconnect the amplifiers and that we 

 throw the imbedding network into its canonical form, as discussed in 

 Sec. 4.1. This is accomplished by a lossless network transformation, and 

 results in an arrangement Hke Fig. 4.2 but with 4:n separate resistances 

 and uncorrelated noise generators. These resistances are arranged 

 in the way shown at the bottom of Fig. 6.8. If we now apply to this 

 canonical form the lossless network transformation inverse to the one 



