Page 41 - The Art and Science of Analog Circuit Design
P. 41

We Used to Get Burned a Lot, and We Liked It


         Figure 3-5.
         A medium-quality table radio of the 1950s. Being decorative, the cabinet and dial are of good quality, in the
         upper-right corner is a magic-eye tube, an oscilloscope-like gizmo that gives an analog indication of tuning-
         accuracy. From the John Eckland Collection, Palo Alto, California. Photo by Caleb Brown.






























                          leakage currents skyrocketed past 100°C to the extent of debiasing cir-
                          cuits. Their Vbe went to zero at 200°C; that is, the whole transistor be-
                          came intrinsic and was a short-circuit. Furthermore, you couldn't find
                          two devices that halfway matched with respect to Vbe and beta and out-
                          put impedance. You didn't bother making instrumentation circuits with
                          those devices; there just weren't any matched pairs to be found. The
                          Vbe's also suffered from terrible long-term drift, I think because germa-
                          nium could never be alloyed adequately for a solid contact. It didn't mat-
                          ter; chopper-stabilized tube op amps were common and worked well. I
                          still have one of the best VTVMs ever made, a Hewlett-Packard chopper-
                          stabilized model that has sensitive DC ranges and a 700MHz active AC
                          probe.
                            What really made my decision to use transistors was the advent of the
                          silicon NPN device. Silicon could tolerate temperature, and was insensi-
                          tive to excessive soldering. It never went intrinsic, and beta control al-
                          lowed for matched pairs. The high-quality differential input stage made
                          the industry of hybrid op amps possible, and some of them could handle
                          the same signal voltages as the tube op amps. Silicon transistors even
                          gave decent frequency responses, although the faster devices were still
                          electrically delicate. Silicon made TVs and radios work better too,
                            Circuit design changed overnight. The threshold voltage of tubes
                          (analogous to the threshold of JFETs) would vary over a 3:1 range,
                          Because of the poor bias point accuracies, most circuits were AC cou-
                          pled. This precluded them from many industrial applications. Although

         24
   36   37   38   39   40   41   42   43   44   45   46