B.S.T.J. Briefs: Semipermanent Memory Using Capacitor Charge Storage andIGFET Read-out

One of the earliest computers used capacitors as its memory. 1 A mechanical means was used for both read-in and read-out operations. Electronic accessing was used in conjunction with vacuum tube or solid-state diodes in relatively modern computers such as the SEAC computer. 2 Capacitor storage is rarely used at present since magnetic memories meet the modern computer requirements much better. The inherent difficulty with capacitor storage was the limited holding time since a nonlinear resistor with small enough leakage currents to allow useful memory holding time was then not readily available. The old capacitor memory was charged through a diode with slow recovery time and with leakage current of 10~10 amp at best and required a large capacitor for any appreciable holding time. Furthermore, the read-out was usually destructive. The capacitor storage merits re-examination in view of the advanced solid-state devices and technology now available. Coupled with an Insulated Gate Field Effect Transistor (IGFET)," the read-out can be nondestructive. Integrated circuit techniques may prove superior to the current magnetic memories for some applications where infrequent recycling is permissible. The inherent speed should be much faster than that of magnetic units. Consider a capacitor C in series with a nonlinear element as shown in Fig. 1. The capacitor may represent the gate capacity of the I G F E T plus any external capacitor in parallel with the gate capacitor. When a positive voltage pulse with amplitude V and duration r is applied at the nonlinear element terminal, it can be shown that the stored charge Q(r) and the decay time constant r9 defined as the time required to reach l/e value of the initial stored charge Q0, can be calculated for various nonlinear resistors.