This paper describes the access-timing control for an embedded SRAM core which operates in low and wide supply voltage (Vdd = 0.3-1.5 V). In the conventional SRAMs, a wiring-replica with replica-memory-cell (RMC) to trace the delay of memory core is used for this purpose. Introducing a wiring-replica control provides the better tracing capability in the wide range of Vdd above 0.5 V than those of using the control with logic-gate-delay. However, as Vdd is reduced below 0.5 V and gets close to the threshold voltage (Vth) of transistors, the access-timing fluctuation resulting from the variation in memory-cell-transistor-drain current (Id) is intolerably increased. As a result, the conventional control is no longer effective in such a low-voltage operation because the RMC can not replicate the variation in transistor-drain current (Id) and Vth of the memory-cells (MCs). Thus, to trace the access-timing fluctuation caused by the variation in Id and Vth is the prerequisite for the SRAM control in the range of Vdd = 0.3-1.5 V. To solve this issue, we have proposed new access-control scheme as follows; 1) a write-replica circuit with an asymmetrical memory cell (WRAM) making it possible to replicate the variation in Id and Vth for the write-access timing control, and 2) a source-level-adjusted direct-sense-amplifier (SLAD) to accelerate the read-access-timing even in low Vdd. These circuit techniques were implemented in a 32-Kbit SRAM in four-metal 130-nm CMOS process technology, and have been verified a low-voltage operation of 0.3 V which has not ever reported with 6.8 MHz. Moreover it operated in wide-voltage up to 1.5 V with 960 MHz. The required 27 MHz operation for mobile applications has been demonstrated at 0.4 V which is 6-times faster than the previous reports. The WRAM scheme enabled the write operation even at 0.3 V. And the access time which can be restricted by the slow tail bits of MCs has been accelerated by 73% to 140 nsec at 0.3 V by using SLAD scheme.

A low power bus architecture with Local and Global Charge-Recycling Bus (Local-CRB and Global-CRB) techniques, featuring virtual stacking of the individual bus-capacitance and the dummy capacitor into a series configuration between supply voltage and ground, has been proposed. These Local and Global CRB schemes make it possible to reduce not only each bus-swing but also a total equivalent bus-capacitance of the ultra multi-bit buses running in parallel. The voltage swing of each bus is given by the recycled charge-supplying from the upper adjacent bus capacitance or the dummy capacitor, instead of the power line. The dramatical power reduction was verified by the simulated and measured data. According to these data, if employing the combination of those CRB schemes in a practical chip, the ultra-high data rate of 25 Gb/s can be achieved while maintaining the power dissipation to be less than 300 mW at Vcc3.6 V for the bus width of 512 bit with the bus-capacitance of 14 pF per bit operating at 50 MHz.

Circuit techniques for realizing fast cycle time of DRAM are described. 1) A high-speed and high-efficiency word-line level Vpp supply can be obtained by a unique static CMOS double-boosted level generator (SCDB) which controls the Vpp charge supply gate. 2) A new write-control scheme eliminates the timing overhead of a read access time after write cycle in a fast page mode operation. 3) A floor plan that minimizes the load of signal paths by employing the lead-on-chip (LOC) assembly technique. These techniques are implemented in an address-multiplexed 16 Mbit CMOS DRAM using a 0.5-µm CMOS technology. A 31-ns RAS cycle time and a 19-ns fast page mode cycle time at Vcc3.3 V, and also even at Vcc1.8 V, a 53-ns RAS cycle time and a 32-ns fast page mode cycle time were achieved. This DRAM is applicable to battery-operated computing tools.

An asymptotically zero power charge recycling bus (CRB) architecture, featuring virtual stacking of the individual bus-capacitance into a series configuration between supply voltage and ground, has been proposed. This CRB architecture makes it possible to reduce not only each bus-swing but also a total equivalent bus-capacitance of the ultramultibit buses running in parallel. The voltage swing of each bus is given by the recycled charge-supplying from the upper adjacent bus capacitance, instead of the power line. The dramatical power reduction was verified by the simulated and measured data. According to these data, the ultrahigh data rate of 25.6 Gb/s can be achieved while maintaining the power dissipation to be less than 100 mW, which corresponds to less than 10% that of the previously reported 0.9 V suppressed bus-swing scheme, at Vcc = 3.6 V for the bus width of 512 b with the bus-capacitance of 14 pF per bit operating at 50 MHz.

An experimental latch circuit is fabricated by using a 0.35µm MT-CMOS technology. This latch circuit has a volume smaller by 30%, a delay time shorter by 10%, and has an active power consumption smaller by 10% over those of a conventional MT-CMOS circuit. Furthermore, at a operation frequency of 100 MHz, an SRAM employing this IPS scheme has a standby current which is 0.4% of SRAM's without using IPS scheme.