An electrically alterable, non-volatile memory cell has more than two memory states that can be programmed selectively. Programming of the cell is conducted by applying a plurality of programming signals having different characteristics to the cell. The programming signals include at least a first programming signal which programs the cell by a first increment and a subsequent programming signal which programs the cell by a second increment smaller than the first increment. As the cell is being programmed to a selected state, its programming status is verified independently of reference values bounding the memory states. For this purpose, a signal indicative of the programming status (e.g., the cell's bit line signal) is compared with a reference signal corresponding to the selected state but having a value different from the reference value or values bounding the selected state. The programming operation can thus be controlled without actually reading the memory state of the cell.

Electrically alterable non-volatile memory with N-bits per cell

Memory states of a multi-bit memory cell are demarcated by generating read reference signals having levels that constitute boundaries of the memory states. The read reference signals may be dependent upon the levels of programming reference signals used for controlling the programming of the memory cell. The memory cell can thus be programmed without reading out its memory state during the programming process, with programming margins being assured by the dependence of the read reference signals on the programming reference signals. Both sets of reference signals may be generated by reference cells which track variations in the operating characteristics of the memory cell with changes in conditions, such as temperature and system voltages, to enhance the reliability of memory programming and readout.

Memory apparatus including programmable non-volatile multi-bit memory cell, and apparatus and method for demarcating memory states of the cell

A monolithic integrated circuit containing an inverting/non-inverting voltage doubler charge pump circuit is disclosed for converting a unipolar supply voltage to a bipolar supply voltage of a greater magnitude. The unipolar input voltage is placed across a first external transfer capacitor by a first set of MOS switches during a first time period. The unipolar input voltage source is placed in series with the first transfer capacitor and this series combination of voltages is placed across a first external reservoir capacitor by a second set of MOS switches during a second time period. The voltage appearing across the first external reservoir capacitor is placed on a second transfer capacitor during the first time period by a third set of MOS switches. The voltage across the second transfer capacitor is placed into a second external reservoir capacitor with its polarity inverted by a fourth set of MOS switches during the second time period. A dual-collector lateral junction transistor, formed during the conventional CMOS processing steps used to fabricate the MOS switches, is connected as a voltage clamp between a ground potential and the two bipolar DC output lines of the power supply circuit to assure correct start-up conditions for the circuit. Gain reduction devices are placed in the semiconductor substrate to collect minority carriers which would otherwise be injected into inherent parasitic four layer PNPN junction devices created as a result of the architecture of the circuit, to prevent latch-up of the four layer devices. In a preferred embodiment, an RS-232 receiver and transmitter are contained on the same monolithic integrated circuit as the dual charge pump power supply.

Integrated dual charge pump power supply and RS-232 transmitter/receiver

A semiconductor device has multilevel memory cells, each cell storing at least three levels of data each. At least a first data composed of first data bits and a second data composed of second data bits are arranged in order that at least a bit of an N-order of the first bits and a bit of the N-order of the second bits are stored in one of the cells, the N being an integral number. A voltage corresponding to the N-order bits is generated and applied to the one of the cells in response to an address information corresponding thereto. Another semiconductor device has multilevel memory cells arranged so as to correspond to a physical address space, each cell storing 2 n levels of data each expressed by n (n≧2) number of bits (X 1 , X 2 , . . . , Xn). A logical address is converted into a physical address of the physical address space. Judging is made whether a logical address space including the logical address matches the physical address space. When matched, the most significant bit X 1 is specified once using a reference value. The specified bit is output from one of the cells corresponding to the physical address. If not matched, the bits (X 2 , . . . , Xn) are specified by n—time specifying operation maximum using maximum n number of different reference values. The data writing/reading operations to/from the semiconductor devices can be stored in a computer readable medium as program codes for causing a computer to execute these operations.

Multilevel semiconductor memory, write/read method thereto/therefrom and storage medium storing write/read program

A pipelined multi-stage analog-to-digital converter (ADC) exhibits high speed and high resolution characteristics in a small chip area using a CMOS process. An optimized high resolution multi-stage ADC improves integral non-linearity errors (INL) and differential non-linearity (DNL) errors and hence increases yield. A binary-weighted capacitor array is used in a multiplying digital-to-analog converter (MDAC) in a front-end stage, and a unit capacitor array is used in the MDACs of the latter stages thereof. Offset, feedthrough and gain errors are removed via digital correction. A digital calibration technique is adopted to reduce the non-ideal effects resulting from component mismatch, by measuring all the code errors of the front-end stage, to thereby minimize the midpoint code DNL error without reference to code symmetry.

Pipelined multi-stage analog-to-digital converter

A pipelined analog-to-digital (A/D) conversion system enhances the effective data rate of the converter in direct proportion to the number of stages in the pipeline. The pipelined A/D converter operates in conjunction with a charge-coupled device (CCD) multilevel storage (MLS) in a three-bit (eight-level) implementation. Three comparators are used in the three-bit system arranged in a sequential successive approximation configuration with control circuits and a CCD shift register.

Sequential successive approximation analog-to-digital converter

A digital-to-analog conversion (DAC) circuit and trigger comparator combination is described for encoding and decoding charge packets in a common-well multi-level signal charge-coupled memory device (CCD). The DAC circuit, which may be of the weighted capacitor type, is used to generate a staircase waveform and to create the common-well under a first gate in the CCD. The trigger comparator adjacent to a second gate in the CCD is a detection circuit which stays in one binary state until an input charge signal is received, whereupon it switches state. In particular, the weighted capacitor DAC contains an extra offset bit which is used in the analog-to-digital or regeneration operation such that when the trigger comparator switches state, the digital input to the DAC at that time correctly represents the signal charge being converted. In one embodiment a circular serial-parallel-serial memory structure is employed as the multi-level CCD memory system.

Multi-level charge-coupled device memory system including analog-to-digital and trigger comparator circuits

Differenz-schaltkreis mit ladungsgekoppelter anordnung

A pipelined analog-to-digital (A/D) conversion system enhances the effective data rate of the converter in direct proportion to the number of stages in the pipeline (9). The pipelined A/D converter operates in conjunction with a charge-coupled device (CCD) multilevel storage (MLS) (9) in a three-bit (eight-level) implementation. Three comparators (AA, BB, CC) are used in the three-bit system arranged in a sequential successive approximation configuration with control circuits and a CCD shift register (8).

A pipelined analog-to-digital conversion system

A digital-to-analog conversion (DAC) circuit (32) and trig­ ger comparator (34) combination is described for encoding and decoding charge packets in a common-well multi-level signal charge-coupled memory device (CCD). The DAC circuit, which may be of the weighted capacitor type, is used to generate a staircase waveform and to create the common-­ well under a first gate (14) in the CCD. The trigger comparator adjacent to a second gate (16) in the CCD is a detection circuit which stays in one binary state until an input charge signal is received, whereupon it switches state. In particular, the weighted capacitor DAC contains an extra offset bit which is used in the analog-to-digital or regeneration operation such that when the trigger comparator (34) switches state, the digital input to the DAC (32) at that time correctly represents the signal charge being converted. In one embodiment a cir­ cular serial-parallel-serial memory structure (37) is employed as the multi-level CCD memory system.

Yee Yen Sung

Papers

Sequential successive approximation analog-to-digital converter

Multi-level charge-coupled device memory system including analog-to-digital and trigger comparator circuits

Differenz-schaltkreis mit ladungsgekoppelter anordnung

A pipelined analog-to-digital conversion system

Multi-level charge-coupled devices memory system