International Business Machines Corporation

Roger Paul Gregor

Steven Frederick Oakland

David James Hathaway

David E Lackey

Keith M Carrig

James Patrick Libous

Edward J Nowak

Sebastian Theodore Ventrone

Daniel Mark Dreps

Kerry Bernstein

Kevin Charles Gower

Michael Richard Ouellette

Frank David Ferraiolo

Stephen Victor Kosonocky

Albert Manhee Chu

Douglas Willard Stout

Alvar Antonio Dean

Larry Wissel

Peter Edwin Cottrell

Terry Cain Coughlin, Jr

Toshiharu Saitoh

Eugene James Nosowicz

Daniel Richard Menard

Janice M Adams

Paul H Bergeron

Eric Lee Carter

Harold E Reindel

Moon Ho Lee

Robert A Gottschall

Gary Francis Yenik

A pulse generation circuit delivers an output pulse whose width is tailored to the load. The pulse generation circuit comprises the following components. A drive circuit has an input coupled to receive a clock signal and an output coupled to drive a load. A comparator has an input coupled to the output of the drive circuit. Another input of the comparator is supplied by a reference voltage. A feedback circuit comprises logic gates and is coupled between the output of the comparator and the input of the drive circuit. The feedback circuit terminates a pulse output from the drive circuit when the pulse voltage output from the drive circuit exceeds the reference voltage. The reference voltage is higher than a voltage required to trigger the logic gates and a voltage required to drive the load. This ensures that the load is driven adequately over a wide range of load currents and capacitances. By setting the reference voltage between the voltage required to drive the load and the supply voltage, the pulse width is not excessive.

Pulse generator with controlled output characteristics

A method and structure of clock optimization including creating an initial placement of clock feeding circuits according to clock signal requirements; identifying clusters of the clock feeding circuits, wherein each cluster includes a distinct clock signal supply device to which each clock feeding circuit within the cluster is connected; changing pin connections between the clock feeding circuits and clock signal supply devices to switch selected ones of the clock feeding circuits to different clusters to reduce lengths of wires between the clock feeding circuits and the clock signal supply devices within each cluster; and adjusting positions of the clock feeding circuits within design constraints to further reduce the lengths of the wires.

Latch clustering for power optimization

A system and method for achieving self-regulated voltage division among multiple serially stacked voltage planes. The system of the present invention is incorporated within a source voltage plane having a source supply node for supplying current and a source ground node for sinking current supplied therefrom. An intermediate voltage supply node is coupled between the source supply voltage node and the source ground node for dividing the source voltage plane into a plurality of intermediate voltage planes. The self-regulated voltage divider of the present invention includes a first capacitor and a second capacitor that are each controllably coupled between either the source supply voltage node and the intermediate voltage supply node, or between the intermediate voltage supply node and the source ground node, such that a voltage level balance is achieved among the intermediate voltage planes.

Self-regulating voltage divider for series-stacked voltage rails

A state saving circuit and method for using the same. The circuit comprises a first latch powered by an uninterrupted power supply, wherein the first latch includes a first pair of cross coupled inverters for storing data, and includes an input cut-off control for isolating the data in the first pair of cross coupled inverters; a second latch coupled to an output of the first latch and powered by an interruptible power supply, wherein the second latch includes a second pair of cross coupled inverters and a clock input for latching the data from the first latch to the second latch; and wherein an interruption of power to the second latch results in a state being saved in the first latch.

CMOS state saving latch

A programmable delay element having a fine delay circuit with fractional units of delay. The fine delay circuit has a fine delay circuit with a plurality of selectable delay paths, each delay path having an associated delay interval. The fine delay element is electrically-coupled to a data terminal for receiving and delaying an input signal. A control circuit is electrically-coupled to the fine delay circuit to select the delay path for the input signal. In a further aspect of the invention, the fine delay circuit is electrically-coupled to a coarse delay circuit having a plurality of selectable delay blocks in a repetitive block configuration. The coarse delay circuit is electrically-coupled to a second data terminal for receiving and inserting a second signal through said fine delay circuit. The control circuit is electrically-coupled to the selective delay path of the fine delay circuit and the coarse delay circuit such that either a fine delay, a coarse delay, or a coarse and a fine delay can be selected.

Programmable delay circuit having a fine delay element selectively receives input signal and output signal of coarse delay element

Power layers of a multi-layer connection structure forming a power distribution network are partitioned to accommodate all necessary voltages for one or more chips connected thereto in each power layer. By doing so, and rearranging vias as permitted by such partitioning via length is reduced while via numbers can be increased to reduce self-inductance of the structure. Transmission lines formed by conductors in the signal layers are referenced to the correct power supply and return/image currents are made of similar path length and substantially symmetrical for both positive- and negative-going signal transitions. These effects reduce delta-I noise to levels which preserve good signal-to-noise ratios to current and foreseeable reduced signal levels.

Multiple power distribution for delta-I noise reduction

This invention reduces power in flip flop apparatuses by providing flip flop apparatuses that have fewer clock trees than prior art flip flops yet still support some or all of the Level Sensitive Scan Design (LSSD) functionality. In preferred embodiments of the present invention, one clock tree is used instead of two to provide lower power, and less switching devices in clocks splitters are used, which also provides lower power. Additionally, a flushable single clock splitter is provided that allows one clock tree to be used up to the flushable single clock splitter and provides two clocks on the output of the flushable single clock splitter. This saves some power yet still allows for dual clock flip flop designs.

Low power LSSD flip flops and a flushable single clock splitter for flip flops

An apparatus comprising a clock for providing a clock signal, means for providing a delayed version of the clock signal, two transparent latches having clock inputs controlled by opposite polarities of the delayed clock signal, a multiplexer having (i) inputs fed by outputs of the latches, and (ii) a select input fed by the clock signal, and means for providing a select signal for selecting the latch whose clock is inactive. Preferably, each of the latches has a scan input gate and a scan output gate, and the scan output of the first latch is applied to the scan input of the second latch to form a scannable latch pair. Also, preferably, the apparatus further comprises a data port for applying data to the first and second latches, and an exclusive OR gate at the data port, whereby the apparatus produces a gated clock signal. Also disclosed is a method of operating this apparatus.

Double-edge-triggered flip-flop providing two data transitions per clock cycle

Rapid set-up is achieved in a programmable delay element having identical pairs of positionally corresponding delay stages in parallel arrays. The pairs of delay elements include identical arrangements of circuit elements and are replicable in a step-and-repeat fashion to simplify delay element manufacture for any arbitrary maximum delay time to be provided. Delay stages of the delay element are comprised of multiplexers. Outputs of respective delay stages are simultaneously stored as a signal transition is propagated through the delay stages in a first order to program the delay element. Thereafter, signals are propagated through selected delay stages in a second order controlled by the simultaneously stored outputs of respective delay stages during the propagation of the signal transition. The selected stages through which a signal is propagated in the second order are the same stages through which the signal transition had propagated in the first order at the time the delay stage outputs were simultaneously stored.

Dual mode programmable delay element

An integrated circuit chip comprises a plurality of clock distribution sub-networks each including a clock input for receiving a clock signal, each of the clock distribution sub-networks having a capacitance, as seen from the clock input, substantially equivalent to others of the clock distribution sub-networks; and a structured clock buffer having a size based on a load of the clock distribution sub-networks, and providing the clock signal to the clock distribution sub-networks.

Method and apparatus for routing low-skew clock networks

A connection array for a chip provides a substantial increase in numbers of signal connection locations and a power distribution arrangement of improved robustness and noise immunity while accommodating multiple power supply voltages by providing pairs of sub-arrays aligned with chip edges and signal connection locations formed in columns orthogonal to a chip edge or segment of the chip perimeter. Signal connections in a column are spaced at a first pitch and columns of signal connections are spaced at a second pitch. Power connections corresponding to different power supply voltages are provided between columns of signal connections and along rows which are centered between rows of signal connections generally parallel to an edge of a chip. Power distribution layers may be formed as a mesh which extends in under the chip in alignment with power connections to the chip and beyond the perimeter of the chip, as well to provide multiple low-impedance power delivery paths to improve noise immunity. The connection pattern allows fewer layers of redistribution wiring to be used to escape the chip, reducing overall product cost. Thus improvements in functionality and performance can be supported at reduced cost, particularly for custom designed application specific integrated circuits.

Dual-pitch perimeter flip-chip footprint for high integration asics

An LSSD MUX D flip flop includes a multiplexer, a master latch L1 and a slave latch L2. The inputs to the multiplexer are functional data D, scan data I, and the control is scan enable SE. The L1 master latch receives its input from the output of the multiplexer and is clocked by the NAND of a -FLUSH (-A CLOCK) with an +EdgeClock (-C CLOCK.) The L2 slave latch receives its data input from the output of the L1 master latch and is clocked with the AND of -FREEZE (B CLOCK) and +EdgeClock. The output of the flip flop is the output of the L2 slave latch. This flip flop structure supports edge sensitive, level sensitive, functional, scan, freeze, flush, and test operations.

VLSI test circuit apparatus and method

A latch comprises first and second NFETs and a first inverter. Data is applied without inversion to the gate of the first NFET and via the first inverter to the gate of the second NFET. A third NFET has a drain connected to the sources of the first and second NFETs. A clock is applied to the gate of the third NFET. Thus, there is only one NFET subject to the constant switching the clock, and therefore the constant power dissipation caused by the clock. To latch the data from the first and second NFETS, first and second inverters are connected in paralle with each other such that the output of each inverter is connected to the input of the other inverter. The input of one of the inverters is connected to the drain of the first NFET and the input of the other inverter is connected to the drain of the second NFET. A second stage of latching is also disclosed.

Low power CMOS latch

A two-port memory cell design which permits simultaneous reading and writing of cells which are on the same wordline but on different Bit Select lines without increase in Read Access Time, and while maintaining memory functionality at low voltages. The memory cell uses a standard 6 transistor design to provide a differential read for fast access plus another three transistors are added to each cell to provide a means of differentially writing the cell and de-gating the write if the bit-select is not active. This cell design has applicability to multi-port memories as well.

High density two port SRAM cell for low voltage CMOS applications

A bidirectional level shifting interface circuit has first and second I/O ports and an FET with a drain-source channel connected between the first and second I/O ports. The first I/O port is connected to an I/O port of a first digital circuit operating at a relatively low supply voltage, and the second I/O port is connected to an I/O port of a second digital circuit operating at a relatively high supply voltage. This channel passes communication signals in each direction between the first and second digital circuit. A latching circuit comprising a P Channel FET is biased by the relatively high voltage supply, has an output connected to the second I/O port, and has a control input. The interface circuit further comprises an inverter circuit having a control input connected to the second I/O port and an inverted output connected to the control input of the latching FET such that when the second I/O port exhibits a binary one voltage caused by the first digital circuit, the inverted output exhibits a binary zero voltage to activate the P Channel FET to latch the second I/O port at sufficient voltage to drive the second digital circuit at binary one level.

Growing community of inventors

Endicott, NY, United States of America

Roger Paul Gregor