Digital Storage TRIZ Cases

Remote glitch detectors and randomized delay harden memory systems against repeated glitch attacks and unreliable local alarms.

Integrated delay elements stagger memory instance and bitcell array startup to cut peak inrush current and ease power supply strain.

A three-transistor stacked sense amplifier cuts kickback noise and lowers minimum operating voltage to improve signal accuracy and speed.

An auxiliary disable control signal stops strobe control during LPDDR4 write-off states, suppressing noise-driven logic uncertainty.

Vertical and horizontal channel portions increase memory density while helping preserve production yield, resistance, and current drive.

By segmenting data lines and multiplexing bitlines outside the sense amplifier region, this case boosts bandwidth without dummy columns or extra chip area.

Local load capacitance in a ring oscillator raises target transistor delay share, enabling more precise SRAM delay measurement.

Stacked cylindrical electrodes with support structures increase 3D DRAM capacitor capacity while preserving chip space and integration density.

A gated clock topology lets a base memory chip wake clocks on stacked chips in time, cutting delay and power in multi-chip RAM.

Multiple per-bank refresh counters let DRAM banks refresh independently, cutting power use and easing controller scheduling overhead.

Grouping memory requests by rank lets inactive ranks enter low-power states, cutting idle power in high-density memory modules.

Per-tile timing adjustment aligns setup, hold, skew, and output delay in stacked memory, easing logic chip design and size constraints.

Handles odd latency counts in divided-clock DDR5 command shifting while stopping unnecessary clocking to reduce power.

Reducing SAN contact pad size or adding counter-voltage cuts coupling-driven column spiking and improves sense margin consistency.

Offset ping-pong counter tables track DRAM row accesses over fixed intervals to detect row hammer attacks without false negatives.

A switched drain readout limits FeFET voltage application time, cutting reservoir computer power use while preserving recognition and throughput.

Bidirectional threshold switching in SRAM feedback paths blocks read noise, protecting stored logic values against process variation.

Dummy bit-lines with tuned width and spacing plus landing pads ease buried contact formation in dense semiconductor memory layouts.

Dummy memory cells are repurposed as bleeder paths to discharge local sense lines, saving semiconductor area for added multiplexer circuitry.

Adjacent wordline activation counts trigger probabilistic extra refresh to curb DRAM row hammer damage without excessive power use.

A unified word line clamp and power shutoff scheme cuts leak current, shrinks control area, and protects memory data during sleep transitions.

Earlier complementary bit-line pre-charging avoids read anomalies and supports lower-voltage memory operation with up to 12% power savings.

Hierarchical word and bit lines with stacked logic tiers cut RC loss in SRAM arrays while preserving write margin and improving EDAP.

Using trained center-eye delays as a starting point, the controller finds eye-margin boundaries quickly to monitor DRAM health with minimal interruption.

A word-line-voltage-controlled protection transistor keeps low-voltage driver transistors within SOA while avoiding extra bias circuits, area, and power.

Boosting capacitors built into the sense amplifier raise bit-line precharge voltage without charge pumps, cutting area and power use.

Risk-based memory scans target read-disturb-prone subportions to preserve data integrity while limiting scan overhead and energy use.

A 2T2M MRAM layout places MTJs at source and drain nodes to raise cell density, lower write voltage, and improve drive capability.

Input current switches both the polarity and magnitude of anomalous Hall voltage, enabling precise multi-bit semiconductor memory states.

High-speed voltage monitoring enables staggered memory bank startup, cutting inrush current without adding unnecessary access delay.

A back gate voltage strengthens read current and lowers unselected word-line stress to reduce read disturb in ferroelectric memory cells.

Hierarchical word and bit lines with local block selection cut RC loss while preserving write margin in larger SRAM sub-arrays.

Read-before-write binning tailors ReRAM programming pulses to each cell’s resistance state, reducing overstress and overprogramming.

A back gate voltage stabilizes channel formation in 3D ferroelectric memory cells, reducing read disturb and improving read reliability.

Randomly precharged dummy columns mask IMC memory power signatures, helping protect stored weight data from side-channel attacks.

A weak program pulse after erase stabilizes resistive memory cells, reducing erase disturb bits and preventing low-resistance reversion.

Three differential write/read ports and a centered N-well SRAM cell layout increase port availability while keeping area compact and readout fast.

Input current tuning in a ferromagnetic/nonmagnetic Hall structure shifts AHE voltage polarity and magnitude for precise memory signal control.

Segmented sub-arrays and multi-directional wiring raise 3D memory density while keeping connectivity manageable for fabrication.

Block control pulses pause division clocks during self-refresh to prevent phase flips, cut data I/O errors, and keep clock phases stable.

On-die buffering in a master memory die cuts signal-line loading, enabling faster multi-die writes without external buffer ICs.

Independent cell storage uses paired bit lines and capacitor-based voltage adjustment to preserve differential sensing while increasing memory density.

Segmented cell bit lines cut capacitance and sense amplifier power while preserving memory access and adding capacity through dummy arrays.

Row-level self-refresh lets memory stop refresh mid-cycle, cutting exit delay to valid commands while lowering idle power use.

A node-coupled pre-emphasis driver and T-coil layout reduces voltage overshoot while preserving transmitter bandwidth and driver longevity.

Logic-memory cells generate random bit streams and execute logic in one circuit, cutting stochastic computing area, power, and complexity.

Duplicating write data across paired memory cells enables fast stuck-at fault correction without ECC delays, preserving read accuracy and speed.

Periodic refresh read voltage and adaptive read setup time stabilize memory-cell threshold voltage and reduce read disturb in NAND reads.

Different coercivities let closely spaced pinned magnets be magnetized in opposite directions without disturbing adjacent spintronic memory elements.

Parallel crossbar resistive memories and paired-value mapping enable compact analog dot-product computing with lower power use.