Innovative Per-Column Timing Tracking Scheme for SRAM Design Yield Improvement

1. Die-Hard SRAM Design Using Per-Column Timing Tracking Shi-Yu Huang and Ya-Chun Lai Feb. 10, 2007 @ Las Vegas (IC-DFN) Design Technology Center (DTC) National Tsing-Hua University, HsinChu, Taiwan

2. Outline • Introduction • Timing Tracking Scheme • Traditional Replica-Based Scheme • Our Scheme • Experimental Results • Conclusion

3. Could trigger a yield crisis! Nanometer Effects on SRAMs Nanometer Effects Worse Device Mismatch Larger Leakage Current Wider Variations of R and C Lower VDD (smaller noise margins) Worse Supply & Coupling Noise Uncertain Delay

4. SRAM Memory Architecture CS bit line WE OE A9 word line A8 Row Decoder . . . A0 Sense Amplifier / Drivers A19 Column Decoder A10 Input-Output (M bits)

5. Bitlines’ Waveforms BL BL BL BL Reading An SRAM Cell pulsed wordline Wordline Q’ Q 0 1 cell current An SRAM Cell

6. Two Types of Sense Amplifiers A Sense Amplifier Continuous Type Latch Type Sense Enable

7. Three Major Problems for SRAM • Mismatch in Bit Cells and Sense Amplifiers • Vt mismatch shrinks the noise margin • Bitline Leakage Current • Could cause failure for READ operations • Timing Tracking • When to turn on sense amplifiers? • When to turn off wordline? (pulsed wordline)

8. cell 0 1 0 1 1 0 0 1 1 0 1 0 1 0 0 1 BL X-Calibration for Leakage Tolerance(Presented in Last IC-DFN) Leakage is calibrated in two steps: Transform the effects of the bitline leakage to a Voffsetbetween (BL, BL) Leakage Current cell cell cell cell cell cell BL cell Deduct Voffset from the input of the sense amplifier When performing sense amplification 1.5V 1.8V X-calibration circuit S.A.

9. X-Calibration BIST 1.373mm BIST Conventional 1.108mm Die Photo of Test Chip

10. Pass Fail Ileak=76.6uA Shmoo Plots Target speed: 150MHz @ 250C Measurement result: Leakage tolerance improved by 317% Supply Voltage (V) Conventional Pass Ours with X-Calibration Supply Voltage (V) Fail Ileak=320uA Injected Leakage Current (uA)

11. Outline • Introduction • Timing Tracking Scheme • Traditional Replica-Based Scheme • Per-Column Timing Tracking Scheme • Experimental Results • Conclusion

12. Traditional Scheme – Replica Bitline Property: replica bitline pair develops a logic signal (i.e., sense enable) when an accessed bitline pair builds up 100mV signal replica bitline pair decoder active wordline accessed logic sense amps CLK Ref: B. S. Amrutur et al., “A replica technique for wordline and sense control in low-power SRAMs,” IEEE Journal of Solid-State Circuits, Vol. 33, No. 8, pp. 1208-1219, Aug. 1998.

13. BL / BL Problems of Replica Bitline Based Timing Control The factors on the speed of a bitline pair: leakage, RC, driving of cell  Each column could have its own bitline development speed  A single sense enable control is susceptible to sensing errors Read cycle Read cycle Voltage (V) SE

14. Read cycle Read cycle BL / BL Voltage (V) SE Adaptive Sensing Control Each sense amp. adapts to its current driving bitline pair!

15. S.E. active ? Operating Flow Typical READ control steps Added timing tracking steps Row address decoding Timing tracker start-up Wordline activation Timing tracker monitoring Bitline discharging ΔVBL>100mV? N Y N Y Sense enable generation Sense amplification Timing tracker disabling

16. BL MC MC MC MC MC MC MC MC Overall Architecture Row Decoder WL Driver BL WL Cell Array MUX2 MUX2 det_en Timing Tracker Timing Tracker se SA SA Controller, Input Buffer, Address Buffer Latch& Buffer Latch& Buffer I/O Circuitry

17. BL MC MC BL BL / MC MC Transient Waveforms for Read Row Decoder WL Driver BL CLK MUX2 WL det_en Timing Tracker se det_en SA Latch& Buffer se Desired property: SE goes high when bitline pair has 100mV!

18. Outline • Introduction • Timing Tracking Scheme • Traditional replica-based scheme • Per-Column Timing Tracking • Experimental Results • Conclusion

19. Effect of Variation on Sense Amp. Vt • As Vt mismatch in sense amplifier becomes excessive, the probability of read failure increases. proposal proposed Pass Rate dummy bitline replica-based Local standard deviation of Vt for transistors in SA (mV)

20. Effect of Variation on Bitline Capacitance • Our is insensitive to bitline capacitance variation. • On the contrary, replica-based method is vulnerable. Proposal proposed 100fF Pass Rate 300fF 500fF dummy bitline replica-based Local standard deviation of Vt for transistors in SA (mV)

21. Compared Capacitor 1.208mm Proposed 1.108mm Layout of Test Chip (Technology): TSMC 0.18um CMOS 1P6M (Creating Nanometer Effects): We used different loadings on different bitlines so as to mimic the different operating speeds in deeper nanometer technologies

22. Layout of Compared SRAM Row decoder Cell array IO circuitry Column decoder & Output buffer Control & Input buffer & Row address buffer Column address buffer

23. Layout of Proposed SRAM Row decoder Cell array IO circuitry & Timing tracker Column decoder & Output buffer Control & Input buffer & Row address buffer Column address buffer

24. Test Chip Characteristics

25. Conclusion • Why Timing Control in an SRAM? • (1) for latch-based sense amplifier enabling • (2) for pulsed wordline control • So as to achieve lower power dissipation • Drawback of Existing Replica-Based Scheme • Replica simply cannot track every bitline pair • Proposed Per-Column Timing Tracking • Adaptive on-the-fly • More tolerant to process variation • Suitable for deeper nanometer technologies