Sitime Corporation · NASDAQ:SITM

Sitime — Technical Presentation (December 2015)

32 slides from Sitime's technical presentation, Dec 2, 2015

  1. Slide 1: Smart Clocking Techniques Extend Battery Life of Wearables

    Smart Clocking Techniques Extend Battery Life of Wearables Jehangir Parvereshi Sr. Manager, Customer Engineering SiTime Corporation Dec 2, 2015

  2. Slide 2: Today's Wearables: Space Sensitive & Battery Driven

    Today’s Wearables: Space Sensitive & Battery Driven  Wearable electronics are designed to: – Fit in ever-shrinking PCB real-estate – Collect and send data in short bursts – Return to the lowest power state – Stay in the lowest power state as long as feasible – Run for days on a small capacity, smal

  3. Slide 3: Wearable Architecture

    Wearable Architecture Sleep Clock RTC/WD Clock  Major power consumption contributors – MCU – Wireless radio 3

  4. Slide 4: Which RF Protocol is a Power-miser?

    Which RF Protocol is a Power-miser? The Bluetooth Low Energy (BLE) protocol has the lowest current drain • BLE has optimized radio power, bit-rate and connection time From: A. Dementeyev, S. Hodges, S. Taylor and J. Smith, "Power Consumption Analysis of Bluetooth Low Energy, ZigBee, and ANT Sensor N

  5. Slide 5: BLE Power Savings Model

    GiTime BLE Power Savings Model DESIGNERS ¢ THINGS

  6. Slide 6: Sleep Clock Accuracy (SCA) & BLE Power

    Sleep Clock Accuracy (SCA) & BLE Power • Average power is directly proportional to the ratio of “ON” time to “Sleep” time Early ON Time ON SLEEP ON ΔT SLEEP TIME TON • Early ON time (ΔT) to accommodate inaccurate sleep clock causes power penalty • ΔT = (SLEEP CLOCK ACCURACY) * (SLEEP TIME) Initiatin

  7. Slide 7: SCA & Window Widening in BLE Core 4.0

    SCA & Window Widening in BLE Core 4.0 ΔT = (SLEEP CLOCK ACCURACY) * (SLEEP TIME) 7

  8. Slide 8: BLE Power Savings Evaluation System

    BLE Power Savings Evaluation System Current waveform reflects Connection event profile SLAVE MASTER PC USB 32768 Crystal TI CC2541 TI CC2540 USB Powered CC2540 BLE CC2541 Key FOB Slave 8

  9. Slide 9: BLE Window Widening Details

    BLE Window Widening Details Rx window width is proportional to (masterSCA + slaveSCA) RX Wake up Pre-process Tx Post-process Window Sleeping Sleeping ON Time 9

  10. Slide 10: Measured 'ON Time' vs. Sleep Time

    Measured ‘ON Time’ vs. Sleep Time  masterSCA + slave SCA= 80 ppm  For Sleep Time = 4s; ON Time = 3.6 ms Sleep ON Time Time (ms) (ms) 100 2.9 2000 3.2 4000 3.6 ON Time = 5.43 ms 8000 4.3 ON Time = 3.6 ms 16000 5.2 10

  11. Slide 11: BLE Power Savings with a 32 kHz TCXO

    BLE Power Savings with a 32 kHz TCXO 200 ppm quartz crystal Quartz RX widening unacceptable at sleep time > 8s 5 ppm TCXO 2 µA Sleep current 30% power savings going from ±200 ppm to ±5 ppm SCA 11

  12. Slide 12: BLE Power Savings Test Results

    BLE Power Savings Test Results  µPower 32 kHz TCXO vs. 32 kHz crystal – µPower TCXO supports tighter SCA of < 5 ppm for maximum power savings across operating temperature • Typical 32 kHz crystal solution can only support SCA > ±200 ppm – The actual frequency stability of Master and Slave sleep clo

  13. Slide 13: Overview of Clock Sources

    GiTime Overview of Clock Sources DESIGNERS ¢ THINGS

  14. Slide 14: Characteristics of Clock Sources in Wearables

    Characteristics of Clock Sources in Wearables Device Today’s Typical Frequency Current Resume Frequency Clock Sources Range (MHz) (uA) Time (ms) Stability (ppm) RC MHz on-chip XO 0.1 to 48 5/MHz** 0.02 ±2500* RC kHz on-chip XO 0.032768 0.12 0.15 ±2500* MCU Ext XTAL + on-chip MHz XO 4 to 48 34/MHz**

  15. Slide 15: Clocking Characteristics for Next Gen Wearables

    Clocking Characteristics for Next Gen Wearables Device Next Gen Frequency Current (uA) Resume Frequency Clock Source Range (MHz) Time (ms) Stability (ppm) High Speed 4 to 24 10/MHz* 2 ±50* MCU Low Speed 0.032768 1* 200 ±5* High Speed 16, 24 10/MHz* <1 ±40* BLE Low Speed 0.032768 1* 200 ±5* * Across

  16. Slide 16: µPower MEMS Clocks

    GiTime uPower MEMS Clocks DESIGNERS ¢ THINGS

  17. Slide 17: Architecture of a µPower MEMS Oscillator

    Architecture of a µPower MEMS Oscillator MEMS Programmable MEMS Resonator Analog IC Oscillator 524 kHz Charge Pump CLK 1.5 mm Dividers, Sustaining Frac-N Drivers, Circuit PLL 0.8 I/O mm CTRL Temperature Temp OTP To Comp Memory Digital • < 100 ppm XO, ± 5 ppm TCXO frequency stability over temp • 1 Hz

  18. Slide 18: 32kHz MEMS XO & TCXO vs. Quartz XTAL

    32kHz MEMS XO & TCXO vs. Quartz XTAL SiT1532 XO / Quartz XTAL SiT1552 TCXO Solution 1.2mm2 total MCU, BLE, or MCU, BLE, or footprint Chipset PMIC 5.5mm2 total Chipset PMIC footprint XTAL+Caps Vdd XO XO XIN XIN 32kHz CLK Osc I/O Osc I/O Features SiT1532 / 52 Quartz XTAL Package Footprint w/ Load 1.2m

  19. Slide 19: µPower MEMS 32 kHz XO & TCXO for Wearables

    µPower MEMS 32 kHz XO & TCXO for Wearables MEMS XO MEMS TCXO 100 PPM Over Temp ±5 PPM Over Temp 2x More accurate than quartz XTAL 30x – 40x more accurate than quartz XTAL Industrial Temp Specification Measured Measured Quartz XTAL -160 to -200 ppm Over Temp 19

  20. Slide 20: µPower MHz MEMS Oscillator for Wearables

    µPower MHz MEMS Oscillator for Wearables  MEMS core power: 60 uA  Smallest footprint: 1.2mm2  Any frequency: 1 to 26 MHz Current Consumption (mA) MCU Operation Mode MHz MEMS + MHz XTAL + MCU MCU Operating Mode 5.2 (7% lower) 5.6 Sleep Mode 2.1 (18% lower) 2.6 20

  21. Slide 21: µPower MEMS Clock Characteristics

    µPower MEMS Clock Characteristics  SiT8021 - MHz clocks for MCU – Frequency range: 1 MHz to 26 MHz – Frequency stability: ± 50 ppm across -20⁰C to 70⁰C, VDD, aging – Core current = 60 uA – Standby current = 0.7 uA – Resume time = 2 ms – Footprint: 1.2mm2 (1508) – Drives multiple loads  SiT1552 – 3

  22. Slide 22: Smart Clocking Techniques

    Smart Clocking Techniques 22

  23. Slide 23: Wearable Using µPower MEMS Clocks

    Wearable Using µPower MEMS Clocks 1.8V μPower 32.768 kHz VDD CLK GND 1.8V μPower Antenna 32 kHz 32 kHz 24 MHz VDD 24 MHz BLE SoC CLK Low-Power GND STB MCU SPI/I2C Optional Synchronize GPIO with BLE radio state • Eliminates 2x 32.768 kHz crystals & 4 load caps • Eliminates bulky MHz crystal & 2 load

  24. Slide 24: BLE: Power Saving Clocking Techniques

    BLE: Power Saving Clocking Techniques • Replace 32 kHz XTAL with µPower 5 ppm TCXO • Shut down on-chip XTAL oscillator • Shut-down 32 kHz RC (500 ppm) • Save CPU cycles used for calibration Dialog Semiconductor BLE: DA14580 Shut-down RCX 24

  25. Slide 25: MCU: Power Saving Clocking Techniques

    MCU: Power Saving Clocking Techniques • Replace with µpower MEMS STMicroelectronics MCU: STM32L476G oscillator • Shut-down on-chip crystal oscillator • MEMS reduces wake-up time • Shut-down RC oscillator • Save CPU cycles used for calibration • Set RC oscillator to 2 MHz • Disable calibration; save

  26. Slide 26: Conclusion

    Conclusion  µPower MEMS 32.768 kHz XO reduces BLE current drain by 30% – 32 kHZ TCXO optimizes BLE RX radio ON time – Supports extended sleep time  µPower MHz MEMS XO reduce MCU power consumption by 17% – Radio synchronized turn-on/off achieves lower active and sleep current  Added benefits of us

  27. Slide 27: Power Savings Demonstration: Dialog Semi BLE Platform

    Power Savings Demonstration: Dialog Semi BLE Platform 27 27

  28. Slide 28: Demonstration HW

    Demonstration HW TCXO adapter board 32 kHz MEMS TCXO HDR Current Monitor BLE Master DA14580 BLE Slave Dialog Semi Pro kit 28

  29. Slide 29: Demo Software: SmartSnippets-1

    Demo Software: SmartSnippets-1 Sleep Time = 1.2s 29

  30. Slide 30: Demo Software: SmartSnippets-2

    Demo Software: SmartSnippets-2 Coulombs consumed ON Time = 7.63 ms • SlaveSCA = 200 ppm • Coulombs consumed during ON time = 6.6 µC 30

  31. Slide 31: Demo Software: SmartSnippets-2

    Demo Software: SmartSnippets-2 Coulombs consumed ON Time = 7 ms • SlaveSCA = 5 ppm • Coulombs consumed during ON time = 4.98 µC ; Power savings = 32% 31

  32. Slide 32: Appendix

    Appendix  White Paper Resources at: http://www.sitime.com/support/sitime-university  Extending Battery Life of Wireless Medical Devices Abstract: To continuously monitor and upload vital data, wireless medical devices need long battery life and long-term connectivity to the cloud. This paper discu

SiTime — Technical Presentation (December 2015) on 6ix