1 / 24

LBNL: 65nm pixel activities

LBNL: 65nm pixel activities. ATLAS-CMS 65nm Pixel ASIC Meeting Dario Gnani for DICES Group (*) November 26-27, 2012 (*) Abderrezak Mekkaoui, Brad Krieger, Jean-Pierre Walder , Carl Grace, Bob Zheng 1 , Henrik von der Lippe , Peter Denes, Devis Contarato ,…. 1: now at Rice University.

taima
Télécharger la présentation

LBNL: 65nm pixel activities

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. LBNL: 65nm pixel activities ATLAS-CMS 65nm Pixel ASIC Meeting Dario Gnanifor DICES Group (*) November 26-27, 2012 (*) Abderrezak Mekkaoui, Brad Krieger, Jean-Pierre Walder, Carl Grace, Bob Zheng1, Henrik von der Lippe, Peter Denes, DevisContarato,… 1: now at Rice University

  2. Talk Summary • 65nm Pixel Readout ASICs for HEP • FE-I5-P1: prototype of next-generation hybrid pixel readout ASIC • Non-HEP 65nm Pixel ASICs • HipPix400: prototype of MAPS ASIC for Electron Microscopy (EM) • Other 65nm ASICs • DCDC65nm: prototype of 4-to-1 DC-DC converter • HIPPO: prototype for new high-speed CCD readout ASIC • POM: prototype for a multi-channel straw readout ASIC for FermiLab Mu2e experiment

  3. 65nm adoption background • Benefit from trend in manufacturing costs to skip a CMOS generation • Longer availability allows for • IP re-use in different designs • Absorb tech transition overhead over longer lifespan • Need of better efficiency for similar performance in future projects (fTgm/Id scaling) • Reduced costs at time of full engineering runs

  4. LBNL 65nm strategy • Consolidate several projects into a single MPW run (chiplets) • Move to new design topologies with larger role for digital design • Develop/acquire new engineering skills in 65nm and mixed-mode designs • Build part and IP libraries • Develop verification and digital P&R flows (e.g. analog functional verification) • Select foundry by tech performance and run availability • Develop prototypes using “tiny” chip runs

  5. FE-I5-P1: Background 1/2 • Explore capabilities of advanced CMOS processes to address future HEP needs (Upgrades, High Luminosity LHC) • Establish an analog front-end baseline • Learn the best way these processes should be used in order to maximize benefit • Evaluate radiation hardness A. Mekkaoui - TWEPP 2012

  6. FE-I4 : Front End chip for the IBL FE-I5-P1: Background 2/2 • The FE-I4B chip is the production version for IBL installation • Process: 130nm CMOS twin-well process • Dimensions: 20mm x 19mm (full-reticle) • Active area: ~90% of the total area • Respects all the specifications • IBL Production : 1 Module =1 Chip • The FE-I4 pixel array is organized in Double Columns (DC) • Double Column is divided into 2×2 pixel Regions • 1 Region: 2×2 pixel • Pixel size: 50x250 µm² • Array size: 336x80=26880 pixels • Radiation tolerance: > 200Mrad • Phase I or Phase II • New pixel detector planned • 2 removable inner layers at radii of 3.3–10cm • 2-3 fixed outer layers at radii of about 15–25cm • FE-I4 fits requirements for outer layers in terms of hit occupancy and radiation hardness • A new development (FE-I5) is required for the inner layers • 20.2 mm • 336×80 pixel array • 16.8 mm • 2 mm • IO pads • Periphery • Main functional core A. Mekkaoui - TWEPP 2012

  7. FIE4 pixel region Vs Pix65nm region (assuming y=50u) FE-I5-P1: Description 1/4 FEI4 2X2 REGION (100X500) “FEI5” 2X2 REGION (100X200) If area to be kept the same as FEI4, about 4X more logic can be added • Substantial area reduction • Ultimately the width of a pixel will limited by practical considerations (power distribution) and not the number of transistors! • Room to add functionality • Need to explore new analog front-ends to keep up with digital scaling A. Mekkaoui - TWEPP 2012 7

  8. FE-I5-P1: Description 2/4 Passive RC: gate leakage limited TDAC (+/- 4b tuning) Preamp: 17fF Feedback cap. Variable “Rff” Inject Block Single to differential + Comparator “preamp” Comparator • Uses only thin-oxide 65nm Transistors • 2mA to 25mA @ 1.2V A. Mekkaoui - TWEPP 2012

  9. FE-I5-P1: Description 3/4 Config. Logic Config. Logic Analog FE Future Digital Region nXm pixels Analog FE Bump opening FE-I4 PIXEL SIZE • Region above same size as 2x1 FE-I4 pixels • 25mm cell y-pitch w/ 50mm bump y-pitch • Power distribution major factor in x-pitch scaling (125mm in FE-I5 ) • Bump mask not part of the submitted layout (same bump size as FEI4) • Test new std-cell-based SEU-tolerant latch A. Mekkaoui - TWEPP 2012

  10. FE-I5-P1: Description 4/4 • 2.6 mm 16 X 32 array 25m X 125m pixels Pixels with added mim-caps (31,27,22,18) A. Mekkaoui - TWEPP 2012 Pixels with added sensors (row 11:31) • 1.8 mm

  11. FE-I5-P1: Results 1/3 Chan 15/32 Qin: 2 to 10ke- Qin=10ke-; 5 IFF settings A. Mekkaoui - TWEPP 2012

  12. FE-I5-P1: Results 2/3 s> than FEI4 (as expected!) Channels with caps or diodes Sigma(untuned) ~350e- rms Sigma(tuned) < 60e- rms A. Mekkaoui - TWEPP 2012 ---- 565e- p-p tuned ----

  13. FE-I5-P1: Results 3/3 Radiation Effects Errors plot for the configuration cell (TRL) All patterns Pattern 0000 Pattern 1111 A. Mekkaoui - TWEPP 2012 M. Menouni- TWEPP 2012 Rows 0-255 • VG of diode-connected FETs at fixed current • Different size transistors • Small/negligible change in threshold voltages of PFETs/NFETs • PFET threshold variation more pronounced! • Some measurement errors in low current regime (meter impedance and ESD diode leakage) • New std cell based SEU tolerant latch + register • triple-redundant • self-correcting • read-back • ROM default • Similar performance to FE-I4 latch • 2-3x10-16 cm2/bit Columns 0-15 Columns 0-15 Columns 0-15

  14. HipPix400: Background/Descr./Results • CMOS detectors offer major advantages over CCD cameras in TEM applications: • Improved PSF • Lower dose • Higher speed 0.18 um CMOS K2 sensor (2010) 5m pixels • 16Mpix, 400 f/s Improved radiation tolerance Commercialproduct B.Krieger, TNS 2012; Commercial partner

  15. HipPix400: Background/Descr./Results 0.18 um CMOS K2 sensor (2010) 5m pixels • 16Mpix, 400 f/s Improved radiation tolerance Commercial product 0.35 um CMOS • TEAM2k(2009) • 9.5m pixels 4Mpix, 400 f/s HIPPIX (2011) 65nm proto • 0.35 um CMOS(2009) TEAM1k 1 Mpix B.Krieger, TNS 2012

  16. HIPPO: Background column-Parallel LBNL CCD Custom 65nm CMOS 35 e- @ 10 Mpix/s • Megapixel square sensor has ~1000 columns @ 50 μm pitch  need custom IC readout • No room for output amplifier need charge-sensitive readout • Ultimate applications require intensive DSP  advanced CMOS process • 65nm CMOS found to be the most adequate C. Grace, TNS 2012

  17. HIPPO: Description 1/2 4200 μm 16 Analog Front ends SERDES (480 Mb/s) 16 SHAs 4 ADCs 12b (80 Msps) Thick-ox input transistor to achieve the required noise level. Nominal transistor is too leaky!

  18. HIPPO: Description 2/2 Preamp J.P. Walder, TNS 2012 ADC

  19. POM: Background • SPECIFICATIONS (FNAL mu2e experiment) • Straw interface: • dissipative transmission line (300 Ohm) • 100ns tail • e- vs p+ discrimination • ADC 8-bit, 16ns samples • Positioning => timing resolution 50ps diff • TDC specs • FE noise specs • Low power: ~10mW/chan • Synchronous inhibit, low occupancy • Radiation tolerance • Two Possible Readout Electronics Schemes: • FNAL • Discrete preamp • Analog signal distribution • LBNL • Integrated preamp • Digital communication • ASIC can work in both configurations

  20. POM: Description/Results chip B chip A • Each channel performs the following functions on each end of the straw: • Front-end (5mW) • FE matching (75 to 300 Ohm) • Low-noise current-to-voltage conversion (3fC for 100fC - min input signal) • Adjustable pole-zero cancellation (60ns base-base) • Discrimination (σt~30ps) • Time stamping (16-bit, 35ps, 1.5mW dual-channel TDC) • Only one end: • Single-to-diff conversion • Amplitude sampling (8-bit, 16ns, 4mW, pipeline ADC) • Digital buffering and serialization (65Mbps) q

  21. POM: Results

  22. DCDC-65nm • General purpose test vehicle • Topologies that offer straightforward gate drive, LV blocking voltage • Able to test internal and external cap operating modes • Non-optimal due to limited die area available on this project • 4:1, 10’s MHz, 100mW/mm 2 • Clkin/Clk out supports interleaved operating mode with mutiplechips • Missing masks in first run… Brad Krieger - ATLAS upgrade 2012

  23. Lessons Learned • Interface IPs • gain experience with small pitch wire-bonding • use staggered pads • use circuit-under-pad layouts • MS/RF compatible with LO libraries • large overhead for initial customization of free resources • Logic IPs • require customization for substrate isolation • unclear total dose performance • Flows • expect out-of-date tech-files for P&R rules • very strong impact of second order effect: tune RC extraction • rgateMod parameter in BSIM4 for schematic-level noise estimates • Foundry • Use “tiny” chip process stack for all designs • Specify the expected stack and options as “special handling”in submission forms even if redundant (esp. RDL)

  24. LBNL 65nm References • FE-I5-P1 • TWEPP 2012 • HSTD-8 • HipPix400 • TNS 2012 • IEEE 2012 - Anaheim • HIPPO • TIPP 2011 • NSS 2011 - Valencia • TNS 2012 • POM • IEEE 2012 - Anaheim • DCDC-65nm • ATLAS Upgrade 2012 - Stanford

More Related