The VeloPix ASIC - CERN...30.9.2016 VeloPix TWEPP 2016 4 VeloPix ASIC module Data rates per ASIC...

29
The VeloPix ASIC 30.9.2016 Tuomas Poikela & VeloPix design and testing team TWEPP 2016, Karlsruhe, Germany

Transcript of The VeloPix ASIC - CERN...30.9.2016 VeloPix TWEPP 2016 4 VeloPix ASIC module Data rates per ASIC...

  • The VeloPix ASIC

    30.9.2016Tuomas Poikela

    & VeloPix design and testing teamTWEPP 2016, Karlsruhe, Germany

  • 30.9.2016 VeloPix TWEPP 2016 2

    Overview● A quick VELO upgrade overview● Chip architecture● First measurements● Summary & future plans

  • 30.9.2016 VeloPix TWEPP 2016 3

    The LHCb VELO upgrade

    Artist's

    impre

    ssion o

    f

    new VE

    LO

    624VeloPixASICs

    TWEPP 2016 : Sneha Naik “On-detector electronics for the LHCb VELO Upgrade”

  • 30.9.2016 VeloPix TWEPP 2016 4

    VeloPix ASIC moduleData rates per ASIC (Gbps)

    The hottest chips 5.1 mm from the beam

    Data per chip: ~15.1 Gbps, 2.9 Tbps for VELO

    The module installation during the CERNLong Shutdown 2 (LS2) 2019/2020.

    Highly non-uniform radiationwithin the hottest chip: From 50 to 400 Mrad.

    Hybrid pixel readout chip:Bonded to a sensor.

  • 30.9.2016 VeloPix TWEPP 2016 5

    Quick comparisonFeature VeloPix (2016) Timepix3 (2013)

    Readout type Continuous, trigger-less, binary

    Continuous, trigger-less, ToT

    Timing resolution/range 25 ns, 9 bits 1.5625 ns, 18 bits

    Power consumption < 1.5 W cm-2 < 1.0 W cm-2

    Pixel matrix, pixel size 256 x 256, 55 um x 55 um

    256 x 256, 55 um x 55 um

    Radiation hardness 400 Mrad, SEU tolerant -

    Peak hit rate 800 Mhits/s/ASIC50 khits/s/pixel

    80 Mhits/s/ASIC

    Sensor type Planar silicon, e- collection Various, e- and h+ collection

    Max. data rate 20.48 Gbps 5.12 Gbps

    Technology 130 nm CMOS, tech A 130 nm CMOS, tech B

  • 30.9.2016 VeloPix TWEPP 2016 6

    Project overview● Design started in June 2013 (after

    Timepix3 submission)● Change of technology (130nm → 130nm)● The chip was submitted May 26th 2016,

    wafers received on 31st August● Fabricated (and diced) chips back at

    CERN on 7th September● Production testing later this year (624

    chips)● Irradiation campaign in the future with

    sensors bondedVeloPix wafers

  • 30.9.2016 VeloPix TWEPP 2016 7

    Chip architecture128 double columns (14.08 mm)

    256

    row

    s (1

    4.08

    mm

    )

    SP63SP63

    SP62SP62

    SP61SP61

    SP0SP0

    SP1SP1

    SP2SP2

    SP63SP63

    SP62SP62

    SP61SP61

    SP0SP0

    SP1SP1

    SP2SP2

    SP63SP63

    SP1SP1

    SP2SP2

    SP63SP63

    SP62SP62

    SP61SP61

    SP0SP0

    SP1SP1

    SP2SP2

    PixelHitProcessorPixelHit

    Processor

    SP0SP0

    SP63SP63

    SP62SP62

    SP61SP61

    EoCEoC EoCEoC EoCEoC EoCEoC EoCEoC

    routerrouterDACsDACs PLLPLL

    SLOW CONTROLSLOW CONTROL TIMING, FAST CTRLTIMING, FAST CTRL

    2.3

    mm

    Super pi xel

    GW

    TG

    WT

    GW

    TG

    WT

    The pixel matrix:256 x 256 pixels128 x 64 super pixels (2x4 pixels each)

    Architecture:Packet-based, 8 pixels/packet + 9 bit time stamp

    Leads to 30% reduction in data rate

    Data-driven, 20Mpackets/s / double column- Timepix3 rate: 1.2 Mpackets/s

    SP61SP61

    SP62SP62

    10-b

    it To

    A @

    40

    MH

    z

    SP0SP0

  • 30.9.2016 VeloPix TWEPP 2016 8

    Front-end architecture

    Global Threshold Vthr

    Front-end (Analog)LeakageCurrentcompensation

    Preamp

    Inputpad

    1 pixel

    V1 V2

    Counters &

    Latches

    4-bit LocalThreshold

    Front-end (Digital)

    Synch.&

    Clock gating

    ToTProcessor

    6-bitLFSR

    Clock(40/80 MHz)

    Super pixel (Digital)

    Common for 8 pixels

    Data nodeData node

    To the next node

    Validevent

    FIFO

    BX ID

    CL

    OC

    Km

    anag

    erC

    LO

    CK

    man

    ager

    4 fFMaskbit

    Shutter

    10 b

    its

    6

    Test bit

    Injected charge:Qin=Ctest (V2 - V1)

    Test pulse

    Synchro noushandsha ke

    Ctest 5 fF

    8

    Testpulse

  • 30.9.2016 VeloPix TWEPP 2016 9

    Double column datapath

    FIFO Node

    NodeFIFO

    FIFO Node

    NodeFIFO

    64 FIFOs0 1

    RegisterRegister

    Arbiter

    Arbiter

    PreviousNode

    FIFOData

    Size = 2packets Up to 20

    Mpackets/s

    NextNode

    8 digitalFront-ends8 digitalFront-ends

    8 analogFront-ends8 analogFront-endsSe

    nsor

    Super pixelcolumn

    Analog pixelcolumn

    Digital pixelcolumn

    Super pixel packet:

    hitmap 8bAddr 6b ToA 10b

    Timestamping @ 40 MHz

  • 30.9.2016 VeloPix TWEPP 2016 10

    Periphery datapath

    64 EoCs 64 EoCs

    LeftData Fabric

    RightData Fabric

    Cen

    ter

    Nod

    e

    Pixel Matrix

    FIFO

    19.20 Gbps19.20 Gbps

    Data Node

    Router/Scramblers/Framing

    4.8 Gbps per bus

    4x 5.12 GbpsGWT link Dataout160 Mpackets/s/node

    64x up to 20 Mpackets/s 64x up to 20 Mpackets/s

  • 30.9.2016 VeloPix TWEPP 2016 11

    TMR/RadHard design● Manual triplication

    (except configuration registers)

    ● Pixel data flip-flops (FF) unprotected

    ● Full TMR in FSM & configuration FFs, pixel config latches

    ● No on-chip SRAM!● NMOS ELT transistors

    used in analog front-end

    ● Std cell library characterized at 400 Mrad, High-Vt NMOS

    ● Corner used for syn/PnR: slow process, VDD 1.08V, 400 Mrad, 25°C

  • 30.9.2016 VeloPix TWEPP 2016 12

    First debugging results

    SPIDR-readout system developed at Nikhef

    VeloPix

    TWEPP 2016: Bas Van Der HeijdenSPIDR, a General-Purpose Readout System for Pixel ASICs

  • 30.9.2016 VeloPix TWEPP 2016 13

    First signs of life

    VeloPix responding to the first slow control command

  • 30.9.2016 VeloPix TWEPP 2016 14

    Tests performed so far

    PLL locked, running at 320 MHz

    Slow control responds, register R/Waccess works

    The pixel matrix resettable, readableand writable

    Analog & Digital front-ends respond to test pulses correctly

    Shutter/test pulse logic works.

    Packet latency (low rate) corresponds toexpected

    Timing and fast control working

  • 30.9.2016 VeloPix TWEPP 2016 15

    First measurements

    ● Power:➢ Analog➢ Digital (IDLE), matrix

    clocking, time stamp bus● DAC (work as expected)● GWT eye diagram

    ● S-curves using analog test pulse

    ● Noise● Threshold variation● Super pixel packet

    latency

  • 30.9.2016 VeloPix TWEPP 2016 16

    Power consumption (1.2V)● Analog: 387 mA, 476 mW● Digital:

    1. After chip power up: 374 mW2. Matrix clock enabled: 694 mW3. ToA counter enabled: 718 mW

    Clock: +320 mW

    ToA bus power +24 mW

    AnalogDigital Pe-ripheryDigital Matrix

    Total: 1.2 W (0.52 W/cm2)(Meets < 1.5 W/cm2)

    TODO: Measurement with high rate

  • 30.9.2016 VeloPix TWEPP 2016 17

    GWT Eye diagram @ 5.12GbpsFirst test runs of VeloPix GWT.2 links running at the same time.

    Very first measurements!BER/jitter to be verified.

  • 30.9.2016 VeloPix TWEPP 2016 18

    Measurements with test pulse

    Global Threshold Vthr

    Front-end (Analog)LeakageCurrentcompensation

    Preamp

    Inputpad

    V1 V2

    4-bit DAC

    Synch.&

    Clock gating

    4 fFMaskbit

    Test bit

    Injected charge:Qin=Ctest (V2 - V1)

    Ctest 5 fF Testpulse

    Test pulse

    Ton

    Test pulse

    #pulses

    ToffCounten

    Pixel 255

    Pixel 254

    Pixel 1

    Pixel 0

    Pixel 255

    Pixel 254

    Pixel 1

    Pixel 0

    TP Generator

    mask#pulsesTonTofTphase

    Slowcontrol

  • 30.9.2016 VeloPix TWEPP 2016 19

    Digital: ToT mode

    Test pulse

    Ton

    Ton = 4 + Pixel X % 32

  • 30.9.2016 VeloPix TWEPP 2016 20

    Analog test pulses in one pixel0.0154 ke-/DACLSB => 1 DACLSB = 0.38 mV => 24.6 mV/ke-

    14-bit DAC Code

    25 test pulses

    Ene

    rgy

    (ke-

    )14-bit DAC Code

  • 30.9.2016 VeloPix TWEPP 2016 21

    Test pulses in 32 pixelsD

    AC

    Cod

    e

    Energy (ke-)

  • 30.9.2016 VeloPix TWEPP 2016 22

    Electronic noise (preliminary)

  • 30.9.2016 VeloPix TWEPP 2016 23

    Threshold equalization (preliminary)

    DAC Code

    Cou

    nts

  • 30.9.2016 VeloPix TWEPP 2016 24

    Summary of pixel measurements

    Digital pixel front-end fully functional.Responds correctly in ToT and Photon counting modes.All measurements without sensor.

  • 30.9.2016 VeloPix TWEPP 2016 25

    Super pixel packet latency (low rate)

    Super pixels onTop of the column haveHigher latency.

    0 10 20 30 40 50 60 70 800

    0.2

    0.4

    0.6

    0.8

    1

    1.2

    Latency greater than (# clock cycles @ 40 MHz)

    % o

    f all

    pack

    ets (

    x 10

    0%)

    Counter

    SP

    BX ID

    T0

    T1

    Latency: T1 - T0

    10

    All packets have latency > 0

    All packets haveLatency < 70

    Matches simulations. Indicates time stampingworking as expected.

  • 30.9.2016 VeloPix TWEPP 2016 26

    Summary● VeloPix ASIC, designed in 130nm CMOS, presented● First silicon for debugging 7th September ● First results show the chip is alive and eyes open:

    ➢ Power 1.2 W/ASIC, DACs working, pixels functional➢ Pixel: Gain ~25 mV/ke-, ENC 63 e-, no systematics➢ GWT serializer working, time stamping works

    ● Debugging and fine-tuning will continue in the following weeks:➢ Full DAQ chain tests, GWT BER and jitter➢ PLL characterization

    ● Production testing later this year at CERN

  • 30.9.2016 VeloPix TWEPP 2016 27

    Thank you!● List of contributors:

    ➢ ASIC designers: Jan David Schipper, Vladimir Gromov, Sandeep Miryala, Xavi Llopart, Rafael Ballabriga, Winnie Wong, Tuomas Poikela

    ➢ Support, readout and testing: Jerome Alozy, Martin van Beuzekom, Henk Boterenbrood, Bas van der Heijden, Jan Buytaert, Marco Daldoss, Edgar Lemos Cid,

    ➢ IP Blocks: Stefano Michelis, Pedro Miguel Vicente Leitao, Rui De Oliveira

    + many others I forgot to mention...

  • 30.9.2016 VeloPix TWEPP 2016 28

    Spares

  • 30.9.2016 VeloPix TWEPP 2016 29

    Power routing and bump padsDigital

    Bumppads

    Powerrails

    Analog

    Supe

    r pi

    xel =

    2 x

    4 p

    ixel

    s

    Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Slide 10Slide 11Slide 12Slide 13Slide 14Slide 15Slide 16Slide 17Slide 18Slide 19Slide 20Slide 21Slide 22Slide 23Slide 24Slide 25Slide 26Slide 27Slide 28Slide 29