timespot - cern indico
TRANSCRIPT
TIMESPOTResults on sensors and electronics developments for future vertex detectors
Adriano Laifor the TIMESPOT teamINFN Sezione di Cagliari – Italy
HSTD12, Hiroshima 14-18 December 2019
3D silicon sensorsproviding time
resolutions below30 ps rms and
relatedelectronics
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 3
Structure, organization and objectives ofMain target:Develop and realize a demonstrator consisting of a complete yet simplified reduced size tracking system, integrating order 1000 read-out pixel channels, satisfying the following characteristics:
• Pixel pitch: ≤ 55 µm• Radiation hardness: 1016 – 1017 1 MeV neq/ cm2 (sensors) and > 1 Grad (electronics)• Time resolution: ≤ 50 ps per pixel (target = 30 ps or better )• Real time track reconstruction algorithms and fast read-out (data throughput > 1 TB/s)
Activities are organized in 6 work packages:
1. 3D silicon sensors: development and characterization2. 3D diamond sensors: development and characterization3. Design and test of pixel front-end with timing measurement4. Design and implementation of real-time tracking algorithms5. Design and implementation of high-speed readout boards6. System integration and tests
3 years work program + 1 year possible extension (2018-2021)
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 4
Places & people:Sezioni: Bologna CagliariGenova Ferrara FirenzeMilanoPadovaPerugiaTorinoTIFPA (Trento)
≈ 60-70 heads, ≈ 20-25 FTE. People from LHCb, ATLAS, CMS + others
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 5
3D Trench geometry and Pixel characteristicsThe starting (break-through) point
5
55 µm
150 µm
135 µm
55 µm
• Total charge deposit for MIP ≈ 2 fC• Full depletion @ few Volts, Velocity saturation @ > 30 V• Pixel capacitance (from simulation) ~140 fF
Biasing el. (p+)Collecting el. (n+)
Elec
tric
field
map
Column geometry (e.g. ATLAS IBL) Trench geometry(TIMESPOT)
HR Si (~ 5kΩ cm)(sensitive volume)
LR Si (<Ω cm)
Single-Side production
method
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 6
dE/dx deposit(3 pixel volume)
400 psMax
charge collection time
https://github.com/MultithreadCorner/Tcode
Full sensor model and signalsSimulated sensor performance in terms of timing
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 7
1st 3D-trench batchDelivered 15 June 2019
Timepix area: 256x256 pixels x 55 µm pitch
See talk #337 S5 by Giulio Forcolin
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 8
88
(A) Pixel-strip (10 pixels connected on the same read-out pad); (B) Single and double pixel; (C) Hexagonal (column) pixel device, based on FBK 3D Double Side Technology.
Devices are connected to electronics by wire bonding (Al, 25 µm diameter, ~ 5 mm length)
3D structures under testIn lab & under-beam measurements
For “static“ tests (IV, Vbd, Capacitance) please refer to talk #337 S5 by Giulio Forcolin …
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 9
System under test
Tests beam @
p+ Beam
DAQ Power Supplies
pM1
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 10
DUT signal (trigger)Exposed Area of DUTs
<< 1 mm2
Each DUT has a different DAQ chain(8 GHz and 4 GHz BW scopes)
Beam RF signal (50 MHz)
2 MCP signals (trigger) (time references st O(10ps))
Exposed area ≈ 5x5 cm2
Tests beam @ PSI pM1Experimental conditions
Cherenkov Time Taggers (MCP)
DUT1 DUT2
Beam E = 270 MeV/c. Beam radius on the spot s ~ 1.5 cm
Original beam: p, e+, µ+, p+ Selectable by TOFUsing a plexiglas degrader and small slitapertures we obtained an almost pure p+
beam (negligible e+ contamination)
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 11
Tests beam @ PSI pM1System inside the black box
beam
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 12
Noise (gauss) + Signal (Landau) + error (step) function fit
MIP MPV ~ 21 mV
3D Pixel @ -80V
The Max/FWHM ratio iscompatible with a Landau
distribution of a 150 µm thicksilicon
Time resolution per bins of amplitude(Genova F/E)
MIP MPV (21 mV)
Time resolution (1)of TIMESPOT 3D trench pixels
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 13
Time resolution (2)of TIMESPOT 3D trench pixels
Trench 3D with «KU» front-end
Leading Edge(Vthr = 10 V)
No ToTcorrection !
st = 46.5 ps
Time resolution becomes 28.9 psafter deconvolving time-tagger jitter
st = 30.7 ps
Trench 3D with
«Genova» front-end
CFD
Tails are the combination of 3 main effects:• Spurious signals (algorithm and in-time EMI noise)• Partial charge deposit (neighbour un-read pixels)• Weak field spots (small contribution)
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 14
st vs electronics F/Eof DUT on PSI beam
Test structure Front-end type st
Pixel strip KU* modified prod. 1 – unshielded 40 – 50 ps
Single pixel KU modified prod. 1 – unshielded 35 ps
Hexagonal column(FBK DS process)
KU modified prod. 1 – unshielded ~ 60 ps (preliminary)
Double pixel GE** board SiGe BJT + BB amp – shielded < 30 ps
Single pixel GE board SiGe BJT + GALI – shielded Bad (Oscillations)
ATLAS Phase2 50x50 with poly connection
KU modified prod. 1 – unshielded High values (>100 ps) (preliminary)
Diamond 110 KU modified prod. 2 ~ 320 ps . Worse S/N ratio
Diamond 55 KU modified prod. 2 ~ 230 ps. Worse S/N ratio
• For practical reasons only collective and wire-bonded pixel structures have been tested (x10 increase in Capacitance).
• Comparing different read-out boards is easy to argue that the measurements are limited by front-end performance and extrinsic (EMI) noise. A better/dedicated electronics is needed.
• Final answer will be given using an integrated front-end
*Kansas University
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 15
Measurementsin the lab
(more «silent» setup)
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 16
Laser setup@INFN Cagliari
• 1030 nm, 100 fs, 40 MHz pulsed laser• Pulse-picker to select pulses from 40 MHz
to single pulse• Monomode fiber to microscope• 5x and 20x optics• Optical filters for light intensity
attenuation• Pin-hole for collimation• Microscope with IR camera
5 µm diameter laser spot on sensor
5x optics cangive a cilinder
inside thesensor volume
(MIP-like deposit)
Abs coeff@1030 nm ≈
30/cmI/I0 ≈ 75% after
100 µm)
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 17
0
20
40
60
80
100
120
0.00 0.50 1.00 1.50 2.00 2.5017
st = 20 ps
Std
dev
Tim
e re
solu
tion
(ps)
Charge (fractions of MPV of MIP deposit, MPV ≈ 2 fC), Vbias = – 60 V @ room temperature
• KU Front-end• Digital CFD (on scope)• 10 pixel strip + wire bond
connection (Ctot ≈ 1.5 pF)• Charge deposit estimate
by separate calibrated CSA
Time response vs Charge
st = smeas / sqrt(2)
Dt
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 18
Cable loss evaluation and recovery
Impact of cable characteristics on measuredsensor timing performance
(in addition to better EMI setup)
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 19
1st prototype in 28-nm CMOSa 1st approach to this technology
• 3 different TDC solutions• 6-bit DAC + SPI I/F• 8-channels CSA+LE
Discriminator • Programmable power
(and speed)• General purpose OPAMP• LVDS Tx/Rx
Integrated cells:
Total area 1.5x1.5 mm2
(mini@sic)
Encapsulated chip prototype. 75 µm bonding pitchBonded at Milano workshop
Ø Main purpose: gain confidence on 28-nm CMOS and test technology withoutpushing in complexity and performance.
Ø All cells are kept independent and directly accessible from external pins(with a few exceptions)
à strongly pad-limited
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 20
DCO (dithering)
Tappeddelay
Time Amplifier
Simulated (rms is theoretical*)
TDCSimulation and tests (still preliminary)
Post-layout simulation
characteristics
TDC scheme à
MeasuredTAP TDC config
*calculated as flat channel bin profile: LSB/(12)1/2
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 21
DCO (dithering)
Tappeddelay
Time Amplifier
Simulated (rms is theoretical)
TDCSimulation and tests (still preliminary)
Post-layout simulation
characteristics
TDC scheme à
MeasuredTAP TDC config
• ≈ 20% degradation of performance in LSB resolutions
• Higher degradation in rms (but dominated by LVDS driver and readout time jitter –still under study)
à sTDC estimatedbetween 35 and 50 ps
?
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 22
Measured performance is x2 worse than expected due to x2 lower gain. Reasons still under investigations (amplifier biasing?)
CSA + DiscriminatorExpected/measured performance
«High» power (7.2 µA)«Low» power (4.1 µA)
Expected performance (from simulations)
HSTD12, Hiroshima – December 2019 TIMESPOT – Adriano Lai 23
Design of 2nd versionTo be submitted Spring 2020
Present, low power
Present, high power
Next version
We are working on the next version of the ASIC:
ü 24x48 55x55 µm2 pixels. Pixel contains: • CSA, LE discriminator, with increased power for
better time resolution (limit is 18 ps @ 22 µW)• New type* TDC (shared among 2 or 4 channels) • TOT real-time correction
ü Other amplifier and TDC test structures (out of the pixelmatrix)
ü Rad-hardening design techniques appliedü Impedance adaptation for diamond sensorü N x1.5Gbps LVDS output links (already designed and
working in 1st version)ü ASIC will be wire-bondable, but TSV ready, with a
Redistribution Layer
*Interpolated–tapped scheme.~ 40 ps LSB, 12 ps rms (theoretical). Size: 22x15 µm2
*Vernier scheme. Conversion time 17 ns~ 30 ps LSB, 8 ps rms (theoretical)
Power vs jitter within our present CSA scheme
• 3D Silicon sensors based on trench geometryhave been produced and characterized
• They show intrinsic time resolution ranging between 20 and 30 ps rms
• Front-end electronics is at the moment thelimiting stage to performance. This is evenmore important when pixel front-end isconsidered, where space and power constraintsenter the game
• A first prototype of pixel electronics has beentested giving best resolutions of 80 ps LSB with40 ps rms (estimated) for TDC and 60 ps rms(estimated) for amplifier (at reduced power consumption of 7µW/channel)
• A second batch of 3D sensor is under design and will be submitted early 2020
• A second prototype of pixel front-end ASIC withmatrix read-out will be submitted is plannedfor Spring 2020
TIMESPOTSummary
HSTD12, Hiroshima 14-18 December 2019