Pixel - Indico

Download Report

Transcript Pixel - Indico

MIMOSA32:
October 2011 Submission in
Tower 0.18 µm Process
Yavuz DEGERLI
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
1
Standard MAPS (Monolithic Active Pixel Sensor) Pixel
• Sensing elements and processing electronics on the same substrate
using a standard CMOS process
• Ionizing particles create e-h pairs in the lowly doped epi-layer
• Electrons diffuse thermally to the N-well/P-Epi diode
A
B
Electrical Potential
N+
N+
N+
P-Well
N-Well
N+
P-Well
MIPs:
100% Fill Factor
e-
80 e-h/µm
P- Epitatial Layer
e-
P++ Substrate
B’
Depth
eIonising
Particle
e-
(total charge)
A-A’
B-B’
A’
Only NMOS transistors can be used in the pixel, since any additional N-well used for PMOS
transistors would compete for charge collection
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
2
MAPS in Quadruple Well Process (Tower 0.18 µm)
Cross-section of a CMOS wafer with the deep P-well implant
NMOS
N+
Diode
N+
P-well
N+
NMOS
N+
N+
N-well
PMOS
P-well
N-well
e-
e-
P+
P+
Deep P-Well
eIonizing
particle
e-
P-Epitaxial Layer
P++ Substrate
Thanks to the deep P-Well implant, PMOS transistors could be used in the pixel
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
3
Tower 0.18µm CIS Process
Standard and Pinned Diodes (available in CIS process)
Standard 3T pixel:
Nwell, deep-Nwell, gated Nwell
Pinned: buried diode with a transfer gate (CCD-like)
(optimized) layout only supplied by Tower!
“Half-pinned”: buried diode with a direct contact
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
4
Binary MAPS Principle
•
In most cases, the voltage drop due to the charges collected on the diode of a pixel is
lower than the natural process mismatches of MOS transistors (10 mV for a MIP)
•
In order to extract the signal and implement efficient on-chip discrimination, one needs
high-precision (low-noise & low-FPN) and fast front-end electronics:
 In-pixel amplification close to sensing element,
Addressing
 Auto-zeroing techniques in every amplification stage of the analogue readout path
Pixels
Discriminators
Zero Suppression & Memory
Rolling Shutter Mode  Low Power dissipation  Suitable for large size imagers,
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
5
Pixel with CDS and Amplification
Feedback
or
Vclp
Clamp
p+/nwell
SF
VDD
RD
CS
Sel_Pix
C
n+
PWRON
n-well
CALIB
p-epi
p-sub
Ib
Low-pass filter and negative feedback
stabilize the operating point of the amplifier
and the baseline of the pixel
 more robust against process
mismatches and ionizing irradiation effects
Pixel
Column
Signal amplification by CS and two “Double Sampling” stages:
First DS stage (capacitor-switch)
 suppresses offsets (FPN) of the CS and polarization device
of the charge collection diode,
 temporal noises (1/f and white) of CS not reduced
fCK=100 MHz
PWRON
RST
Clamp
RD (1)
CALIB (2)
Second Correlated DS stage (column circuitry)
 suppresses offset and 1/f noise of SF,
 white noise of SF not reduced
LATCH
T
Static power dissipation: ~100 µW/pixel (mainly dominated by the output stage) [MIMOSA26: ~180µW/pixel]
Readout time: < 80 ns/row possible according to simulations
[MIMOSA26: 160 ns/row]
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
6
Column-Level Auto-zeroed High-Speed Discriminator
Vr2
S1
Vref1


SF
S2
Vref2

S1’
VRD
Vin
VCALIB

S4
S3

C1
C2
- +
A
+
- +
A
+
C1’
S2’
- +
A
+
C2’


Q
S3
- A
+
+
Latch
Q

3
S4’
Vr2
• Both IOS (Input Offset Storage) and OOS (Output Offset Storage) techniques,
• All offsets of gain stages removed using only 2 control signals and phases.
t1: S1-S1’, S3-S3’, S4-S4’ ON; S2-S2’ OFF
Sampling of
- offsets of gain stages,
- signal level of the pixel (VRD),
- threshold voltage Vref1.
t2: S2-S2’ ON; S1-S1’, S3-S3’, S4-S4’ OFF,
Offset correction (auto-zero) and comparison with
- offset of the output stage of the pixel (VCALIB),
- the common mode of the threshold voltage Vref2.
t3: S2-S2’ ON; S1-S1’, S3-S3’, S4-S4’ OFF,
Latching
[email protected]
Threshold:
Vth=Vref1-Vref2
Signal:
Vsig = VRD-VCALIB
Static power dissipation: ~120µW
ALICE-MFT Meeting – March 26th, 2012, Orsay
1 (RD)
2 (CALIB)
3 (LATCH)
t1
t2
t3
[MIMOSA26: ~250µW/discri]
7
Column-Level Auto-zeroed High-Speed Discriminator
Layout of the
discriminator of M32
Gain Stage
Vout-
Vout+
Vin+
(20µm x 230 µm)
Latch
LATCH
LATCH
Q
Q
Vi1
Vi2
Vin-
LATCH
Ib
A4
• Output common mode voltage well defined,
doesn’t need common mode feedback circuit
• Dynamic latch: No static power dissipation
Vcm(out)= (Vo+ + Vo-)/2 = VDD – ( ib /2).R
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
8
3.3 mm
MIMOSA32: Overall Chip
13 mm
Overall chip dimensions: 13x3.3 = 43 mm²
• Prototype for charge collection optimization, radiation studies and in-pixel amplifier optimization,
• Prototype for in-pixel amplification & correlated double sampling and column-level discrimination,
• Various test structures (LVDS I/O, JTAG, ramp generator, very large diode, etc.),
• Prototype including pixels in-pixel discrimination (extension of the 3D RSBPix architecture),
• Small prototype including pixels with ampli-shaper-discri,
• SPAD: CMOS SiPMs (InESS).
* Submission: Octobre 24th, 2011
* Total surface paid: 50 mm²  82.3 kUSD (60 chips) + 4.4 kUSD (2x20chips)
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
9
3.3 mm
MIMOSA32: Charge Collection & In-Pixel Amplifier Study Chip
Pixel Array (128 x 256)
5.2 mm
• 32 different sub-arrays of 64x16 pixels,
• Basic pitch 20µm, except two sub-arrays (20x40 µm² and 20x80 µm²),
• 3T basic pixels (with standard NMOS, Enclosed Layout NMOS, « logic gate » as disturber to check deep P-well, etc.),
• Various diode types and sizes (N-well/P-epi, Deep N-well, half pinned),
• Pixels with NMOS, PMOS and CMOS based amplifiers (N-well/P-epi diode),
• 5 bit decoder to choose one of 32 sub-arrays, 16 parallel analog outputs,
• Integration time < 32µs,
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
10
Sequencer
1.9 mm
MIMOSA32: Column-Level Discrimination & In-Pixel Amplifier Study Chip
Pixel Array (128 x 64)
Discriminators (128)
Output Multipexer (128 to 16)
3.3 mm
• 4 different sub-arrays of 64x32 pixels,
• Pitch 20µmx20µm, standard N-well/P-epi diodes,
• Pixels with NMOS, PMOS based amplifiers and diode-connected transistor or P+/N-well diode as bias element
(same pixels also repeated on the previous prototype),
• Column-ended discriminators (2 variants),
• Simple digital sequencer,
• 16 parallel digital outputs,
• 3 bit decoder to choose a sub-array of 64x16 pixels & 16 discriminators,
• Common threshold voltage for 128 discriminators
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
11
Thank you for your attention
[email protected]
ALICE-MFT Meeting – March 26th, 2012, Orsay
12