0% found this document useful (0 votes)
67 views16 pages

Timepix4: Specs, Features & Plans

Timepix4 is an upgraded version of Timepix3 that aims to improve time and energy resolution. Key planned improvements for Timepix4 include: 1. Smaller 65nm technology node compared to 130nm for Timepix3 for improved performance. 2. Ability to butt four sides of chips for larger arrays compared to three-side buttability of Timepix3. 3. Expected time resolution of ~200ps compared to 1.56ns for Timepix3 through latching of an internal VCO phase. 4. Expected energy resolution of <1keV compared to <2keV for Timepix3 through improvements to the front-end and use of a lower supply voltage. 5

Uploaded by

w96ppcqykn
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
67 views16 pages

Timepix4: Specs, Features & Plans

Timepix4 is an upgraded version of Timepix3 that aims to improve time and energy resolution. Key planned improvements for Timepix4 include: 1. Smaller 65nm technology node compared to 130nm for Timepix3 for improved performance. 2. Ability to butt four sides of chips for larger arrays compared to three-side buttability of Timepix3. 3. Expected time resolution of ~200ps compared to 1.56ns for Timepix3 through latching of an internal VCO phase. 4. Expected energy resolution of <1keV compared to <2keV for Timepix3 through improvements to the front-end and use of a lower supply voltage. 5

Uploaded by

w96ppcqykn
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 16

Timepix4: Specs & Plans

X. Llopart

1
Timepix3  Timepix4
Timepix3 Timepix4
Technology IBM 130nm TSMC 65nm
Pixel Size 55 x 55 µm ≤ 55 x 55 µm
3-side buttable 4-side buttable
Pixel arrangement
256 x 256 256 x 256 or bigger
Data driven PC (10-bit) and TOT (14-bit) CRW: PC and iTOT (12...16-bit)
Operating Modes
Frame based TOT and TOA
Zero-Suppressed Data driven < 80 MHits/s < 500 MHits/s
Readout Frame based YES YES
TOT energy resolution < 2KeV < 1Kev
Time resolution 1.56ns ~200ps
Readout bandwidth 5.12Gb (8x SLVS@640 Gbps) 20.48 Gbps (4x 5.12 Gbps)
No volcano  Dynamic gain
Front-end “with” Volcano
But supply only 1.2V

2
TOT High Resolution
Tpeak < 25ns

Preamp Out

Disc Out

Pixel Readout Starts


Clk (40MHz) (475ns→ 19 clock cycles)

Global TOA (14-bit) 16382 16383 16384 0 1 2 3 4

FTOA_rise (4 bits)=7 FTOA_fall (4 bits)=11


VCO Clk (640MHz)

TOA (14-bit) X 16383 TOA (14 bits)=16383

TOT (10 bits) =4


TOT Clk (40MHz)

TOT_ HiRes=(TOT) + FTOA_rise - FTOA_fall

3
Simulation
900

800
TOT (Timepix3) measured
700 TOT (Timepix3) simulated
TOT resolution [e- FWHM]

TOT (Timepix4) simulated


600

500

400

300

200

100

0
0 100 200 300 400 500 600
FE return to zero [ns/1ke-]

4
Timepix4
• Improve TOA resolution Latching of internal VCO phase
• Technique used by “Time-of-Flight 3D Imaging based on a SPAD-TDC Pixel
Arrays”  ~25 ps resolution

5
TPIX4 VCO
• Improve TOA resolution Latching of internal VCO phase
• 5-tap VCO has 10 unique phases
• TOA reconstructed from:
1) TOA @ 40MHz (N-bits) tbd stage Phase[4] Phase[3] Phase[2] Phase[1] Phase[0]

2) FTOA @ 640MHz (4-bits) Init 10 1 0 1 0 1


(Enable=0) 1 1 0 1 0 0
3) UFTOA @ 6400MHz (4-bits)
2 1 0 1 1 0
3 1 0 0 1 0
4 1 1 0 1 0
5 0 1 0 1 0
6 0 1 0 1 1
7 0 1 0 0 1
8 0 1 1 0 1
9 0 0 1 0 1

Vcntr
VDD

Enable
NAND

Phase[0] Phase[1] Phase[2] Phase[3] Phase[4]


6
Possible approach for Timepix4: DLL
N+1

 Control the phases that arrive in the super


N
N+2 pixels of a double column using a voltage-
controlled delay line tuned by a delay-locked
N+3
loop (DLL)
N-1

 In lock, Td = Tref/2N, i.e., the buffers’ delays, Td,


2N-2
are a well-defined fraction of the reference
4
clock (can be calibrated offline)
 Robust to PVT variations in a first-order analysis
3
2N-1
 The possibility of using a single PD/CP/LF for
the whole matrix is under investigation
2N  Power and area benefits
2

 We are also investigating the possibility of


implementing the control loop in the digital
1
domain
Phase Charge Pump &
Detector Loop Filter
 Scalable and easily portable to more advanced
CKref Vctrl technology nodes
7
Timepix4 (also) as a photon counting chip?

• CRW (Continuous Read-Write):


• Packet based and zero-suppressed
• 2 x 12..16 bit counters per pixel:
• PC / PC
• iTOT / iTOT
• Possible operation modes:
• Controlled by external pin (as in Medipix3)
• Data controlled  Data packet is generated after counter reaches
an on-pixel programmed value
• Single counter:
• Packet based and zero-suppressed
• 24..32 bit  PC or iTOT
• Auto Threshold Calibration?
• On-pixel gain (TOT) calibration?
8
Timepix4 arrangement
26 mm
24.2 mm
Timepix3
Timepix4
256 x 256 pixels
1 superpixel  2x4 pixels 55um x 55um
220x64 superpixels (65 Kpixels)
436 x 581 pixels
32 mm

28.16 mm
55µm x 55µm x1
(253316 pixels)
440 x 512 pixels
55µm x 55µm
(225280 pixels)
x3.86 x3.4

9
Large Timepix4:
TSV (4-side buttable) vs WireBonds (2-side buttable)

24.2 mm 24.2 mm

Timepix4-WB Timepix4-TSV

1 superpixel  2x4 pixels 1 superpixel  2x4 pixels


220x64 superpixels 220x64 superpixels

28.16 mm
30.16 mm

440 x 512 pixels 440 x 512 pixels


55µm x 55µm 55µm x 55µm
(225280 pixels) (225280 pixels)

10
64B/66B encoding
• IEEE 802.3ae-2002 amendment which introduced 10 Gbit/s
Ethernet
• 3.125% bandwidth overhead
• Timepix3  25% in 8b/10b
• 64 bit event data (scrambled)
• Timepix3  48 bit (25% less)

11
Data Packet in Event Data Driven
Timepix3 Timepix4
Event Data bits 48 64
Status 4 4
Address 16  65536 pixels 18  225280 pixels
TOA 14  409.6 µs @40MHz 18  6.55 ms @40MHz
FTOAr 4  1.5625 ns 4  1.5625 ns
FTOAf - 4  1.5625 ns
UFTOA - 4  ~200ps
TOT 10 12

Status[63:60] Address[59:42] TOA[41:24] UFTOA[23:20] FTOAR[19:16] FTOAF[15:12] TOT[11:0]

12
Serializer1

Readout

Digital readout

Digital readout
1 2
& configuration

Analog biasing
• Readout (based in a 40MHz clock): Analog Periphery

• (Fast) 2x 5.12Gbps serializers:

Digital readout

Digital readout
• One on each digital domain 3 4
• 2.56Gbps, 1.28Gbps also available
• 64b/66b encoding
• (Slow) 2x 640Mbps SLVS drivers: Serializer2
• One on each digital domain
• 320Mbps, 160 Mbps, 80 Mbps and 40 Mbps also available
• 64b/66b encoding

• Slow Control:
• SPI bus addressing up to 80MHz:
• Control operation
• Write and Read registers
• Write and read pixel configuration

13
Readout Rates
Timepix3 Timepix4
Readout Bandwitdh Up to 5.12 Gbps Up to 20.48 Gbps
Links x8 LVDS x4 CMOS serializers
Encoding 8b/10b 64b/66b
Number bits in full frame > 3.9 Mbits > 14.8 Mbits
Readout full Frame 768 µs 726 µs
Maximum chip hit rate 85.3 Mhits/s 310.3 Mhits/s
Maximum hit rate area 43.0 Mhits/s/cm2 45.5 Mhits/s/cm2

14
Plans
• Original Plan was submit Medipix4 and Timepix4 simultaneously on the
same reticle
• Large Timepix4 is a recent proposal  Not approved by collaboration:
• The photon counting (Synchrotrons…) requires a fast frame readout mode
• Studies on different key blocks already started:
• Pixel input RDL (Edinei/Xavi)
• Re-shaping of IP blocks to fit narrow layout constraints (Viros)
• DLL clock distribution (Edinei)
• On-pixel VCO and serializers (Nikhef)

• Medipix4 will come later maybe in 130nm

• Designers team from CERN and NIKHEF

15
Conclusions
• Proposal for Timepix4:
• Particle tracking:
• ~200ps TOA
• ~1KeV FWHM TOT resolution

• Photon counting (deadtime ~300ns@20Kev):


• 1 counter 24..32 bits
• 2x counter 12..16bits in CRW zero supressed

• 1 threshold with 5-bits threshold tuning


• >500 e- minimum threshold
• Maybe: Auto threshold calibration and/or ToT gain tunning

• 4-side buttable
16

You might also like