UCHOLA design tang - University of...

19
Dual HOLA Design for FTK Project Mel Shochet At K li Anton Kapliy Fukun Tang Daping Weng Daping Weng Presentation for Dual HOLA Board Design Review University of Chicago Enrico Fermi Institute Jl 18 2011 July 18, 2011 F. TANG 1

Transcript of UCHOLA design tang - University of...

Page 1: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Dual HOLA Design for FTK Project

Mel ShochetA t K liAnton KapliyFukun Tang Daping WengDaping Weng

Presentation for Dual HOLA Board Design Reviewg

University of ChicagoEnrico Fermi Institute

J l 18 2011July 18, 2011

F. TANG 1

Page 2: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Dual HOLA Card

Dual HOLA for ATLAS DAQ and FTK

MezzanineData Port

Optical  Optical link1 CERN HOLA

ATLAS DAQAltera Cyclone IV

EP4CGX15

Dual HOLA  Card

FPGAWith built in

pTransceiver

HOLA LDC

built‐in Serdes Optical 

Transceiver

Optical link2CERN HOLA LDCLDC

FTK

Brief  Design Specification:Data Input at Data Port: 32b@40MHz = 1.28GbpsData output to Serdes:   32b@50MHz = 1.6GbpsLi R t f O ti l Li k ( ith 8B/10B) 2 0GbLine Rate of Optical Links (with 8B/10B) = 2.0Gbps  

Link1 (to ATLAS DAQ): Full S‐Link protocol implemented on transceiver.

Links designed to run 1.6‐2.5Gbps

p pLink2 (to FTK): 

Transmitter: Full S‐Link protocol implementationReceiver:     Only “Flow Control” used. 

Number of Dual HOLA card in production: 250F. TANG

2

Page 3: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

UCHOLA Schematic (1/6 ‐‐‐ Top Level)

Top Level Schematic and FPGA Power Pin Connections

6 LV Powers required for FPGA:• P3V3• VCCA2V5• VCCINT1V2• VCCLGXB1V2• VCCHGXB2V5• VCCDPLL1V2

FPGAFPGA

F. TANG 3

Page 4: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

UCHOLA Schematic (2/6 –FPGA Configuration)

FPGA

• Serial EPROM Configuration (EPCS4)

F. TANG 4

Page 5: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

UCHOLA Schematic (3/6 ‐‐‐ S‐Link Data Port & FPGA IOs)

Power_on Reset

FPGAFPGA

• Slink Data Port and FPGA I/Os

50MHz Transceiver Reconfig. Clock 

F. TANG 5

Page 6: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

UCHOLA Schematic (4/6 ‐‐‐Clocks & Optical Transceivers)

• Optical Transceivers

T i R f Cl k G

FPGA

• Transceiver Reference Clock Generators

F. TANG 6

Page 7: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

UCHOLA Schematic (5/6—Low Voltage Powers)

• LV  DC/DC Module• 3.3V to 2.5V• 3.3V to 1.2V

F. TANG 7

Page 8: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

UCHOLA Schematic (6/6—Low Voltage Powers)

• LV DecouplingLV  Decoupling

F. TANG 8

Page 9: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

PCB LAYER STACK‐UP & IMPEDANCE CONTROL

Stack up and trace routing are designed for the best signal integrityStack‐up and trace routing are designed for the best signal integrity

F. TANG 9

Page 10: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Dual HOLA Prototype PCB (Top View)

Board Size: 149 x 74mm 

F. TANG 10

Page 11: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Dual HOLA Prototype PCB (Bottom View)

Board Size: 149 x 74mm 

F. TANG 11

Page 12: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Static & BERT Tests

• Low power consumption: [email protected] (2 Watts Total) with 2 optical transceivers.

• BERT: Error free with 1.35E15 data tested for dual channels.

• Test setup: With CERN’s PCI/HOLA test setup a 7dbTest setup: With CERN s PCI/HOLA test setup, a 7db multimode 850nm attenuator is inserted in each optical fiber for both transmitters and receiversoptical fiber for both transmitters and receivers.   

• See Anton’s report for details

F. TANG 12

Page 13: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Dual HOLA Production Board

Two modifications have been made from the prototype board:p yp

(1) Incorporating with S‐link HW specification, move two optical transceivers 10mm to themove two optical transceivers 10mm to the right edge of the card.

(2) Disconnect Power‐on Reset Enable signal from 1.2V power supply.1.2V power supply.

F. TANG 13

Page 14: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Dual HOLA Top Side View

Component height clearance area

These 2 connectors will not be installed in production version.

41mm

F. TANG 14

Page 15: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

HOLA Hardware Clearance SpecificationRefer to S‐link specification page 44

Slink HW specification: 31mm from  the right edge of the board)

31mm

41mm

Prototype card violates S‐link clearance specification byPrototype card violates S‐link clearance specification by extending optical transceivers 10mm inside the card!

F. TANG 15

Page 16: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Signal Integrity

• We did not see any noticeable affects on the signal integrity from SI simulation by moving g g y y gthe optical transceivers 10mm to the right edge of the board (FPGA transceivers nowedge of the board. (FPGA transceivers now only drive differential lines of 31mm with a rate of 2Gbps (21mm in the prototype card)rate of 2Gbps. (21mm in the prototype card).

F. TANG 16

Page 17: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Removing “Power_good” (1.2V) Enable Signal

1 2V is not high1.2V is not high enough to enable “power‐on RESET”. We disconnected it in e d sco ec edprototype card test.(pin‐3 has internal pull‐up)p p)

No connection to 1.2V in the production boardsproduction boards

Ab h ill t ff t f ti l f !Above change will not affect any  functional performance!

17F. TANG

Page 18: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Summaryy

• Approval of Production Runpp

• Purchase order of PCB/Components

• PCB assembly

• Production board testsProduction board tests

F. TANG 18

Page 19: UCHOLA design tang - University of Chicagoedg.uchicago.edu/~tang/ATLAS_FTK/UCHOLA_design_tang.pdf · 2011. 7. 7. · Removing “Power_good” (1.2V) Enable Signal 12V1.2V is not

Acknowledges

• Special thanks to Stefan Haas of CERN for his constant psupports and advices on the schematics and PCB designs, as well as test setups.  Also thank him for providing us HOLA source and test bench firmware codes.HOLA source and test bench firmware codes.

• Thank Ted Liu of Fermilab for his very valuable advices on the Dual HOLA prototype board tests.

• Thank many others including Arrow Electronics and Altera• Thank many others including Arrow Electronics and Alteratechnical support teams for the valuable discussions and assistances for the FPGA timing model simulations.

F. TANG 19