ESODAC Study for a new ESO Detector Array Controller - PowerPoint PPT Presentation

About This Presentation
Title:

ESODAC Study for a new ESO Detector Array Controller

Description:

ESODAC Study for a new ESO Detector Array Controller – PowerPoint PPT presentation

Number of Views:35
Avg rating:3.0/5.0
Slides: 14
Provided by: mmeyer
Learn more at: http://www.eso.org
Category:

less

Transcript and Presenter's Notes

Title: ESODAC Study for a new ESO Detector Array Controller


1
ESODACStudy for a new ESO Detector Array
Controller
2
Introductionand Key Points
  • Idea is to realize a modular system with a basic
    Front-end unit for a four channel system on one
    card of standard VME 6U size.
  • Power Consumption on Front-end less than 10
    Watts.
  • Very low noise design anticipated.
  • Add on cards are 32 channels on a board of the
    same size and additional 10 Watts of power
    consumption.
  • ( Number of channels can be remotely set ).
  • There will be no processor on the front-end side.
  • Data distribution on backend side possible free
    topology .
  • Connection between Back and Front-end only by
    fibers.
  • Weight below 1 Kg.
  • This system should not require active cooling.

3
System Design
4
System Block
5
Back-end
  • Function is based on the XILINX Virtex Pro FPGA
    XC2VP7 FF 672 .
  • Back-End PCI is a 64 Bit PCI board downscale to
    32 Bit PCI possible.
  • FPGA contains PCI interface, protocol engine PCI
    to transceivers
  • for communication and data reception and
    RocketIO transceivers.
  • Direct interface from FPGA to PCI without glue
    logic.
  • Independent PCI master and PCI slave.
  • Communication and data transfers all on serial
    link.
  • Data rate on one channel between front and
    back-end 200MByte
  • More bandwith possible ( one FPGA contains 8
    transceivers space limit for PCI card size
    might be four ).
  • Routing capability by high speed fibers e.g. to
    VME or other systems.
  • Selective data reception and routing possible (
    like in old times shift and add with IRACE ).

6
Front-End Basic Module
  • The front-end Basic Module is based on the XILINX
    Virtex Pro FPGA XC2VP7 FF 672.
  • Main functions of this module are
  • Communication
  • Data transfer
  • Sequencer
  • Clock driver, biases and associated DACs
  • Four data acquisition channels ( each either 16
    or 18 Bit ADCs ) and preamps
  • Utilities ( Markers, Synchronization )
  • Telemetry

7
Front-End Basic Module
  • Design ideas are
  • Communication and data transfer to the back-end
    is handled with the FPGAs Gigabit transceivers.
  • Protocol engine to serial link contained within
    the FPGA.
  • Sequencer is completely contained within the
    FPGA.
  • The digital clock driver lines of the sequencer
    connect out of the FPGA without glue logic to the
    clock driver switches. Clock driver alternatives
    have to be evaluated, one possibility is type
    used in IRACE.
  • Sequencer will contain provisions for high speed
    external trigger inputs and status outputs.
  • The ADC outputs of the four acquisition channels
    connect without glue logic to the FPGA due to the
    high pin count available there. Favorable ADCs
    are the AD76xx types from Analog Devices. The
    preamplifier is fully differential, input range
    will be /- 2.5V. There will be no clamp/sample
    implemented in the analog chain.
  • Provisions will be taken to incorporate different
    ADC types ( e.g. high speed lower resolution
    types) .

8
Front-End Basic Module (cont)
  • Design ideas are
  • Synchronization will be foreseen to additional
    Basic Modules (more clocks, biases).
  • For high-drive clocks (e.g. high capacitive loads
    by big arrays) provisions to external drive
    modules must be foreseen.
  • Connection to the additional multi channel AQ
    modules is a connection by fiber or copper on
    high speed links with FPGA transceivers. This
    would give a very low data bus noise coupling to
    the analog part. Communication and set-up of
    front-end modules also runs on this link.
  • Alternatively possible could be connection by a
    64 Bit bus. The bus principle is a daisy chain as
    in IRACE. This bus will directly sort out of the
    FPGA without glue logic. Different driver
    possibilities (LVTTL, GTL ) are provided within
    the FPGA. A low speed serial bus connecting all
    front-end modules needed for set-up and tests
    would also sort out of the FPGA.
  • Provisions for adaptation to detector ASICs must
    be foreseen ( at present missing protocol
    definition of ASIC).
  • Telemetry will be on board.
  • Digital Outputs for shutter control, test markers
  • Monitoring of clocks, biases and detector signals
    will be on basis of an external module.

9
Front-End Basic Module
Connectors for detector signals, clocks and
biases have to be defined.
10
Add on cards
  • AQ Module ( 16 Bit )
  • The front-end AQ Module is based on the XILINX
    Virtex Pro FPGA XC2VP7 FF 672.
  • On board are 32 acquisition channels on 16 Bits.
  • ADC outputs of the acquisition channels connect
    with little glue logic on a bus structure to the
    FPGA (tests on crosstalk are needed).
  • Favorable ADCs are the AD76xx types from Analog
    Devices.
  • The preamplifier is fully differential, input
    range will be /- 2.5V. There will be no
    clamp/sample implemented in the analog chain.
  • A low speed serial bus for set-up and test must
    be implemented. AQ modules are always slaves on
    this bus.
  • AQ Module ( 18 Bit )
  • On the basis of the 16 Bit AQ module a 18 Bit
    version can be build. Same layout and printed
    board might be possible.
  • High Drive Module
  • External Monitor

11
Status
12
Back-end
  • Back-end Prototype ( PCI 64 Bit board ) designed
    in scheme and layout.
  • FPGA design (Communication and Scatter-Gather DMA
    in 32 Bit functional)
  • Fiber optics system realized as functional study.

PCI 64 Bit Design ( DXP Image )
13
Front-end
  • Fiber optics system realized as functional study.
  • Sequencer realized as functional study.
  • Acquisition module ( Adcs and preamp ) are
    already tested as a piggy-back back-up monolithic
    replacement for the Analogic hybrid ADCs. Preamp
    will have to be revised (single ended gt
    symmetrical).
  • Clock driver could be realized like IRACE module
    alternatives have to be studied.
  • If not implemented by high speed links
  • Bus system mixture between IRACE ( daisy chain)
    and Compact PCI.
  • Neither studies nor design.
  • Power supply design not studied. Separate and
    remote from Front-end desired.
Write a Comment
User Comments (0)
About PowerShow.com