daq hardware status - overview r. stokstad
DESCRIPTION
DAQ Hardware status - overview R. Stokstad. DOM Main Board Schedule (Minor) Design (Przybylski) Firmware (Stezelberger) Testing (Goldschmidt) DOR Board (Sulanke) Master Clock Unit (Nygren, Sulanke) ---------------------- Time Calibration, Cable measurements - PowerPoint PPT PresentationTRANSCRIPT
DAQ Hardware status - overviewR. Stokstad
• DOM Main Board– Schedule (Minor)– Design (Przybylski)– Firmware (Stezelberger)– Testing (Goldschmidt)
• DOR Board (Sulanke)• Master Clock Unit (Nygren, Sulanke)
– ----------------------
• Time Calibration, Cable measurements(Stokstad, Sulanke, Hardtke)
Individual presentations for the above to be in Mons proceedings
Rev 2 board
Icecube DOM MB Status and Development
• April – Oct '03– Rev 2 test and development (22 boards)
• Oct – Nov ’03 – Rev 3 fabrication and test (16 boards)
• Nov ’03 – Mar ’04– Rev 4 fabrication and test (60 boards)
• March – June ’04 – Rev 5 for deployment (420 boards)
Schedule
• October – November ‘03• First 4 cards due Oct 22, 12 more to follow 5 days
after approval– Changes from Rev 2 address memory performance at
temperature and signal quality issues– Testing at LBNL will include manual tests and STF
testing– Temp cycling of bare and loaded boards
• +65, -40degC 10 cycles
• First 4 cards to UW, later to LBNL• 8 cards to UW, 8 cards to LBNL
Rev 3 status
• November – March ‘04– Design modifications as indicated by Rev 3
tests and internal review– Final parts selection for reliability– 60 cards to be fabricated– “Qualification test” at UW
• Longer term temperature tests• String operation tests• Some boards used for development at LBNL
Rev 4 plans
Rev 5 plans• Initial production review November ‘03
– Based on Rev 3 and Rev 4 tests– Earlier than usual to allow long-lead parts purchasing
• Production readiness review - January ‘04– Based on initial Rev 4 tests– Qualification testing– Vendor qualifications– Fabrication process approval
Rev 5 Production (March – June ’04)• Full production and Q/A procedures in place• All parts as selected for reliability• ATWD production run including qualification
tests• Any differences between Rev 4 and Rev 5
require formal Engineering Change Procedure and re-qualification testing
• Fabrication begins March ‘04• Testing begins April ’04• Delivery April - June ‘04
• EPXA4 Baseline (The larger FPGA)• Hi-Reliability Part Substitutions
– Primary Oscillator; Corning (Hi-Rel 2560A-0009)
– High-Rel DC-DC Converter; Power-One brand
• Low impedance Power Distrib.to SDRAM• 2 x 12 Channel ADCs vs. 1 x 8 Channel
for monitoring • Q/A Tests Added to Plan; Coupons on PCB• Noise Related Layout Changes• Fabrication procedure changes
Some Changes made in Rev 3
•On-Board Flasher Performance Tuning- UV LED gives < 10ns FWHM pulse- Blue LED gives ~13 ns FWHM pulse
•x0.25 x 2 x 8 gain for ATWD PMT inputs•Extend Input Clipping Range [+2.4..-3.0]•Stronger ATWD Drivers•Optimize ATWD Driver Output Impedance to minimize ringing•fADC channel gain adjustment for 12 bits
Tuning for Physics Performance
DOMMB Block Diagram
Firmware• CPLD (lowest level programmable logic device)
Almost finished. Smallr changes needed for Rev 3 and to make the design more robust.
• FPGA – STF 99% Done for testing– ConfigBoot Preliminary Version minimal boot– IceBoot Preliminary Version normal boot– DOMAPP ~50% Done for data taking
For software development purposes the SFT FPGA can be used for the ConfigBoot and the IceBoot FPGAs
DOR card• Firmware Status• Production Status, Rev. 0• Planned Production, Rev. 1
Data Buffer
SRAM2 x 256Kx16
Cable Interface #1
Cable Interface #2
Cable Interface #3
Cable Interface #4
PCI-Core
FLASH1M x 8
AlteraPLD
EPM7064
Config JTAG
Altera FPGA
EP20K200E
JTAG
PCIBus
DOM 1..4
JTAG
Clock
96 V
DOM 5..8
Reload
DOR, DOM Readout card 13
ICECUBE Meeting, Mons October 2003
FPGA FLASH SRAM
96VDOM Power
Clock & Time String port
JTAG
PLD
Comm. ADC
Comm. DACQuad Cable Con.
Comm. Rec.
Power Switch
DOR Rev. 0, Main Firmware Status• PCI Bus control including DMA (..120 MB/s) &
Interrupts• Parallel control of 2 wire pairs / 4 DOMs• 1MBit/s serial data rate, (50KB/s per DOM)• DOM_a/b polling• Automatic cable length adaptation• Software initiated but firmware controlled time
calibration• Three running firmware versions:
– DOR_TEST, after production test– vDOM, includes two virtual DOMs for Linux driver test– DOR, final version, under development
Production Status• Rev. 0, 10 DORs running at : UW 2x,
Bartol 1x, LBNL 7x• 20 more in production now, ready in Dec.
2003• 4 stay at DESY, 16 go to ?
Planned Production, Rev. 1• Redesign ready in January• first test in Feb., production of 60 if o.k.• 60 ready in April 2004• 60 DORs -> sufficient to control 480 DOMs
DOMMB Test Plan• Select well functioning DOM MBs (including
delay board) suitable for integration into DOMs to be deployed in IceCube/IceTop.
• For boards which are not well functioning, provide information to aid understanding, debugging and fixing.
• Provide some DOMMB characterization: optimum running parameters, etc.
• Be able to handle an expected “steady state” flow of 55 boards/week.
Initial Tests• Visual inspection, Power to board• Load “released” firmware and software in
dedicated setup (1 board at a time)• Get “boot prompt” and tests that cannot be done in
DOR-based setup (if any)
• Multiple heating and cooling cycles (+65C to -40 C) to catch infant mortality cases.
• Boards powered and some testing during burn-in • Power cycles during burn-in • Full operational test at the end to identify bad boards.
Burn-in
Testing Setup
• STF (Simple Test Framework) based: STFServer (DOMMB) and STFApp (Test-Control PC)
•All tests defined with Pass/Fail and Input and Output Parameters
•All Info saved to database (db)
•A db query program to see waveforms, output parameters, histograms, etc.
Example of STF test atwd_pulser_spe test
Parameter Type Valueatwd_sampling_speed_dac input 850atwd_ramp_top_dac input 2097atwd_ramp_bias_dac input 3000atwd_analog_ref_dac input 2048atwd_pedestal_dac input 1925atwd_ch0_clamp input 0atwd_chip_a_or_b input 0atwd_channel input 1pulser_amplitude_uvolt input 50000pedestal_subtraction input truen_pedestal_waveforms input 50loop_count input 1000triggerable_spe_dac output 496atwd_baseline_waveform output plotatwd_waveform_width output 4atwd_waveform_amplitude output 65atwd_expected_amplitude output 50atwd_waveform_pulser_speoutput plotpassed output truetestRunnable output trueboardID output 3c62a6c4
Average of 1000 waveforms in 1 sec.
Master Clock Unit: Function
• Create and distribute stable 20 MHz source to all DOM Hubs in IceCube DAQ.– DOR cards mirror MCU time, calibrate DOMs
• Link “IceCube Time” to GPS time .• Provide robust real-time time verification.
– Motivation: detect any error condition in less than 1 sec
Some System Requirements• The DOM Hub will be capable of stand-
alone operation, (use the DSB 20 MHz clock)– This facilitates many testing activities
• The IceCube MCU will be capable of full operation in the absence of GPS input– Could happen that GPS signal interrupted
• IceCube MCU operation shall not require that all DOM Hubs are operational. – Independence of strings during
commissioning.
Accuracy and Precision•GPS satellites are synchronized to the US Naval Observatory.•The USNO atomic clocks maintain time to < 2x10-15 day-1.
•GPS stability degraded to ~4x10-8 second-1 by triangulation errors.•DOM local clocks stable to < 1 x 10-10 second-1.
GPS much less stable than DOM local clocks (seconds scale)
•However, GPS will display better long-term stability (hours scale)
Master Clock unit must employ a source with better short-term stability than GPS, yet must still match long-term GPS time.
Use commercial atomic clock “linked” to GPS over long-term
Master Clock Unit (MCU) Status
• Draft Requirements Document exists• Specific Implementation proposals have
been made by Sulanke and Przybylski.• Plan is that Sulanke will design and
fabricate this subsystem at DESY.