Title: Electronics Status
1Electronics Status
- Philippe Farthouat
- CERN
- ATLAS Plenary June 25, 2004
2Outline
- Production of ASICs
- Production of Front-end Systems
- Production of Back-end Systems
- Production of Power Supplies
- TTC issues
- Radiation hard electronics
- Infrastructure
- EMC in ATLAS
- Next developments
- Summary
3Production of ASICs
4Production of ASICs (cont)
5Production of ASICs. Comments
- All ATMEL (DMILL) wafers have been delivered
- TRT ASDBLR
- 55 parametric yield (Enough for the complete
detector) - Test injection capacitor failures observed
- LArg wafers
- Good yield for all the chips
- Dispute with ATMEL not yet finished but out of
our control - Two ASICs on the critical path
- RPC CMA which required a major redesign because
of a change in the RPC cable lengths (input
adjustment pipeline too short) - Submitted now and expected end of the year
- TGC Slave Board ASIC (SLB)
- Last iteration received this summer needs a fix.
Resubmitted and expected this month
6Front-end Systems (1)
- All systems are either in full production or
starting it - PRR done for most of them
- To be noticed
- Liquid Argon Calorimeter
- 58 Front-end crates
- More than 2000 custom boards
- Very complex boards (in particular the FEB)
- Tile Calorimeter
- Production to be finished before next March
- MDT Chambers
- All the mezzanine boards produced
7Front-End Systems (2)
- TRT
- Barrel front-end boards
- Available for all barrel modules
- 14 layers with buried vias
- fpBGA on both sides
- End-cap front-end boards
- Pre-production on-going
8Front-End Systems (3)
- Test stand of the LArg FEB
9Back-end Electronics (ROD)
- Pixel-SCT PRR passed
- Production to start now
- TRT full scale prototype available
- Successfully used in the test beam
- Change in the read-out granularity to be done to
allow a complete staging of the C-Wheels
electronics - Double-density ROD to save money
10Back-end Electronics (ROD)
- LARG/Tile
- PRR done
- MDT ROD
- On-going redesign based on FPGA SHARC
- Design finished
- Lay-out to be done
- Expected prototype end of this year
11Power Supplies (1)
- Radiation hard STm Voltage regulators
- Positive and Negative versions available
- Available through CERN store
- Last negative version not yet delivered
- CERN store enquiring
- 2004 should have been the last year to order
these components. Well still be able to order
next year
12Power Supplies (2)
- Pixel
- Power supplies (LV and HV) in USA15 and US15
- ST regulators at patch panel 2 level
- FDR last June
- Commercial devices
- Good candidates for LV and HV from WIENER, CAEN
and ISEG - SCT
- Power supplies in USA15 and US15
- Production on-going
- HV modules in Krakow (problem with a failure mode
being fixed) - LV modules in Prague
- TRT
- LV in UX
- Good candidates from Wiener or CAEN
- Tendering process started Market survey done
and tender to be issued soon (by PH-ESS) - HV in USA15
- Price enquiry out soon to about 10 companies
- Several candidates (CAEN, Dubna, ISEG)
13Power Supplies (3)
- LArg LV
- DC-DC converters close to the calorimeter
- 280V DC input
- 250 kW total
- Production starting
- End in July 2005
- Very long development to obtain a device
sufficiently radiation-hard - LArg HV
- Power supplies from ISEG
- Production done
- A problem with broken capacitors being
investigated
14Power Supplies (4)
- Tile HV
- Bulk supply and distributor in the drawers
- Production on-going
- Tile LV
- DC-DC in the fingers
- Design finalised
- A few boxes in production
- Still some radiation tests to be done
- PRR expected late fall
- Muon System
- Same scheme for all chambers
- Commercial power supplies in UX15
- CAEN and Wiener are very good candidates
- Granularity adapted to the cost
- Several chambers/layers per power supply (both HV
and LV) - Tender in preparation
15TTC Issues
TiTiSee
16TTC issues
- TTC components in the pit
- It is the responsibility of level-1 team to
provide (and pay for) to each system the TTC
components from the TTCvis up to the end of the
fibres - TTCvi, TTCxx, Fibres, TTCoc
- It is responsibility of level-1 to provide to
each system the pin diodes, TTCrx and QPLL. These
devices are paid by the systems - TTCrx, Pin diodes available (56 ChF and 9 ChF)
- TTC components for test systems
- Test beam, commissioning stands, lab set-up
- Requests through the electronics co-ordinators
- Each sub-system should have given its needs about
a year ago - Some needs are not yet covered
- Last order of components
17QPLL Problems (1)
- Because of the protocol used by the TTC, the
extracted clock at the receiving end (TTCrx) has
some jitter - 30 100 ps rms depending on traffic on the
B-Channel - This jitter is too high for the Gbit link
serialisers or some LVDS serialisers - QPLL is a PLL driven by a crystal oscillator and
it filters the jitter on the clock delivered by
the TTCrx - Jitter less than 100 ps peak-to-peak
- It recently appeared that in about 25 of the
cases a huge jitter (3 ns) appears at certain
frequency and/or temperature - Low frequency drift ? does not disturb the Gbit
Links - Destroys any timing measurement
- Very serious problem for the ATLAS LArg
- So far only ATLAS LArg has reported on this
problem
18QPLL Problems (2)
Bad one
Good one
19QPLL Problems (3)
- On-going effort to try and understand this
problem - Extensive measurements made by the engineer in
charge of the TTC with the help of Stefan Simion
(LArg) - Two engineers in the CERN MIC group running
extensive simulation and trying different
solutions - Quite a lot of power injected in the crystal
- Addition of extra capacitance or resistance cure
the problems but - Addition of a sufficiently high value resistor
increases the high frequency jitter at a level
which could disturb the Gbit Links - Addition of capacitors reduces the working
frequency range (still OK for us) - An engineer from the company which delivered the
crystals is investigating - No results yet and interaction a bit slow
- LArg (Nevis) has contacted another company
- Samples measured at CERN with QPLL. All (almost)
OK (only 3 devices measured) - Problem to be solved asap as it blocks the LArg
front-end boards production
20Radiation Hardness (1)
Report from Francis Anghinolfi
- Components pre-selection ? 99 achieved
- Missing
- Some Pixel and TRT PP2 electronics
- One FPGA on MDT CSM Board
- RPC Coincidence Matrix (ASIC)
- Production qualification ? 72 achieved
- Missing
- Pixel Production going on
- TRT PP2
- LArg Optolink
- MDT CSM
- EC MDT Alignment
- TGC 50 production
- RPC 50 production
Fully Completed - SCT - CSC - TILE CAL - Barrel
MDT Alignment
21Radiation Hardness (2)
- Power Supplies
- All components pre-selection achieved
- No production phase presently started up
- LARG PS PRR done 2 pre-series units Nov04
- TILE CAL PRR Q404
- Muon LV/HV commercial devices
- Validation of the production to be organised
22Radiation Hardness (3)
- We should be careful about qualification of Power
Supplies (schedule) - It appears some late electronics developments
- E.g. Pixel and TRT PP2 boxes
- Single component added at the last minute on PCB
should be identified - E.g. protection devices
- Verification Process during installation or
commissioning (incl. safety/monitoring devices,
crane control, cooling/cryogenics control, etc )
to be defined
23Infrastructure
Report from Georges Blanchot
- USA15. Levels 1 and 2, transformers and
switchboards - 250 racks are in place, ready to receive power
and equipment - Power distributed with CANALIS bus-bars 800A
(500 kW) primary, 160A (170 kW) secondary - On the critical path now
- UX15. HS structures, levels 0 to 8
- Need to specify power requirements and to update
the EMDH database - First round with responsibles of racks done
- Standard switchboards and cabling
- Not on the critical path now
24Power to Racks in USA15
- The power distribution is made of
- Canalis busbars
- Carry the power to the rack
- Control bus
- Control bus to communicate with the Twido box
- Twido box
- On/Off and other monitoring functions with a
TWIDO PLC connected to control bus - Distribution Box
- 3 plugs for racks, mono or triphased versions,
with remote control of the switch - 1 auxiliary plug
- 1 plug for ventilation unit
25USA15 Racks Room
- Where required, temporary plugs will be
supplied ASS on level 2, DCS on level 1. General
services plugs are already in place - Some racks require UPS/Diesel, installed at a
later stage. Must run on temporary plugs DCS,
DSS, ASS
- First branch to get power level 1, 4 rows,
including Tilecal, end october - Once working, all other branches to be installed
- Next milestone LArg needs its racks in February
05 - Cooling parts (turbines, deflectors, heat
exchangers) january 2005
26UX15 Racks
- Need to define power requirements.
- Need to form power groups of racks.
- Too little information on EMDH first round of
enquiries done by email.
27Summary Infrastructure
- Electrical distribution to USA15 in progress
- Power is delivered on the basis of data available
in EMDH - ? System must make sure this is up to date
- Tilecal to get power by end october
- ? Others by end november
- Cooling parts ordered and to be delivered (and
installed) in January 05 - Confirm power needs for racks in UX15
- Rack configuration in EMDH too
- ? Systems must fill in the Rack Wizard database
here also - Preliminary data available
28ATLAS EMC Policy
Reference document is available
at https//edms.cern.ch/file/476490/1/ATLAS-EMC-P
OLICY.pdf Additional Information is available
at http//atlas.web.cern.ch/Atlas/GROUPS/FRONTEND
/EMC/
- EMC Procedures in ATLAS
- Safety and Installation Report.? The setup must
in all cases comply with the CERN safety rules. - EMC Report.? The required noise performance must
be achieved.? The system must not be a
disturbance for other systems. - Commissioning.? Insure that installlation is
done as defined in steps 1 and 2.
29EMC Issues Specific to ATLAS
- Grounding
- Large dimensions of the experiment ? long
grounding wires - Ground wire is a safety requirement
- Difficult to insure equipotentiality between
detectors and their racks located 100 meter away - Common mode noise
- Results from the long cables and limited ground
equipotentiality - Common mode noise radiates more than 103 times
than differential mode noise - Noise coupling mainly caused by CM emissions of
cables - Power cables to switched mode power supplies and
other power equipment (motors, pumps) - Digital transmission (fast transients) copper
links, such as fieldbuses, serial buses, relays - Near field
- Couplings mostly to happen in the near field
region inside cable trays - Distinguish electric (capacitive) and magnetic
(induction) near field couplings - Type of coupling conditions the type of shielding
requirement - Long Conductors
- Results in CM/DM conversions along the cable
(multiconductor transmission lines analysis) - Results in noise amplification and attenuation
for given frequencies. - Capacitive couplings between systems
- Electrically isolated detectors share large
surface gaps capacitive coupling used by CM
currents.
30Measurements CM Emissions
31Measurements Immunity
32EMC Summary
- ATLAS EMC Policy
- Quality and risk management issue
- Establishes methods and procedures to insure the
systems electromagnetic compatibility in the
experiment environment - Compliance requirements specific to the ATLAS and
CERN environments - Procedures
- Clear description of the setup
- To insure compliance with safety rules
- First document from LArg in approval process
- To define the valid configuration for EMC
measurements - EMC Measurements
- Specific for each system
- Started this summer (MDT in H8, Tile Cal)
- To understand how noise couples into a system and
affects its performance, within the established
compatibility limit - Guidelines
- On grounding configurations, coupling modes,
cable shields, shield terminations routing
paths, test methods, tools
33Future developments
- Although still very early, future developments
for LHC upgrades have to be considered - Main points to be looked at
- Which technology(ies) could be used
- Power management
- Does not aim at being a complete study of the
situation but at pointing potential problems
which may influence our working habits - Most of the material presented from Sandro
Marchioro
34Technologies
- One of the possibilities is to continue very good
relations with DSM vendors and to start looking
at CMOS very deep sub-micron technologies - 130 or 90 nm
- First evaluation of 130 nm shows that the
radiation hardness is very good - Linear transistors look very promising with minor
weaknesses - No guard-ring needed
- Sensitive charge smaller, higher SEU sensitivity
- NB without enclosed transistors, error rates can
be considerably higher than for present 0.25mm
designs - Latch-up not observed and not expected to be a
issue - However there are difficulties
35Cost
- We clearly cannot afford hundreds of
developments and prototypes
36Developments in 0.25µ
37Cost comparison
From Sandro Marchioro
Assuming one iteration and ABCD type chip
- Clearly points towards
- Very few different designs produced in large
quantity - A single iteration
38Power
- These technologies leak
- Without careful design the power will not
decreased - Nor the current (Vdd is going down)
39Future Developments
- Cost of the new technologies will force us to
co-ordinate the efforts - Whatever the choice is, the power and the way of
getting it inside the detector volume is a major
problem which requires RD - Not mentioned was the complexity (and cost) of
the design tools - Requires training and practice
- The existing expertise among the different
institutes must be maintained
40Summary
- Front-end ASICs
- All are produced except
- Pixel FEI which is still in production (no
problems) - CMA from the RPC which has just been submitted
- SLB from TGC which needed a fix
- Front-end systems
- All in production
- Several systems have started the
production/procurement of power supplies - Solutions exist for the others (Pixels, TRT,
Muons) - Radiation hardness validation not to be forgotten
- Radiation hard electronics
- Be careful that nothing is forgotten during
installation - EMC policy (Grounding and shielding) to be applied