Title: Integrated Stave Electrical/Mechanics/Cooling Update
1Integrated StaveElectrical/Mechanics/Cooling
Update
2Prototypes and Designs
60 cm, 9 cm strip, 6 segments/side
Stave-06
1 meter, 3 cm strip, 30 segments/side 192 Watts
(ABCD chip), 2.4 Xo support structure
6 x 3 cm, 6 chips wide
Stave-07
Build and test
Study
1 meter, 2.5 cm strip, 40 segments/side 200-250
Watts (_at_0.25 W/chip) 1.9 2.2 Xo support
10 x 10 cm, 10 chips wide
Stave-08
3Introduction
- With regard to the Single Sided approach, the
arguments about simplicity, material, and cost,
have been presented already. - This effort is tightly coupled to the alternative
powering efforts, in particular serial, but we
would hope to include aspects of DC-DC as well,
in the future. - We are concerned that in a large, for example,
serial system, grounding, shielding, and
modularity issues may be critical - Our goal, therefore, is to build and test a
realistic scale system early-on, and include
enough options and flexibility to be useful. - With regards to the above, and also to assembly
and production issues, we want to confront as
much of the full problem as possible, now.
4Aspects
- Status overview
- Components
- Fixtures
- Electrical testing
- Additional critical tests
- Alternatives
5Issues From Valencia
- Progress on testing and fabrication of Stave-07
- Irradiation of hybrids glued directly on silicon
surface - Thermal performance of bridged hybrid
- Planning for Spring 2008 module review
6Status Overview
- Central goal is assembly and test of Stave-2007.
Confront assembly, test, and measurement issues
relevant to future - Have built 30 hybrids , and operated 5 in a
serial chain with good performance - Have built and are studying 4 modules with hybrid
on the silicon and 1 reference module with
hybrid off the silicon - Effort on gluing and alignment procedures
implicit in module building - Bus cable has been fabricated
- Extra clock lines have been added to allow
options for clock distribution included 1 clock
for 30 module or 1 clock for 10 modules - All components required to build Stave-2007 are
in-hand. - Bridged hybrid has been further simulated and
clarified - Irradiation plans underway
7Components
- 3 cm p-in-n sensors based upon ATLAS-98
- Fanouts from SCT
- 6 ABCD chip, serial powered ceramic hybrid
- Approx 30 built and tested with good yield
- Stave mechanical core
- Ready and waiting
- Assembly fixtures
- In use
- Interface pc boards
- Bus cable fabrication complete, delivered
- DAQ system (NI-PXI card LV software)
- Power supplies
86 ABCD BeO Ceramic Hybrid
- 30 fabricated and tested
- Yield is (surprisingly) good!
- Represents a density maxima
- Includes HV-GND options within serial scheme
- Analog performance is right-on-target
- Ceramic flatness is engineered by printing but
not perfect.
Analog current
Ground layer
Analog power
Power layer
Digital power
LVDS section
Serial power section
9Hybrid HV-GND options
AC gnd
AG-MOD
HV Gnd
HV in
AV-MOD
10Fixtures for Assembly AND Test
11Module Assembly and Test Fixture
Evolve a single fixture for assembly, bonding,
inspection, and test.
12Bus Cable Signal Layout
Clock Command lines
Data Readout 1/hybrid
Port Card
HV distribution
Serial current return
Serial current link
13Bus Cable Shielding
Al foil, 50 um thick, can be grounded to each
hybrid
14DAQ
15HV Supply for 30 step serial system
- GENH 150-5
- 0-150 V
- 0-5 A
- Constant current or constant voltage mode
- 30 step ABCD system will require 120 V and 0.75
A. - Supply is in-hand, preparing to test using hybrid
test board daisy chain and then bus cable
16Electrical testing
- Hybrid performance
- verified
- Serial powering with increasing drops
- 5 OK, new supply allows full 30 drop test
- Data transmission in a multi-drop system
- 5 test board system OK, now confront full bus
cable, Santa Cruz - Module performance
- In progress, see slides
- Grounding and shielding on a stave
- upcoming
- Effect of glue
- Radiation effects
17Module Performance
- Reference module (hybrid off silicon) shows
excellent performance, low leakage, low noise,
and correct gain - Assembly of 4 hybrid-on modules has been a
learning curve and not without incident - One module shows good leakage and analog
performance while the others have larger
currents. - Adopting additional safeguards and procedures in
order to control performance. - Note for Stave-06 multiple good glued-on
modules were built and characterized
ref
Glued on
Input noise _at_ 2fc
18continued
Glued on
Ref
19Additional Critical Tests
- Glue studies
- Once we get the module assembly process
under-control, plan a systematic glue testing
program - Temperature cycling
- Load with thermal filler (BN)
-
- Needs to be repeated on n-in-p sensors as well.
- Irradiations
- See slides
- Bridged hybrid
- See slides
20Irradiation plans
- Irradiations are in planning phase both at BNL
and LBNL - BNL is looking at BNL, Boston, and Los Alamos
sites - LBNL would use on-site 55 MeV protons
- Hope to have first runs in the next few months
- A key set of questions here are what would
constitute a meaningful measurement? - Conditions
- Particle type
- Specifics of detector design
- One point of view holds this to be purely a
surface issue which could be addressed with a
gamma sourceopinions? - Ultimate skepticism?
- Note CDF and D0 have run with glued hybrids for
gt5 years
21Alternatives
- The bridged hybrid has always been the ATLAS
preference - FEA indicates reasonable performance for this
alternative - However the ultimate material reduction would
come from reducing the hybrid substrate even more - A kapton flex or other thin film hybrid with
essentially no substrate, glued directly on the
silicon, would minimize material - We are not wed to the ceramic technology it was
convenient and low risk for us. - We would be happy to see others pick this
challenge up and move it forward! - Since the 6 chip serial hybrid is known to work
electrically, we are importing the layout into a
fine-pitch printed board design which could be
adopted for flex or other etched approaches. We
can make this available to the community.
22FEA Models of Bridged Hybrid - I
- Since the Valencia meeting we have completed FEA
analysis of basic thermal performance with
bridged hybrid - Multiple models to achieve reliability and
understanding - ¼ simple model, similar to ANSYS model by others,
air treated as solid under bridge - ½ model with air box to allow for 3D air effects
(not just under bridge), no air flow
Air box for 1/2
¼ model
½ model
Air gap
23FEA Models of Bridged Hybrid -II
- Multi-hybrid model, with air flow included(so far
0.01 m/sec) - For simplicity, all studies so far done with tube
wall temperature fixed at -28C, 0.25 W/IC and no
detector heating. - Goal is to compare first with glued hybrid under
same conditions - More details are here and here
Multi-hybrid model
24Short Summary
- Good agreement among models
- Modest effect of gas flow
- Reduced K means lower than nominal K in
structure. No optimization of structure yet - For comparison, nominal design (hybrids glued on
silicon) for comparable assumptions yields max
sensor temperature of about -22.5C
Model IC Peak Temp(C) Bridge Gradient(C) Sensor T Max(C)
¼ model -5.8 6.6 -20
½ model -6.3 6.8 -20.2
With air flow -15C gas -5.9 7.6 -18.7
¼ model Reduced K 3.4 5 -16.7
25Conclusions
- Focused attempt to address the plans and issues
discussed in Valencia - All components for Stave-2007 are in-hand
- Approach is friendly to alternatives bridges
etc - We welcome the participation, input, or
suggestions of the community - Look forward to preparing for the June review