FP420 – CERN 17-18 February 2008
FP420 Detector Cooling
Thermal Considerations
Berend Winter
FP420 Workshop CERN 17-18 Feb 2008
The Detector Array
Flexible Link (electronics signals)
Prox. Electronics
Tracker Cell
Detector Plane (CE7)
Beam Line
Thermal Interface to Cold Sink
•
•
The detector array is moved together with the beam line tube (not shown here) which has flexible
links (preserving primary vaccum) with the main beam line. This means that the interface between the
detector array cold sink and the vacuum enclosure (secondary vacuum) moves together (no relative
displacement).
Thermal Design Goal: tracker cell operating at -20°C
FP420 Workshop CERN 17-18 Feb 2008
Dissipated Heat
• The detector consists of 5 super modules
– Each super module consists of 2 detector planes (Al/Si) with each
holding 2 detectors on each side (4 in total per super module)
– Each super module has an ASIC underneath each detector and the
4 silicon detectors within each super module share one proximity
electronics module
– Each detector/ASIC combination dissipates 0.5 W or less
– Each proximity electronics module dissipates 1.0 W
– This makes for a total of 3 W per super module and 15 W in total
for each detector array head (5 super modules)
– As we will see later on 5 W needs to be added due to parasitic
heatloads
FP420 Workshop CERN 17-18 Feb 2008
Heat Paths
• Each super module has two thermal interfaces to a cold
sink (5 on each side)
• Each detector (10 in total for a detector array) is connected
to (relatively) warm front end electronics via a flexible link
• The detector array is mounted on an interface (vacuum
enclosure) at room temperature (20 to 30 deg-C)
• The detector array is radiation coupled with the vacuum
enclosure
• The thermal hardware (straps) are radiation coupled with
the vacuum enclosure
• The cold sink has to interface through the vacuum
enclosure and is somehow conductively coupled with it
FP420 Workshop CERN 17-18 Feb 2008
Thermal Network Around the Detectors
(one detector plane)
Conducted Heat
Radiated Heat
Cold Sink Interface
Si tracker cell 0.5 W
Flex link
Al/Si detector plane (support structure)
Prox. Electronics (0.5 W)
Si tracker cell 0.5 W
cables
Vacuum Enclosure
support
Cold Sink Interface
Flex link
FP420 Workshop CERN 17-18 Feb 2008
Parasitic Heat Loads
• The detector array as a whole is supported by the vacuum
enclosure, heat will flow from the warm vacuum enclosure
to the cold detector array.
– Using Torlon support rods the parasitic heat load can be limited to
less than 1.5 mW/Celcius (6 rods, 5x5x50 mm) with the detector
array running at -30°C and the vacuum vessel at +30°C this gives
less than 0.1 W parasitic heat load for the detector array.
• The proximity electronics are connected through the
vacuum vessel wall, 30 conductors (28 AWG)
– 30 wires, 28 gauge with a minimum length of 100 mm and a
thermal jump of -30°C to +30°C gives 0.5 W per proximity
electronics PCB (one for each super module). This is 2.5 W for the
whole detector array.
FP420 Workshop CERN 17-18 Feb 2008
Parasitic Heat Loads (Continued)
• Radiated Heat
– Detector Array absorbs 2.2 W total (worst case)
– Thermal straps absorb 0.05 W total (worst case)
ASIC/Detector
Proximity Electronics
Detector Support
Harness
Radiated Heat
Total
Unit [W] Number # Total [W] Type
0.50
20
10.00 Dissipated
1.00
5
5.00 Dissipated
0.02
6
0.10 Parasitic
0.50
5
2.50 Parasitic
2.25
1
2.25 Parasitic
19.85 W
FP420 Workshop CERN 17-18 Feb 2008
Tracker Cell Model
Tracker layer 200um
Bumps 50um
ASIC 200um
Glue 100um
FP420 tracker model
CE7 500um
30
20
Tr middle
10
TR edge
ASIC edge
ASIC middle
CE7 middle
°C
CE7 Cedge
0
-10
0
2
4
6
8
-20
-30
•
Time [sec]
In 2006 an ESATAN analysis was performed to assess the thermal gradient over the
tracker cell. With updated materials and dimensions the results are still the same.
Gradient between tracker cell and the CE7 carrier plate is 2.5°C (worst case)
FP420 Workshop CERN 17-18 Feb 2008
Thermal analysis of Al/Si CE7 structure
-21.3°C
• A thermal analysis was
performed on the CE7 support
structure.
• Applied heatloads:
– 1 W for two tracker cells
– 0.5 W for prox. elec.
– 0.5 W for parasitics
• Applying a boundary condition
of -29.3°C the interface with the
tracker cell is -22.5°C
– As a result the tracker cell itself
is sitting at -20°C.
-29.3°C
-22.5°C
FP420 Workshop CERN 17-18 Feb 2008
Thermal Gradients
•
•
Since we know the heat loads we can calculate the thermal gradient at each
stage of the design.
– Analysis has shown a gradient of 2.5°C from the active part of the tracker
cell to the support structure (CE7 plane) with a heatload of 0.5 being
dissipated in the tracker cell. This is worst case (but not extreme) since
most of the heat will be dissipated in the ASIC underneath.
– The heat dissipated in the proximity electronics and the parasitics
(radiation and conduction) were analysed and resulted in a gradient of
6.8°C between the CE7 area where the tracker cell is mounted and the
cold interface. The total heat flow per CE7 plane is 2 W (dissipated and
parasitic)
– The gradient between the tracker cell and the cold interface is 9.3°C in
total.
For the rest of the design we need a cold interface to the CE7 planes at -30°C
and we need to conduct 2 W per interface. (Two sinks on each side of a super
plane makes ten in total, conducting 20 W in total)
FP420 Workshop CERN 17-18 Feb 2008
Peltier Cooling
• Peltier Cooling forces a thermal gradient over a series of solid state
elements capable of pumping heat
• Not a very efficient way of cooling. About 2% efficient for the first stage
of detector cooling in our case (with updated heat loads)
• Maximum gradients in order of 70°C (without pumping capacity)
FP420 Workshop CERN 17-18 Feb 2008
Peltier Cooling no longer Viable
• Last year the assumptions on the total amount of heat dissipated and
the total amount of parasitics were seriously underestimated (factor of
3). Simple hand calculations now show that the Peltier cooling option
is no longer viable. The efficiency has dropped by a factor of ~4. So
we need to dump significantly more heat with less efficient Peltier
elements (around 2% efficiency).
• We need to dump an awful lot of heat to the outside of the vacuum
vessel and use fins with forced convection as a heat sink to go all
electric (peltier and cooling fans) on the cooling. For example
computer CPU fans only get rid of a Watt or two in optimal conditions.
That is by using a significant thermal gradient (about 70°C)
• Using Peltier devices we estimate the total heat we need to dump
(somehow) into the tunnel to be significant more than 1 kW…
FP420 Workshop CERN 17-18 Feb 2008
Fluid Cooling
• Using a simple fluid cooling loop with a cold plate
on the inside of the vacuum vessel we should be
able to dump the heat efficiently. This is by means
of a copper tube soldered to the back of a copper
plate. The copper plate (gold plated) has holes
where thermal straps from the detector can be
bolted on firmly. This is a technique routinely used
in cryogenics/vacuum chambers. The interface
through the wall is via a standard (off the shelf)
thin wall stainless tube feed through.
FP420 Workshop CERN 17-18 Feb 2008
Heat Transfer
• The heat flows per CE7 plane into a copper shim
and then into a copper block where the heat of
two CE7 planes combines before it flows into the
cold plate.
-29.3°C CE7
1W
Cu block
2W
-29.3°C CE7
1W
Fluid Cooled
Copper Plate
FP420 Workshop CERN 17-18 Feb 2008
Heat Transfer Continued
Conductive Path
Pressure contact between CE7 and copper shim
Bulk conductance copper shim (0.25 x 15 x 30 mm³)
Pressure contact between copper shim and copper block
Bulk conductance copper block
Pressure contact between block and busbar (cooling plate)
Bulk conductance of busbar (cooling plate)
Total Jump
Heat [W]
W/k
1
1
1
2
2
2
0.25
0.17
0.25
10
0.25
0.55
delta-°C
4.0
5.9
4.0
0.2
8.0
3.6
25.7
• The thermal jump between the CE7 planes and the cooling plate is
25.7°C making the total jump between tracker cell and cooling plate
35°C.
• Conclusion: The cooling plate has to be at least -55°C and be able to
pump 10 W (one cooling plate on each side of the detector arrays)
FP420 Workshop CERN 17-18 Feb 2008
Cooling Thermostats
• Cooling thermostats come
in wide range.
• Lauda is a prime supplier
of cooling thermostats
capable of cooling down to
-90°C, most likely not
radiation hard
• Can CERN provide for a
cooling fluid line?