Air Cooling Status Report

Jehanne Simon-Gillo and Jan Boissevain

Los Alamos National Laboratory

February 1, 1996 Revised March 26, 1996

PHENIX-MVD-96-4

Introduction:

The PHENIX MVD subsystem has been investigating the possibility of using an air-based system to cool the front-end electronics. The power dissipation of the front-end electronics components has not yet been defined. Our initial beliefs were that the total power dissipation would be 5mW/chan. ORNL's best current estimate is approximately 13.5mW/chan. As these values have changed significantly, we have addressed below whether air cooling is still a viable option. As shown, air cooling does remain an option at 15mW/chan. Safety margins are discussed in the last section of this document. If the power dissipation significantly increases while the MCM size remains constant, then an alternative cooling system must seriously be considered. The following information must be known about the front-end electronics before a final decision can be made between air and other cooling systems: power dissipation, number of chips on MCM, size of chips on MCM, MCM technology choice. The latter three points define the MCM size.

Parameters:

Use cooling relationships developed by LANL Engineering Group, ESA-DE
From dewpoint experiments, assume minimum inlet air T of 10 C (5 C)
Assume size of MCM will increase: perform exercise for current size (3.5x4.8x0.11cm) and largest MCM size (5.0x4.8x0.11cm) that will fit in given physical dimensions.
As suggested by ESA, vary MCM spacing, massflow rate and inlet air temperature.
Current MCM spacing is 6mm- could increase to 8mm.
Keep electronics below 40 C
Assume power dissipation of 15mW/channel
- Specs originally 5mW/channel
- Current estimate 12.3mW/channel

Current Power dissipation estimates:

Preamp: 1.25mW/chan
Discriminator: 0.5mW/chan
AMU/ADC (2.3mW/chan) + correlator (1mW/chan): 3.3mW/chan
Heap Manager: 6.25mW/chan
OPAMP(converts current sum to voltage): 1mW/chan
LED driver(may be part of FPGA): 1mW/chan

Total: 13.3mW/chan

Results:

Example:

Three temperature gradients: T1 = heating of air through plenum, T2 = MCM-to-air thermal exchange, T3 = heat spreading on MCM

T1+T2+T3 < 30 C

From ESA: To cool 10mW/chan 3.5 x 4.8 x 0.11cm MCM below 40 C, 6mm spacing, 15m/s

T1 = 10 C
T2 = 9 C (Effected by power load, area)
T3 = 6 C (Effected by power load, area - assume fixed 6 C)

So, to cool an MCM size of 3.5 x 4.8 x 0.11 cm @ 15mW/chan, 6mm spacing:

Heat transfer coefficient = 80W/m^2 C
Heat load: 15mW/chanx256chan = 3.84W per MCM
Area to cool: 3.5x4.8cm = 16.8x10-4m^2
80W/m2oC x 16.8x10-4m^2 = 0.1344W/C
3.84W/MCM x 1/0.1344W C = 28.57 C / 2 =14.3 C = T2 (Divide by 2 as cooling both sides of MCM)
T3 = 6 C
14.3 + 6 = 20.3 so 9.7 C left for T1
Mass flow linearily related to T1
(New heat load/old heat load) x 15m/s = 3.84/2.56 x 15m/s = 22.5 m/s

Now change spacing to 8mm spacing:

Area increases by 8/6 = 1.33 This will affect T1
T1 = 10 C x 3.84/2.56 x 1/1.33 = 11.3 C goal is 9.7 C so need to increase velocity by 11.3/9.7=1.2
15m/s x 1.2 = 18m/s

For MCM size of 5 x 4.8cm

T2 = 10 C,

T3 = 6 C so 14 C left for

T1 (6mm spacing)

10 C x 3.84/2.56 x 1/1.43 = 10.5 C
10.5/14 x 15m/s = 11.3m/s
spacing increases to 8mm:
11.3m/s x 6/8 = 8.5 m/s

MCM Size	Spacing	Mass flow	Inlet air T (degrees C)
3.4 x 4.8 cm	6 mm	22.5 m/s	10 C
	8 mm	18 m/s	10 C
5.0 x 4.8 cm	6 mm	11.3 m/s	10 C
	8 mm	8.5 m/s	10 C

Implications and loose ends:

Air tubing must be insulated
Water lines to heat exchanger must be insulated
How do we cool the silicon
Test effect of really cold air on mechanical components
Define final power dissipation
MCM technology choice not yet defined
ESA states that cooling calculations often (but not always) 30-40% pessimistic - test via 6 layer cooling prototype
Error on dewpoint experimental hall

Safety margins:

The current safety margins are listed below. It is difficult to quantify some of these, especially the second one, "Commercial components can operate up to 70 C". If one does not include this particular safety margin, then it is estimated that the remainder constitute a 50% safety margin for the smaller size MCM and an 80% safety margin for the larger size MCM.

MVD goal for custom electronics is 40 C - Actual limit is 50 C.
Commercial components can operate up to 70 C (FPGA is hot spot - is commercial)
Assume 15mW/chan instead of 13.3mW/chan
High end estimates for LED driver
Assuming fully populated detector
Fix minimum inlet air T to 10 C (5 C)
Size fan unit for excess capacity