March 4 - 7, 2001Hilton Mesa Pavilion Hotel
Mesa, Arizona
Sponsored By The IEEE Computer SocietyTest Technology Technical Council
Burn-in & TestSocket Workshop
IEEE IEEE
COMPUTERSOCIETY
COPYRIGHT NOTICE• The papers in this publication comprise the proceedings of the 2001BiTS Workshop. They reflect the authors’ opinions and are reproduced aspresented , without change. Their inclusion in this publication does notconstitute an endorsement by the BiTS Workshop, the sponsors, or theInstitute of Electrical and Electronic Engineers, Inc.· There is NO copyright protection claimed by this publication. However,each presentation is the work of the authors and their respectivecompanies: as such, proper acknowledgement should be made to theappropriate source. Any questions regarding the use of any materialspresented should be directed to the author/s or their companies.
Session 6Tuesday 3/06/01 10:30AM
Thermal Management Approaches
�Dynamic Junction Temperature Control For Lidded C4 Packages�Joe Hovendon - Schlumberger
�Approaches To Thermal Management Of High Power Devices�Paul Nesrsta - Reliability Incorporated
�Dixie Chips - 'Too Hot To Handle'�Jim Ostendorf - Dynavision
Technical Program
Dynamic Junction TemperatureControl for Lidded C4 Packages
2001 Burn-in and Test Socket Workshop
Joe Hovendon Product Manager, Schlumberger
Dynamic Junction TemperatureControl for Lidded C4 Packages
2Schlumberger
Agenda
� Package Shift to C4
� MPU Power Roadmap
� Lidded C4 Package Detail
� C4 Package Thermal Circuits
� Lidded vs. Non-Lidded Tj error (passivecontrol)
� Active Conduction Thermal Control Solution
Dynamic Junction TemperatureControl for Lidded C4 Packages
3Schlumberger
Packaging Shift for MPUs
� MPUs are shifting to C4 packaging forimproved electrical and thermalperformance
Dynamic Junction TemperatureControl for Lidded C4 Packages
4Schlumberger
Power Density Roadmap
MPU Power & Power Density Trends
0
20
40
60
80
100
120
140
160
180
200
1999 2002 2005 2008 2011 2014
Year
Pow
er (W
)
0
10
20
30
40
50
60
70
80
90
100
Pow
er D
ensi
ty (W
/cm
2)
Px Pow er Dissipation (at Introduction)
Px Pow er Dissipation (at peak volume)
Px Pow er Density (Intro: 4 sq.cm. die)
Px Pow er Density (Peak: 1.25 sq.cm. die)
Dynamic Junction TemperatureControl for Lidded C4 Packages
5Schlumberger
Lidded C4 Package Details
Heat Exchanger
Package Lid
Interface Material
Die
Substrate
HeatExchanger &Lidded C4Package
Dynamic Junction TemperatureControl for Lidded C4 Packages
6Schlumberger
Lidded vs. Non-Lidded C4 PackageThermal Circuits
Tj(error)= (θθθθ(j-hs) + θθθθ(hs-hx) ) * Power
θ(j-hs)θθθθ(j-hs)
θθθθ(hs-hx)
Tj(error)= θθθθ(j-hx)* Power
θθθθ(j-hx)
Dynamic Junction TemperatureControl for Lidded C4 Packages
7Schlumberger
Lidded vs. Non-lidded Tj Error(Passive Conduction Thermal Control)
Passive Tj Error vs. MPU Power Forecast
0.00
20.00
40.00
60.00
80.00
100.00
120.00
140.00
160.00
180.00
200.00
1999 2002 2005 2008 2011 2014Year
Pow
er D
issi
patio
n
20
40
60
80
100
120
140
Pass
ive
Tj E
rror
Unlidded C4
Lidded C4 Lidded Tj(error)= (0.3 + 0.4 ) * Power
UnLidded Tj(error)= 0.3 * Power
Dynamic Junction TemperatureControl for Lidded C4 Packages
8Schlumberger
Active Conduction Thermal Control Heat Exchanger
Dynamic Junction TemperatureControl for Lidded C4 Packages
9Schlumberger
Active Conduction Thermal ControlSystem Architecture
Thermal ControlHardware &Software
HeaterPowerSupply
DUT
Temp.Exchange
Heat ExchangerDrive Signal
CoolantSystem
HeaterTemp.
Coolant
DUT PowerDUT Temp.
HeatExchanger
HeaterPower
Dynamic Junction TemperatureControl for Lidded C4 Packages
10Schlumberger
Active Conduction Thermal ControlData
Approaches to ThermalManagement of High Power
Devices
Paul NesrstaReliability Incorporated
Reliability Inc. 2
Scope of Presentation
� This presentation deals with methodsfor control dut junction temperatures ina parallel test or burn-in environment.
� What is the maximum dut power that canreasonably be accommodated usingcirculating air as the heat transfer medium?
Reliability Inc. 3
Presentation Road Map
� Process Cost reduction is the Goal� System Capacity vs Throughput� Dut Stress Uniformity vs throughput� System capacity determinants� Max dut power vs system capacity� Max reasonable dut power in air
Reliability Inc. 4
The Real Goal: Reducedprocessing cost for higher
power devices� Cost determinants;
� tooling cost� operating cost� System throughput� Yield
Reliability Inc. 5
Two keys toThroughput
� System Capacity� Shouldn�t be compromised if possible
� Stress Condition Uniformity
Reliability Inc. 6
Uniformity affectsThroughput
� All duts subjected to exactly identicaltest/stress conditions
� Why?� Closer tolerances allow higher stress
temps without compromising yield� Higher stress temps allow shorter BI
duration� Shorter BI duration allow higher throughput� Higher throughput = lower cost
Reliability Inc. 7
Uniformity determinants
� All test fixturing is not alike� Socketing variations� chamber variations
� Ambient Temperature� Air flow velocity
� All duts are not alike� Mechanical� Electrical performance
Reliability Inc. 8
To achieve uniform dietemps
� The thermal management systemmust compensate for variations in:� Dut to dut packaging� Socket to socket thermal impedance� Ambient temperature� Dut to dut power dissipation
Reliability Inc. 9
Dut to Ambient Interface
� Dut to Ambientinterface comprises 3elements:� Dut to heat sink
interface (Primary)� Heat sink to ambient
(Secondary)� Dut to socket heat
leakage (small)
Primary Interface
Secondary Interface
die
Socket
Reliability Inc. 10
Dut to Heat Sink interface
� Ideal interface would:� Provide low thermal impedance� Uniform from dut to dut� Provide even coverage across die
surface� Leave no residue� Robust� Low cost
Primary Interface
Reliability Inc. *Source: Thomas,Lindon,Heat Transfer, Prentice-Hall NJ 1992 11
Two choices for low primarythermal impedance
� Near perfect flatness and coplanarityor
� Conforming thermal interface
�Any air space between two heat conductingsurfaces greater than 50x10-6� adversely
affects heat conduction�
Reliability Inc. 12
Primary interfacecomparison
� Hard copper surface� θ j-hs: 3-5 ºC/W/cm2
� Variations due to lack ofconformability
� die coverage variations� Requires high contact
pressure� Die cracking� solder ball
� Conformableelastomer pad onCopper� θ j-hs: 1-2 ºC/W/cm2
� >15psi contact pressure� May leave residue� maintenance issue
Reliability Inc. 13
Primary interfacecomparison
� RI�s conform-ableinterface oncopper� θ j-hs:
~0.5ºC/W/cm2
+/- 5%~15 psi contactpressure
� Hard copper surface� θ j-hs: 3-5 ºC/W/cm2
� Variations due to lack ofconformability
� die coverage variations� Requires high contact
pressure� Die cracking� solder ball damage
Reliability Inc. 14
Yield vs Throughput
� Pressing hard enough on a hardinterface to insure even, consistentcontact may lead to mechanicaldamage.
� Not pressing hard may lead toinconsistent thermal management
� Allowing for inconsistent thermalmanagement compromises throughput
Reliability Inc. 15
Heat sink to Ambientinterface
(Secondary interface)� Ideal� low thermal impedance to ambient� uniform from dut to dut� cheap
� Our answer: Finned copper in highvelocity air
Secondary Interface
Reliability Inc. 16
Coolant uniformity
� Ambient temperature must be the samefor all duts
� Secondary thermal interface is afunction of heat sink to coolant contact
� HS to coolant contact is a function ofheat sink design & coolant velocity
Reliability Inc. Source: Christopher A. Soule PCIM Magizine August '97 pp104-111 17
Coolant velocity� Thermal boundary layer gets thinner as
velocity increases
Heat sink fin
Air FlowEffective boundary layer
Thermal boundary layer
Reliability Inc. Source: Christopher A. Soule PCIM Magizine August '97 pp104-111 18
Heatsink effectiveness vs airflow velocity
� Boundary layer changes less above 1000 lfm
Effective Boundary Thickness
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
100 300 500 700 900 1100 1300 1500 1700 1900 2100 2300 2500 2700
Velocity (LFM)
Inch
es
Laminar Flow Turbulent flow
Reliability Inc. 19
RI Criteria System Air flowCalculated:
Velocity Min/Max = 1800/2533 lfm
Measurement ResultsTop Backplane
Min = 2105 lfmMax = 3680 lfm
Bottom BackplaneMin = 1715 lfmMax = 2905 lfm
Reliability Inc. 20
Total ∆∆∆∆T Vs Air velocityDUT Power (Watts) 50
14.7 14.7Heatsink Area (Sq.In.) 20.0 0.52
Temperature Difference Between Die and Air in a C20 ChamberFor a 20 Sq. In. heatsink with a 2.2 Sq.Cm.Die Running at from 50 to 0 Watts
Laminer Flow Turbulent Flow
DUT to Heatsink InterfaceDie size (mm) X (mm)
Thermal Resistance in °C per Watt per Sq. CM
82.6
74.5
68.1
62.9
58.5
51.8
48.9
46.4
44.3
42.4
40.8
39.3
38.0
36.9
35.8
34.9
34.0
33.2
32.4
31.8
31.1
30.5
30.0
29.5
29.0
28.5
28.2
27.7
27.3
26.9
26.5
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
Velocity of Air in Feet Per Minute X 100
Del
ta T
°C (D
ie to
Air)
Reliability Inc. 21
Thermal impedance stack-up
� θ j-hs ~0.5 C/W/cm2
~2 cm2 = 0.25 C/W/cm2 +/-5%Calibrated per device
� θ hs-a = 0.42 C/W +/-15%� Total θ j-a = 0.67 C/W +/- 20%
Junction Heat Sink Ambient
θ j-hs θ hs-a
Reliability Inc. 22
Temperature uniformitycompensation
� Contributors:� Dut to Dut power variation� θ hs-a variation� θ j-hs variation
� Can compensate for dut power, θ j-hs & θ hs-a
Reliability Inc. 23
Per DUT Compensationmethods:
� Control thermal impedance path
� Add heat with fixed thermal impedancepath
Junction Heat Sink Ambient
θ j-hs θ hs-aAdjust
Junction Heat Sink Ambient
θ j-hs θ hs-a
Add heat
Reliability Inc. 24
Individually Controlledimpedance path per dut
� modulated coolant� complex electromechanical mechanism� thermally efficient� not robust� expensive
Modulate coolant
Reliability Inc. 25
Method 2� Thermo-electric cooler impedance
modulation� allows higher chamber ambient
temperature� robust� low efficiency� expensive
Impedance modulation
Reliability Inc. 26
Added heat compensation
� Resistive element heater� robust� cheap� not efficient
Add make-up heat
Junction Heat Sink Ambient
θ j-hs θ hs-a
Add heat
Net effect is reduction of ∆∆∆∆T j-hs
Reliability Inc. 27
Heater sizing� Assume 100% variation in dut power� θ j-a varies from 0.54C/W to .80C/W� Therefore 50W * .80C/W = 40C� Smallest θ hs-a is 0.36C/W� Therefore we must add ~112W to make
up 40C
Junction Heat Sink Ambient
θ j-hs θ hs-a
Add heat
Reliability Inc. 28
Dut TemperatureMeasurement
� To compensate for variations we must know:� Dut power� Ambient temp� Die Temp
� Plea to Device Designers:Please put accessible, accurate, tempmeasurement devices on the die!
Reliability Inc. 29
Measurement Method 1
� Die measurement outsidethe thermal path
� disturbs heat removaluniformity
� susceptible to local variationsin die temp
Measurement device
Reliability Inc. 30
Temperature MeasurementMethod 2
� Measure heat sink temp� maintains uniform die
coverage� provides die temp
averaging� allows compensation for
hs-a� must compensate for dut
to hs thermal interface
MeasurementDevice
Reliability Inc. 31
Dut tempprediction/compensation
� If we know:� dut power� ambient temp� thermal impedance from dut to
ambient� Then we can calculate dut temp
and correct
DUTSet Point
Cntrl
Reliability Inc. 32
Plot of DUT Temperature/Power versus Time
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
0 50 100 150 200 250 300 350 400 450 500
Time [s]
Tem
pera
ture
[C]/P
ower
[W]
DUT Temp. (calc.) DUT Temp. (meas.)T/S Temp. DUT Power
ParametersDUT Temp. sp = 85 CAirflow = 1700 (+/- 30) lfmAir Temperature = 22 CDUT Pwr. = 16-50-0 W Trans
16 W Steady StateMax. Temp. = 85.6 CMin. Temp. = 84.8 CAverage = 85.23 Cmeas/calc error = 0.11 C
50 W Steady StateMax. Temp. = 86.2 CMin. Temp. = 84.9 CAverage = 85.39 Cmeas/calc error = 0.06 C
0 W Steady StateMax. Temp. = 85.9 CMin. Temp. = 85.0 CAverage = 85.64 Cmeas/calc error = 0.5 C
50-0W TransitionUndershoot = 75 CDuration = 67s
16-50W TransitionOvershoot = 90.35 CDuration = 42s
Reliability Inc. 33
So far we have provided:
� a consistent low thermal impedancefrom the die to ambient
� Worst cast = 0.8 º C/W� 0.80 º C/W * 50W = 40 ºC rise from dut
to ambient� IF target Tj is 125 ºC then Ambient must
be 125 - 40 = 85 ºC
Reliability Inc. 34
System Power Sizing
� Assume normal distribution of 0 to 50Wduts
� Therefore avg Power = 25W� Therefore must make up 25W/dut� If 115W make up for 0W then avg
make-up heat = (25/50)*112W =56W� 56W + 25W = 81W/dut
Reliability Inc. 35
System Capacity
� Standard C18 dissipates 5500W @ 85ºC
� 5500W / 81W/dut = 67 duts/system
� C20 dissipates 24,000W @ 50 ºC� 24000 / 81 => 296 duts/system
or >12 duts/burn-in board
Reliability Inc. 36
The total system must:
� Combine:� Low thermal impedance
path to coolant� dut power variation
compensation� High power handling at
low ambient set points
Reliability Inc. 37
Conclusions
� �Conventional� circulating air Test &BI isviable for dut power dissipation <50W
� This is desirable because it:� High throughput� Robust� Maintains current process paradigm� Saves money!
Dixie ChipsDixie Chips
�Too Hot to Handle�
byJim Ostendorf
DYNAVISION
2
BibsBibs
�Electrical
�Mechanical
�Thermal
�Cost
3
ElectricalElectrical
�Noise; Cross Talk�Rise/Fall Times�Functional Test�Bist
4
MechanicalMechanical
�Loader/Unloader�Rigidity�Bowing�Accuracy�Plating
5
ThermalThermal
�Objectives:- Temperature- Voltage- Fail Safe Protection- Cost
6
ThermalThermal
�Design Goal:- Tight Die Temperature Control forDiffering DUT Power Dissipation
7
ThermalThermal
�Smart Heat Sink�Thermocouple�Heat Pump�Control Module
8
ThermalThermal
� Voltage:- Low Voltage 1.8 V
� Current:- High Current Up to 30.0 A
� Power Characteristics of Target DUTS:- Wide Range of Power Dissipation
9
ThermalThermal
� System Description:- An intelligent heat sink is assigned to each DUTallowing bi-directional transfer of heat betweenDUT and heat sink surfaces.- The intelligent heat sink pulls heat out of hotdevices and pushes heat into cold devicesmaintaining the desired uniform temperature oneach DUT.- The intelligent heat sink consists of a bi-directional heat pump sandwiched between twoparts of a passive metal heat sink.
10
ThermalThermal
� Temperature Control:- A temperature sensor and one side of themodified heat sink get into direct contact with thesurface of the DUT.- The other portion of the modified heat sink hasthe fins that interface with the oven air flow.- The temperature of the oven is set at anappropriate level that would allow the thermalcontrol system to work.
11
ThermalThermal
� Temperature Control cont.:- Imbedded between the two portions of the metalheat sink is the heat pump. the heat pump iselectric current controlled.- The direction of the heat flow is determined bythe direction of the current through the heat pump.- The direction of the current depends on whetherthe DUT temperature that is sensed is higher orlower than the target temperature.
12
ThermalThermal(Voltage)(Voltage)
� Voltage:� DC to DC Converter� Sense at DUT
13
ThermalThermal(Voltage)(Voltage)
� Design Goal:- Tight Voltage Control Throughout the FullTarget Current Range
14
ThermalThermal(Voltage)(Voltage)
� System Description:- One dc to dc converter is dedicated toeach DUT cell, allowing very tight voltageregulation and control over each cell.- Voltage sensing for regulation is pickedup right at the DUT load thereby reducingdrastically the voltage variations caused bythe large current draws.
15
ThermalThermal(Protection)(Protection)
� Design Goal:- Provide Fail Safe Control to ProtectDevices Under Test
� Shut Down Mode
16
ThermalThermal(Protection)(Protection)
� Fail Safe Features:- Power is turned off to a DUT on thermalrunaway, over voltage and over current conditions.- No power is applied to a cell on �DUT absent"condition.- No power is delivered to the bib while the theheat sink and power control assembly is "intransit".- Burn-in board can not be withdrawn unless thecontrol assembly is in the retracted position.
17
ThermalThermal(Protection)(Protection)
� System Description:- Some standard, widely-used burn-in systemshave user-designed back planes and driver boardsfor exercising and stressing microprocessor andasic products.- New burn-in board design plugs into theseexisting systems but with provisions forindividually controlling the power to each DUTcell and individually controlling the temperatureof the device in each cell.
18
ThermalThermal(Protection)(Protection)
� System Description cont.:- These power and thermal control functions maybe housed in a separate assembly plugged into theslot next to the burn-in board it is intended tocontrol.- This second approach allows for use of thesecontrol overhead over several application devicesprovided that a standard density and geometry ofthe burn-in board is implemented. choice of socketis very important in order to allow efficiency inthermal interface and transfer.
19
ThermalThermal(Cost)(Cost)
� Design Goal:- Design into a wide-based, standard, andfamiliar burn-in system
� Air Heat Exchange� Existing Hardware� Total Solution
20
ThermalThermal(Mechanical)(Mechanical)
� System Description:- The mechanical design of the voltage andcontrol assembly and the kinematics ofmechanical motion required to move theheat sink and power tabs against therequired surface contact pressure betweenthe DUT surface and the heat sink is themost challenging part of this project.
21
Front Front
22
Left Left
23
Right Right