Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
Roger Stout, PE Research Scientist Corporate R&D: Technology Development
RPS • 2010 April
<roger.stout@onsemi.com>
Corporate Research & Development Packaging Technology
Course outline
• Part I: Characterization (75 minutes)
– Why Everything you Thought You Knew (about device thermal characteristics) is Wrong – Characterization Techniques – Miscellaneous Measurement Techniques
• Part II: Linear Superposition (120 minutes)
– – – – – Basic Theory The Reciprocity Theorem A Detailed Example and its Implications Building a System Model Time Varying Heat Sources
• Part III: Thermal Runaway (45 minutes)
– Theory – Datasheet example – HighTemperature Reverse Bias “qual” example
1 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Part I Characterization
2
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Can this device handle 2 W?
3
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Why is ON’s SOT23 thermal number so much worse than the other guy’s?
• ON
– – – – – – SOT23 package 60x60 die Solder D/A Copper leadframe Minpad board Still air
• some other guy
– – – – – – SOT23 package 20x20 die Epoxy D/A Alloy 42 leadframe 1” x 2oz spreader Big fan
4
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Fundamental ideas
• Heat flows from higher temperature to lower temperature • The bigger the temperature difference, the more heat that flows • Three modes of heat transfer – Conduction (solids, fluids with no motion) – Convection (fluids in motion) – Radiation (it just happens)
5
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermalelectrical analogy
temperature power Δtemp/power <=> voltage <=> current <=> resistance
energy/degree
<=> capacitance
6
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
“Junction” temperature?
Historically, for discrete devices, the “junction” was literally the essential “pn” junction of the device. This is still true for basic rectifiers, bipolar transistors, and many other devices. More generally, however, by “junction” these days we mean the hottest place in the device, which will be somewhere on the silicon (2nd Law of Thermodynamics).
This gets to be somewhat tricky to identify as we move to complex devices where different parts of the silicon do different jobs at different times.
7 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
What’s wrong with thetaJA?
2
JA
TJ Ta Pd
JA
TJ
Jtab
Ttab Pd
TJ
8
Pd
Ta
TJ
Corporate R&D : Packaging Technology
Jtab
Pd
Ttab
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
ThetaJA vs. copper area
9
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Typical thermal test board types
Minpad board
1inchpad board
Minimum metal area to attach device (plus traces to get signals and power in and out)
Device at center of 1”x1” metal area (typically 1oz Cu); divided into sections based on lead count
10
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Theta ( ) vs.. Psi ( )
• JEDEC <http://www.jedec.org/> terminology – Z JX , R JA older terms ref JESD233, 234 – JA ref JESD 51, 511 – JMA ref JESD 516 – JT, TA ref JESD 512 – JB, BA ref JESD 516, 518 – R JB ref JESD 518 – Great overview, all terms: JESD 5112
11
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
“Theta” (Greek letter
We know actual heat flowing along path of interest
Tx
xy
Ty
qpath
Ty
Tx
true “thermal resistance”
12 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
“Psi” (Greek letter
We don’t know actual heat flowing along path of interest
Tx
xy
Ty
qtotal
??
Tx
… all we know is total heat input
13 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Ty
a reference number only
Corporate R&D : Packaging Technology
Facts and fallacies
• Basic idea: – “thermal resistance” is an intrinsic property of a package • Flaws in idea: – there is no isothermal “surface”, so you can’t define a “case” temperature
• Plastic body (especially) has big gradients
– different leads are at different temperatures – multiple, parallel thermal paths out of package
14
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
An example of a device with two different “Max Power” ratings
• Suppose a datasheet says: – Tjmax = 150°C • But it also says: – JA = 100°C/W – JL = 25°C/W – Pd = 1.25W (Tamb=25°C) – Pd = 3.0W (TL=75°C)
25 100 *1.25 25 125 150
75 25 * 3 75 75 150
Where’s the “inconsistency”?
15 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Where’s the inconsistency?
TJ =150°C
25°C/W ( JL) 100°C/W ( JA)
What’s TL? Not 75°C !!
(try about 119°C)
TA =25°C
…¾ of the way from ambient to Tj
16
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Back in the good old days ...
metal can fair approximation of “isothermal” surface axial leaded device only two leads, heat path fairly well defined
17
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Which lead? Where on case?
18
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
“Archetypal” package
10%
convection case
wire/clip
silicon
die attach
flag/leadframe
10%
20%
60%
circuit board
Corporate R&D : Packaging Technology
19
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Then we change things …
add an external heatsink …
optional heatsink
flip the die over …
optional heatsink
40%
60%
mold compound/ case wire/clip silicon
die attach
optional “case”
die attach pads/ balls optional underfill
silicon
flag/leadframe
20%
40%
application board
20%
application board
20%
20
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
A bare “flip chip”
10%
silicon pads/ balls underfill application board
90%
21
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Same ref, different values
J tab
1.2 C/W
J tab
Pd Tc
50 W 25 C
0.8 C/W
Pd Tc
1.5W 25 C
1 GPM of H 2 O
still air
22
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Even when it’s constant, it’s not!
Tj
R1 (path down to board) constant at 20
package environment
R3 (path through case top) constant at 80
TL
TC
R4 (case to air path resistance) constant at 500, or 20x R2
1000 900 800 700 600 500 400 300 200 100 0 1
thetaJA  var brd only thetaJA  var airflow
JA
TJ Tamb Q total
1 1 R1 R 2
board
1 R3 R 4
1000
R2 (board resistance) vary from 1 to 1000
10 rstnc. [C/W] 100
thetaJA
Tamb
25 psiJL  var brd only psiJL  var airflow 60 50 40 15 20
psiJT
psiJC  var brd only psiJC  var airflow
psiJL
JL
T J TL Q total
R1 10 R1 R 2 1 5 R3 R 4
0 1 10
board rstnc. [C/W]
30 20 10 0 100 1000 1
JT
TJ TC Q total
10
R3 R3 R 4 1 board R1 R 2 rstnc. [C/W] 100
1000
23
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Tj
JA
Ta
24
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Fallacies recap:
• “Package resistance” isn’t fixed: – multiple heat paths exiting package – boundary conditions dictate heat flow
• • • • • Heat sinks Neighboring devices/power dissipation Single vs.. doublesided boards Local convection vs.. boardedge cooling Multiple layers/power/ground planes
• Therefore, different application environments will see different “package resistance”
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
25
Corporate R&D : Packaging Technology
Characterization Techniques
Typical TSP behavior
calibrate forward voltage at controlled, small (say 1mA) sense current
Characterization Techniques
125°C sense current
T
Vf
25°C
0.5 V
Vf
0.7 V
26
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
How to measure Tj
true const. current supply
(1 mA typical)
approximate const. current supply
10K DUT
OR
10.7V DUT If V f0.7V, then I1mA
27
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
How to heat
sample current is off while heating current on sample current is always on
10K heating power supply
10K heating power supply
10.7V
DUT
10.7V
OR
DUT
28
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
The importance of 4wire measurements
+(1.00 V) Power Output = 1 W supply (0 V) 0.18 V 0.64 W 0.70 W 0.05 V
29 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
1A
0.82 V 0.90 W 0.85 V
0.15 V
0.95 V
Corporate R&D : Packaging Technology
Which raises an interesting question:
+(1.00 V) Power Output = 3 W supply (0 V) 0.45 V 1.3 W 0.02 V 0.55 V
3A
1.3 W 0.3 W 0.98 V
Is this a fair characterization of a lowRdson device?
30 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Bipolar transistor
• TSP is Vce at designated “constant” current • Heating is through Vce • Choose a base current that permits adequate heating
bias resistor TSP supply
10K
switch
TSP=Vce
bias supply heating supply
31
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Schottky diode
• TSP is forward voltage at “low” current • Voltages are typically very small (especially as temperature goes up) • Highly nonlinear, though maybe better as TSP current increases; because voltage is low, higher TSP current may be acceptable • Heating current will be large
32
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
MOSFET / TMOS
• Typically, use reverse bias “back body diode” for both TSP and for heating • May need to tie gate to source (or drain) for reliable TSP characteristic
TSP supply
10K
switch
TSP=Vsd
heating supply +
33
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
MOSFET / TMOS method 2
• If you have fast switches and stable supplies • Forward bias everything and use two different gate voltages
TSP supply
10K
close switch to heat
close switch to heat
close switch to measure
+ Vgate for heating 
+ Vgate for measure 
TSP=Vds
+ heating supply 
34
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
RF MOS
• They exist to amplify high frequencies (i.e. noise)! • Feedback resistors may keep them in DC
TSP supply
10K
close switch to heat
TSP supply + + heating supply 
close switch to heat
close switch to measure
+ Vgate for heating 
TSP = body diode
35
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
IGBT
• Drainsource channel used for both TSP and heating • Find a gate voltage which “turns on” the drainsource channel enough for heating purposes • Use same gate voltage, but typically low TSP current for temperature measurement
TSP supply
10K
switch
TSP=Vds
gate voltage heating supply
36
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thyristor
• Anodetocathode voltage path TSP supply used both for TSP and for heating 10K • typical TSP current probably lower switch than “holding” current, so gate must be turned on for TSP anode gate readings; try tying it to the anode (even so, we used 20mA to test SCR2146) TSP=Vac • Hopefully, with anode tied to gate, cathode enough power can be dissipated to heat device without exceeding gate voltage limit
37 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
heating supply
Logic and analog
• Find any TSP you can – ESD diodes on inputs or outputs – Body diodes somewhere • Heat wherever you can – High voltage limits on Vcc, Vee, whatever – Body diodes or output drivers – Live loads on outputs
• (be very careful how you measure power!)
38
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Heating curve method vs.. cooling curve method
39
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Quick review: Basic Tj measurement
first we heat then we measure
10K heating power supply
10K heating power supply
10.7V DUT
10.7V DUT
40
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Question
• What happens when you switch from “heat” to “measure”?
Answer: stuff changes
• More specifically, the junction starts to cool down
41
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Basic “heating curve” transient method
Vf
calibrate forward voltage @ 1mA sense current convert cooling volts to temperature
.5V
measurements
voltage current
1 ma poweroff cooling poweroff cooling poweroff cooling poweroff cooling
steady state reached
highcurrent heating
highcurrent heating
highcurrent heating
highcurrent heating
T
25°C
Temperature
125°C
measured temperatures Time
Vf
.7V
42
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Heating curve method #2
voltage
1 ma
highcurrent heating
highcurrent heating
highcurrent heating
highcurrent heating
Temperature
poweroff cooling
poweroff cooling
poweroff cooling
poweroff cooling
Time
measured temperatures
43 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Time
current
measurements
voltage
current
Basic “cooling curve” transient method
Vf
calibrate forward voltage @ 1mA sense current Temperature
.5V
1 ma
highcurrent heating
poweroff cooling convert cooling volts to temperature
125°C
T
25°C
steady state reached
Temperature
Vf
.7V
Time (from start of cooling)
Time transient cooling period (data taken)
heating period
subtract cooling curve from peak temperature to obtain “heating” curve equivalent
44
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Whoa! … that last step there …
• Heating vs.. cooling – Physics is symmetric, as long as the material and system properties are independent of temperature
45
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Heating vs.. cooling symmetry
Start of constant power input (“step heating”)
Start of (constant) power off
junction flag
lead
(all the same curves, flipped vertically)
back of board edge of board
46
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
• For a theoretically valid cooling curve, you must begin at true thermal equilibrium (not uniform temperature, but steady state) • So whatever your JA, max power is limited to:
power
47 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
T j max
Tambient
JA
Corporate R&D : Packaging Technology
By the way … Steadystate vs.. transient ?
• Since you must have the device at steady state in order to make a full transient coolingcurve measurement, steadystate JA is a freebie. (given that you account for the slight cooling which took place before your first good measurement occurred)
48
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Effect of power on heating curve
24x steadystate power 6x steadystate power 3x steadystate power 2x steadystate power
Tjmax
junction temperature
steadystate max power
< steadystate max power
Tamb
49 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
time
Corporate R&D : Packaging Technology
Some initial uncertainty
a few initial points may be uncertain
highcurrent heating Temperature steady state reached
but once we’re past the “uncertain” range, all the rest of the points are “good”
poweroff cooling
transient cooling period (data taken)
heating period Time
50
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Heating vs.. cooling tradeoffs
HEATING starting temperature ambient limited by tester closer to ambient all points similar error
Corporate R&D : Packaging Technology
COOLING ? limited to steadystate closer to Tjmax error limited to first few points
heating power
temperature of fastest data error control
51 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Still air vs.. moving air
• Varying the air speed is mainly varying the heat loss from the test board surface area, not from the package itself • You just keep remeasuring your board’s characteristics
52
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
100
total system thermal resistance
90
80
70 60 50 40 30 20
little package
package resistance
thetaJA [C/W]
mediumsized package
board resistance
10 0 0.1
big package
1
10
air speed [m/s]
53
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Different boards
• Minpad board
• 1” heat spreader board • You’re mainly characterizing how copper area affects every package and board, not how a particular package depends on copper area
54
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
1" pad vs minpad
350
Roger Stout 5/25/2000
300
Source: Underated thermal data from old PPD database
SOD323
250
TSOP6
SOT23
1" pad thetaJA (C/W)
SOT23
200
SOD123 SOD123 TSOP6(AL42) SOT23
150
Micro 8
TSOP6
SO8
SOD123 SMA & Pow ermite
100
SMB Dpak D2pak & TO220 Top Can Top Can SMC SOT223 SO8
50
overall linear fit is: 1" value = [0.51*(minpad value)  7]
0 0 100 200 300 400 500 600 700 min pad thetaJA (C/W)
55
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Standard coldplate testing
• “infinite” heatsink (that really isn’t) for measuring thetaJC on highpower devices • If both power and coldplate temperature are independently controlled, “two parameter” compact models may be created
56
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Standard coldplate testing
• Detailed design and placement of “case” TC can have significant effect on measured value
57
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
2parameter data reduction
Q
heat up, Q1
Q1 Q2
T1
Q
R1
1 T j T1 R1
(
)
1 T j T2 R2
(
)
This has the form of a twovariable linear equation: heat in, Q Tj R2
y
m1 x1 m2 x2 b
where:
m1
T2 heat down, Q2
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
m2
1 R1 1 R2
x1 x2
(T
j
T1 )
j
(T
T2 )
b
0
58
Corporate R&D : Packaging Technology
A “single coldplate” test
ambient Ta Rja
80.0 0 120. 00 100. 00
Tj
Ta
Tjc ( ¡C)
Tc
coldplate
Tj Rjc Tc coldplate
60.0 0 40.0 0 20.0 0 Incr easing Power, Chuck H eld Cold No P ower, Chuck T emper ature Incr eased 20.0 0 40.0 0 60.0 0 80.0 0 100. 00
0.00 20. 00 0.00 20. 00
Tja (¡C)
59
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
ambient Ta Rba Tb Rjb Tj Rjc Tc coldplate
Tjc ( ¡C)
A “single coldplate” test, package down
120. 00 100. 00 80.0 0 60.0 0 40.0 0 20.0 0 0.00 10. 00 0.00 20. 00 Incr easing Power, Chuck H eld Cold No P ower, Chuck T emper ature Incr eased 10.0 0 20.0 0 30.0 0 Tjb (¡C) 40.0 0 50.0 0 60.0 0
Tj
Ta
Tb
Tc coldplate
60
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Miscellaneous Measurement Topics
• Thermocouple Theory 101 • Infrared Theory 202
61
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermocouple Theory 101
62
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermocouple Theory
region “B”  20cm of wire length between the junction and the door of the temperature chamber region “C” – 10cm of TC wire length passing from inside of chamber door to outside of chamber door
TC junction B A
perfectly uniform temperature chamber at 100°C
C D
TC scanner
perfectly uniform ambient temperature of 25°C
region “A”  a 1cc box around the junction
region “D” – 100cm of TC wire length from outside of chamber to TC measuring instrument
63
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermocouple Theory
• Question #1: – where is the temperature actually being “sensed”? or to put it another way, which region of the four defined above (A, B, C, D), is the one that matters, as far as generating the EMF being measured by the TC scanner? Why?
The emf is generated where there is a temperature gradient
B C D
TC junction A
perfectly uniform temperature chamber at 100°C
perfectly uniform ambient at 25°C
64
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermocouple Theory
• Question #2: – if the purpose of the TC scanner is to measure an EMF, theoretically how much current has to flow in the wire to make the correct measurement?
None. Ideally, a galvanometer circuit balances the EMF at the scanner until no current flows.
Corollary: wire size, and therefore resistance, may affect how long it takes the galvanometer circuit to stabilize.
65 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Thermocouple Theory
• Question #3: – what do the two wires in the TC actually do, and why do they have to be made of different materials?
100°C ΔV=0.0025V
100°C
Different materials have different Seebeck coefficients, i.e., V/°C. So if you have two different materials, the same temperature gradient appearing in along both wires will result in two different EMF’s.
ΔV=0.0037V
25°C
25°C
66
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermocouple Theory
• Question #4: – what does the “junction” in a TC actually do?
25°C
ΔV=0.0025V
100°C
Provides the common electrical point that relates the EMF’s in the two wires to each other, hence the net difference appears at the “open” end of the circuit.
Net ΔV=0.0012V
ΔV=0.0037V
The junction DOES NOT “measure” the temperature; the wires do!
25°C
Corollary: if the junction itself is isothermal, it can be made of any material whatsoever!
67 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Thermocouple Theory
• Question #5: – if you really wanted to “calibrate” this thermocouple, where would you need to apply the test conditions?
TC junction B A C
Everywhere along its length!
D
68
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infrared Theory 202
69
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infrared Theory
A opaque surface radiates thermal energy in proportion to its absolute temperature (to the fourth power), and some other things
q
70 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
AT
4
Corporate R&D : Packaging Technology
Infrared Theory
It’s also absorbing infrared energy from everything else around it (since everything else has a temperature and is therefore emitting)
It turns out that the “absorptivity” equals the “emissivity.”
71 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Infrared Theory
There is thus a net energy transfer which, in the simplest situations, may be described by:
q
obj
Aobj (Tobj
4
Tencl
4
)
(typically applies to a small, isothermal object in a big, isothermal enclosure)
72 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Infrared Theory
But it quickly gets more complicated when there are multiple objects (or temperatures) hanging around. • Each object has a “view factor” of every other object • Each object has its own emissivity (usually NOT unity)
73
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infrared Theory
The “emissivity,” , is also related to the “reflectivity,” , as follows:
1
So the worse something emits, the better it reflects, and viceversa.
74
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infrared Theory
In other words, unless you’re looking at a perfect emitter (aka a “blackbody”), you don’t simply see less radiation than you would for a blackbody, you see some of the radiation of what’s behind you!
75
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infrared Theory
In even the simplest reallife situation, you usually have four “objects,” and their associated temperatures, to consider: • The test specimen (130°C) • The surrounding room (25°C) • You (33°C) • The detector (195°C)
76
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infrared Theory
(the room) The “target”
(the detector itself)
(you)
(the room)
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
77
Corporate R&D : Packaging Technology
Infrared Theory
So if you’re looking at a shiny, low emissivity specimen, you’ll see a combination of these four temperatures (that is, the radiation representing these four temperatures) all bouncing into the detector! Depending on the angles involved, therefore, it is quite possible to see vastly higher temperatures than are really there, or vastly lower temperatures are really there.
78 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Hot water loop
Footprints
79
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Direct view
In mirror
80
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Reflection
Emissivity
81
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
View at 10 am
View at 10 pm
82
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
83
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infrared Theory
Moral:
For accurate temperature measurements, you must account for the emissivity of the target, and you must control the background!
84
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infrared Theory
Here’s one way to approach it:
isothermal enclosure, high emissivity, opaque to infrared
IR camera
DUT
85
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infrared Theory
Governing equation
Two reference scans
I
I back a T 4
1
I1
I2
I back a T14
I back a T2 4
Solve for Temperature
Solve for the unknowns
T
I
I back a
4
a
I2 T2
4
I1 T1
4
I back
T2 4 I1 T14 I 2 T2 4 T14
In terms of the reference scan quantities:
T
T2 (I
4
I1 ) T1 (I I 2 I1
4
I2 )
1
4
86
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Part II Linear Superposition
87
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Linear superposition – what is it?
• The total response of a point within the system, to excitations at all points of the system, is the sum of the individual responses to each excitation taken independently.
Tsource 2
Tcomposite
Tsource1
Tsource n
88
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Linear superposition – when does it apply?
• The system must be “linear” – in brief, all individual responses must be proportional to all individual excitations.
Tnet A
TA
89 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
TA
B
TA
C
TA
D
2 qB
3 qC
Corporate R&D : Packaging Technology
1.2 qD
Linear superposition doesn’t apply if the system isn’t linear.
T
a(T , q1 ) q1 b(T , q2 ) q2
T
a q1
n1
b q2
n2
90
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Linear superposition – how do you use it?
Tamb
Tref5
Tj6 Tj5
Tj3
Tref3
Tj2
Tref1 Tj4 TB
91 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Tj1
Linear superposition – when would you use it?
When you have multiple heat sources (that is, all the time!)
92
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
junction temperature vector
theta matrix assembled from simplified subsystems
power input vector
T j1 Tj2 T jn
93
J 1A 21
12 J 2A
1n 2n
n1 n2
JnA
q1 q2 qn
Ta
selfheating terms
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
board interactions
Corporate R&D : Packaging Technology
junction temperature vector
theta matrix assembled from simplified subsystems
power input vector
T j1 Tj2 T jn
JB1 21
BA1 JB 2
12 BA2
1n 2n
n1 n2
JBn BAn
q1 q2 qn
Ta
device resistance
94 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
board resistance
Corporate R&D : Packaging Technology
visualizing theta and psi
heat in here measurements here are s
J1A J1B
(idle heat source “x”)
measurements here are s (idle heat source “y”)
xA
BA
yA
thermal ground
95 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
theta matrix doesn’t have to be square
junction temperature vector one column for each heat source
JA1 12 x1 L1 1 B1 12 JA 2 x2 L1 2 B2 13 23 x3 L1 3 B3
power input vector
T j1 Tj2 TxA TL1 A TBA
q1 q2 q3
(why is this and not ?)
one row for each heat source
one row for each temperature location of interest
96 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Electrical reciprocity
+ 5V 
+ 
+ ?V 
+
0.3 V V

I 0.3A 2 V
97
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermal reciprocity
heat input here
response here
same response here
98
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Another thermal reciprocity example
(r)
response here
heat input here same response here
(s)
(s)
(r)
99
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
When does reciprocity NOT Apply?
• Upwind and downwind in forcedconvection dominated applications
B
C
airflow A D Heat in at “A” will raise temperature of “C” more than heat in at “C” will raise temperature of “A” “B” and “D” may still be roughly reciprocal
100
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
(square part of) matrix is symmetric
columns are the heat sources J1 J2 J3 J4 rows are the response locations J5 J6 75 65 55 60 22 10 73 65 71 60 55 25 11 65 55 60 65 61 21 15 55 60 55 61 73 18 11 59 22 25 21 18 125 14 22 10 11 15 11 14 180 10
R1
R3 R5 B
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
55
20 65
60
24 63
63
14 62
61
19 63
21
95 21
15
15 12
101
Corporate R&D : Packaging Technology
Superposition example
Tamb=25 Tref5=47.0 Tj6=36.0 Tj5=49.2
Tj3=85.5 Tref3=85.5
Tj2=96.5
Device 1 heated, 1.1 W
Tref1=105.3 Tj4=91.0 Tj1=107.5 TB=96.5
102
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Reduce the data
Tj1 Tamb
j1A
j1A
75 65 55
q1
Tj2 Tamb q1
107.5 25 1.1
96.5 25 1.1
75
j2A j3A
j2 A
65
j4A j5A
60
22 10 73 55 20 65
TB
BA
j6A r1A
Tamb q1
96.5 25 1 .1
65
r3A r5A BA
103
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Device 2 heated, 1.2 W
j1A j2A
65 71
Tamb=25 Tref5=53.8 Tj6=38.2 Tj5=55.0
j3A
j4A j5A j6A r1A
60
55 25 11
65
60 24 63
Tj3=97.0 Tref3=97.0
Tj2=110.2
r3A r5A
Tref1=103.0 Tj4=91.0 Tj1=103.0 TB=100.6
104 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
BA
Device 3 heated, 1.3 W
j1A
55 60
Tamb=25
j2A
Tref5=43.2 Tj6=44.5
j3A
65
61 21 15
Tj5=52.3
j4A j5A j6A
Tj3=109.5 Tref3=106.9
Tj2=103.0
r1A r3A
55
63 14 62
Tref1=96.5
Tj4=104.3
Tj1=96.5
TB=105.6
r5A BA
105
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Device 4 heated, 1.1 W
j1A
60 55
Tamb=25
j2A
Tref5=45.9 Tj6=37.1
j3A
61
73 18 11
Tj5=44.8
j4A j5A j6A
Tj3=92.1 Tref3=92.1
Tj2=85.5
r1A r3A
59
61 19 63
Tref1=89.9
Tj4=105.3
Tj1=91.0
TB=94.3
r5A BA
106
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Device 5 heated, 0.7 W
j1A
22 25
Tamb=25
j2A
Tref5=91.5 Tj6=34.8
j3A
21
18 125 14
Tj5=112.5
j4A j5A j6A
Tj3=39.7 Tref3=39.7
Tj2=42.5
r1A r3A
22
21 95 21
Tref1=40.4
Tj4=37.6
Tj1=40.4
TB=39.7
r5A BA
107
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Device 6 heated, 0.5 W
j1A
10 11
Tamb=25
j2A
Tref5=32.5 Tj6=115.0
j3A
15
11 14 180
Tj5=32.0
j4A j5A j6A
Tj3=32.5 Tref3=32.5
Tj2=30.5
r1A r3A
10
15 15 12
Tref1=30.0
Tj4=30.5
Tj1=30.0
TB=31.0
r5A BA
108
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Collect the /
J1 J2 J3 75 65 55 60 22 10 65 71 60 55 25 11 55 60 65 61 21 15
values
60 55 61 73 18 11 22 25 21 18 125 14 10 11 15 11 14 180
columns are the heat sources
J4
rows are the response locations J5 J6 R1 R3 R5 B
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
73
55 20 65
65
60 24 63
55
63 14 62
59
61 19 63
22
21 95 21
10
15 15 12
109
Corporate R&D : Packaging Technology
Now apply actual power
Tamb=25
Tref5=106.3
Actual power in application
Tj5=124.7
Tj6=139.1
Qj1
.4
Q j2
Tj3=134.9 Tj2=140.1
.4
.4 .4 .5 .2
Q j3 Q j4 Q j5 Q j6
Tref3=134.1
Tref1=138.8 Tj4=135.8 Tj1=140.0 TB=139.1
110 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Compute some effective /
values
Take Tj1, for instance. Remember when it was heated all alone, we calculated its selfheating thetaJA like this:
Tj1 Tamb
j1A
q1
107.5 25 1.1
75
Now let’s see:
Tj1 Tamb
j1A
q1
140 25 0.4
288
111
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
And that’s not just a single aberration!
Junction to Reference
Self heating
j1A j2A j3A j4A j5A j6A j1R1
3.0 2.0 36.8
vs. 1.5x vs. 1.0x vs. 1.2x
2.0 2.0 30.0
288
288 274 277 199 309
vs. 3.8x vs. 4.1x vs. 4.2x vs. 3.8x
75
71 65 73
j3R3 j5R5
Junction to Board
j1B
2.2
vs. 0.2x
10.0
vs. 125 1.6x vs. 180 1.7x
j2B
j3B j4B
2.5
10.5 8.3
vs. 0.3x
vs. 3.5x vs. 0.8x
8.0
3.0 10.0
112
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Is the moral clear?
• You simply cannot use published thetaJA values for devices in your real system, even if those values are perfectly accurate and correct as reported on the datasheet and you know the exact specifications of the test conditions. • Not unless your actual application is identical to the manufacturer’s test board – and uses just that one device all by itself.
113
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
So is it really this bad?
Only sortof. Let’s revisit the math for one device …
Tj1 Tj2 Tjn
J1A 12
12 J2 A
12 q2
n
1n 2n
1n
Tj1
Tj1
JnA
1n qn
2n
J1A q1
q1 q2 qn
Ta
Ta
Ta
1nqn
J1A q1 2
“effective” ambient
114
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
A graphical view
Isolated device
Tj1
J1A q1
power, q
1
Ta
Ta
J1A
junction temperature , TJ1
Device in a system
n
shift in effective ambient
Tj1
J1A q1 2 J1A q1
1n qn
Ta
Ta’
1
J1A
still the same slope
Ta
Ta
junction temperature , TJ1
115
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
What about that “system” theta we saw earlier that was so different?
power q1 the “system” thetaJA 1
J1A J1A
device #1 power/temperature perturbations will fall on this line
NOT this one
1
n
Ta
116
1n qn 2
Ta’
the isolateddevice thetaJA TJ1
junction temperature
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
How does effective ambient relate to board temperature?
“system” slope for isolated device
if any of these are nonzero, Ta will be higher than Ta
n
Tj1
j1a
Q1
i 2
B1a
(
i1
Qi ) Ta
(
j1B
)Q
B1a
1
effective Ta ambient
j1B
Q1
Q1
TB1a
Ta
Ta
Tj1B
temperature rise, board to J1
117 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
when Q1 is not zero, both zero, both of of these be these willwill be nonzero zero
temperature rise, ambient to board
Corporate R&D : Packaging Technology
Predicting the temperature of high power components
• The device and system are equally important to get right
Predicting the temperature of low power components
• The system is probably more important than the device
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
118
Corporate R&D : Packaging Technology
Using the previous board example …
theta array
J1 J2 J3
75 65 55 60 22
65 71 60 55 25
55 60 65 61 21
60 55 61 73 18
22 25 21 18 125
10 11 15 11 14 Qj1
power vector
0.5
J4
J5 J6 R1 R3 R5 B
Q j2 0.5
Q j3 0.5 Q j4 0.5 Q j5 0.2 Q j6 0.02
10
73 55 20
11
65 60 24
15
55 63 14
11
59 61 19
14
22 21 95
180
10 15 15
65
63
62
63
21
12
Corporate R&D : Packaging Technology
119
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Observe the relative contributions
For junction 1 (a high power component) we have: the device itself … the other devices … = (75 x 0.5) + (65 x 0.5) + (55 x 0.5) + (60 x 0.5) + (22 x 0.2) + (10 x 0.02) + 25
=
37.5
+ 32.5 + 27.5 + 30 + 4.4 + 0.2
+
25
= 37.5 +
120 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
94.6
+
25
Corporate R&D : Packaging Technology
Graphically, a highpower device looks like this:
power q1=0.5 W note the “embedded” thetaJA looks like 264 C/W decreasing power
increasing power
1
264 C/W
1
75 C/W
=94.6 C
n
25 C
1n qn 2
=37.5 C (θJ1Aq1)
157 C TJ1
junction temperature
121
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Relative contributions to TJ6
the other devices … = (10 x 0.5) + (11 x 0.5) + (15 x 0.5) + (11 x 0.5) + (14 x 0.2) + (180 x 0.02) + 25 the device itself …
= 5.0 + 5.5 + 7.5 + 5.5 + 2.8 + 3.6 = 26.3 + 3.6 + 25 + 25
122
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Graphically, lowpower device #6 looks like this:
power and just in case you were wondering, the “embedded” thetaJA looks like 1495 C/W !
1 =3.6 C
q6=0.02 W
=26.3 C
180 C/W
25 C
54 C TJ6
junction temperature
123
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Controlling the matrix
How to harness this math in Excel®
124
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
3x3 theta matrix, 3x1 power vector Excel® math
Matrix MULTiply obtained by using multicell placement CtrlShiftEnter rather {=array formula notation} than ordinary Enter of array formula
theta matrix
power vector
array reference to theta matrix
array reference to power vector
125
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
7x3 theta matrix, 3x1 power vector Excel® math
don’t forget to use theta matrix is no longer square – CtrlShiftEnter # of columns still must equal # of rows of power vector to invoke array formula notation array formula now occupies 7 cells
126
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
7x3 theta matrix, 3x2 power vector Excel® math
power “vector” is now a 3x2 array – each column is a different power scenario, yet both are still processed using a single array (MMULT) formula the single MMULT array formula now occupies 7 rows and 2 columns (one column for each independent power scenario result)
127
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Temperature direct contributions and totals
120
120
80
temperature rise [C] sum of sources
J1 J2 J3 J4 J5 J6 R1 R3 R5 B
temperature rise [C], each source
100
100
80
60
60
40
40
20
20
0 result location
0 J1 J2 J3 J4 J5 J6 R1 R3 R5 J3 at 0.4 W J6 at 0.2 W B result location J3 at 0.4 W J6 at 0.2 W J1 at 0.4 W J4 at 0.4 W J2 at 0.4 W J5 at 0.5 W
J1 at 0.4 W J4 at 0.4 W
J2 at 0.4 W J5 at 0.5 W
128
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Normalized responses at each location due to each source
200
normalized response [C/W], each source
180 160 140 120 100 80 60 40 20 0 J1 J2 J3 J4 J5 J6 R1 response location R3 R5 B J1 at 1 W J2 at 1 W J3 at 1 W J4 at 1 W J5 at 1 W J6 at 1 W
129
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Filling in the thetamatrix
• Handy formulas for quick estimates • Utilizing symmetry
130
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Conduction resistance
basic heat transfer relationship for 1D conduction
q dT k A dx k A T L
if we define
R T q
then
R L k A
Corporate R&D : Packaging Technology
131
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Convection resistance
basic heat transfer relationship for surface cooling
q h A T
if we define
R T q
then
R 1 hA
Corporate R&D : Packaging Technology
132
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Radiation resistance
basic heat transfer relationship for surface radiation
q F A (T 4 Ta4 ) F A (T 2 Ta2 )(T Ta )(T Ta ) F A (T 2 Ta2 )(T Ta ) T
if we define
R T q
then
R 1 FA (T 2 Ta2 )(T Ta )
temperatures must be expressed in degrees “absolute”!
133
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermal capacitance and time constant
Capacitance is ability to store energy specific heat is energy storage/mass Based on simple RC concept, relate rate of storage to rate of flux, result is
C
so if
cp V
and if
= RC
R
then
L and C k A
c pL2 k L2
ρc p (L A )
R
then
1 and C h A
c pL h
ρc p (L A )
134
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Some useful formulas
• conduction resistance…………..……… R
• convection resistance…………...……… R
L k A 1 h A c pV c pL2 k c pL T h Q A 2 cp k t
• thermal capacitance……………...…….. C
• characteristic time…………………..…. – (dominated by 1D conduction) • characteristic time……………………... – (dominated by 1D convection) • shorttime 1D transient response……...
135
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Terms used in preceding formulas
• • • • • • • • • • • L  thermal path length A  thermal path crosssectional area k  thermal conductivity k  density cp cp  heat capacity  thermal diffusivity cpk  thermal effusivity h  convection heattransfer “film coefficient”) T  junction temperature rise Q  heating power t  time since heat was first applied
136
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
When do these effects enter?
hundreds of seconds tens of seconds a second or so
junction temperature
mainly environmental convection and radiation effects
mainly local application board conduction effects
typical heating curve for device on FR4 board in stillair
time
mainly package
materials/conduction effects
137 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
if
R
then
and
2R
4R
138 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Cylindrical and spherical conduction (through radial thickness) resistance formulas
ro ln ri k L
[solid angle]
Halfcylinder
R
R
1 ri 2 1 ri 4
1 ro k 1 ro k
Hemisphere
[included angle]
Full cylinder
R
r ln o ri 2 k L
• • •
R
Full sphere
where
L – cylinder length ri – inner radius ro – outer radius
139
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Packageshrink “gotcha”
Remember how much of thetaJA depends on what isn’t the package?
Well, what if your cooling depends significantly on convection from the board surface (whether free or forced air)?
q
h A
T
A
q h T
So never mind the package resistance, the board can only transfer a certain amount of heat to the air:
*Special thanks to Dave Billings for the idea behind this slide
140 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Packageshrink “gotcha”
1000 A 2 mm
1.0q W
W 1E5 mm2 °C
h
100°C T
So if 4 SOT23’s each have 0.25 W, then T=0.25*200=50°C PROBLEM: No problem! 4 SOT23’s on a 1000 mm2 board, effective JA=400°C/W
1 SOT23 on a 250 mm2 spreader, JA=200°C/W
Fortunately, 0.25 W each x 400°C/W→100°C
*Special thanks to Dave Billings for the idea behind this slide
141 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Whew!
Packageshrink “gotcha”
NOW your favorite supplier comes out with the next generation version (much smaller) of the same component.
SOT23 4 SOT723’s on a 1000 mm2 SOT23’s board, JA=400°C/W, T=100°C, power = 1 W total, power = 1 W total, 0.25 W each each
Decrease size (but not power and reduce power dissipation dissipation) (RDSON or other electrical performance)
How many SOT723’s can you put in that same 1000 mm2? ANSWER: 4 8
8 SOT723’s on a 1000 mm2 board, JA=800°C/W, T=100°C, power = 1 W total, 0.125 W each
SOT723
*Special thanks to Dave Billings for the idea behind this slide
142 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
PERIODIC and NONPERIODIC POWER INPUT
• Dutycycle curves – Normalized and nonnormalized • Linear superposition applied to time domain
143
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Squarewave duty cycles (aka ratios)
• What the heck are they?
– show peak junction temperature as a function of duty cycle and “pulse width”
• Where do they come from?
– “heating curve” is equivalent to the limiting case of a 0% dutycycle, where once the “pulse width” is over, there’s never another cycle – “linear superposition” allows you to generate the whole family from the heating curve
144
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Representative “duty cycle” response chart
2.5 R(t), Thermal Resistance [°C/W]
2
1.5
d=0.5
1
0.2 0.1
0.05 Single pulse
0.5
0 0.00001
0.0001
0.001
0.01 t, time (s)
0.1
1
10
145
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
It begins with one of these:
100
transient thermal response for axial lead rectifier JunctiontoLead, 1/4" lead length
10
R(t)JL [°C/W]
1
0.1 0.0001
0.001
0.01
0.1
1
time [s]
10
100
1000
A “singlepulse” response is the same as a 0% duty cycle  if you think in terms of a 0% duty cycle meaning you have only one pulse (however long it lasts), but once you turn if off, you never turn it back on again. So any finite “on” time is zero percent of infinitely long “off”.
146 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
“published” formulas
Nondimensionalized versions
r(t , d ) d
r(t , d ) d
dimensionalized versions
R(t , d ) d R
(1
(1
d ) * r(t )
d ) * r(t p) r(t ) r( p)
(1
d ) R(t )
r (t , d ) d
(1
d)*r t
t d
r (t ) r
t d
R(t , d ) d R
(1
d) R t
t d
R(t ) R
t d
lim R(t , d ) d R
t 0
lim R(t , d )
t
R
147
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
A “normalized” curve:
1
JL=27.4°C/W
r(t)JL [normalized]
0.1
normalized transient thermal response for axial lead rectifier JunctiontoLead, 1/4" lead length
0.01 0.0001
0.001
0.01
0.1
1
time [s]
10
100
1000
Is this really such a bright idea?
148
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Consider this family of “nonnormalized” curves:
100
Transient thermal response for axial lead rectifier JunctiontoLead, varying lead length
10
R(t)JL [°C/W]
R(t)1" R(t)3/8" 1 R(t)1/32"
0.1 0.0001
0.001
0.01
0.1
time [s]
1
10
100
149
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
The danger!
100
Transient thermal response for axial lead rectifier JunctiontoLead, varying lead length
10
r(t)JL [normalized]
See what happens to the shorttime response when you “normalize”!
R(t)JL [°C/W]
R(t)1" R(t)3/8" 1 R(t)1/32"
0.1 0.0001 1
0.001
0.01
0.1
time [s]
1
10
100
JL1/32"=20.0°C/W JL3/8"=34.9°C/W
0.1
JL1"=54.9°C/W
0.01
r(t)1/32" r(t)3/8" r(t)1"
normalized transient thermal response for axial lead rectifier, JunctiontoLead varying lead length
0.001 0.0001
0.001
0.01
0.1
time [s]
1
10
100
150
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Compare the families of curves:
100
R(t)JL [°C/W]
Different steady states
squarewave duty cycles for 1/32" lead length
10
single pulse 1% duty cycle 2% duty cycle 5% duty cycle
1
10% duty cycle 20% duty cycle 50% duty cycle
Same shorttime 0% (singlepulse)
0.1 0.0001
0.001
0.01
0.1
time [s]
1
10
100
100
squarewave duty cycles for 1" lead length
R(t)JL [°C/W]
Different everything between !!
10
single pulse 1% duty cycle 2% duty cycle 5% duty cycle
1
10% duty cycle 20% duty cycle 50% duty cycle
0.1 0.0001
0.001
0.01
0.1
time [s]
1
10
100
151
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
What to do if you’re given a normalized curve
• Find out what reference value it was normalized from
– Undo it
• If you can’t find out
– Throw the curve away
152
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Linear Superposition Applied to Time Domain
153
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Basic heating curve  a “single pulse”
R(t)
t
Q
t
power input corresponding to single pulse heating curve
154 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Single pulse (actually turned off)
Finite pulse, decomposed into two infinite steps
Q
equals
a t
Q
plus
t
Q a t
Temperature response of a finite pulse (constructed from superposition of two single pulse responses)
R(t) R(t) R(t) a t
plus
t
equals
a t
155
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Two differing pulses
Two finite pulses decomposed into infinite steps
Q1 a Q2 b c Q1 t a Q1 Q2 b c Q2 t
is made up of
Temperature response constructed for two finite pulses (constructed from superposition of four single pulse responses)
R(t)
R(t)
results in this
156
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
An arbitrary pulse train
• Temperature at end of nth pulse
n
Tn
i 1
Pi R(t2 n
[
1
t 2i
2
)
R(t2n
1
t 2i
1
)]
Notes: times and pulses must be in strictly increasing chronological order
P1 Pn P2
When there is no “off” period between two consecutive pulses, set t 2i 2 t 2i 3 (i.e. do not “combine” them into a single value)
t2n1
Power
t0 t1 t2
t3 time
t2n2
For temperature at the beginning of the nth pulse, set t 2n 1 t 2n 2
157
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
An example
Item P1 P2 P3 t1 t3 t31 t32 t5 t51 t52 t53 t54
t0 t6 t4
Junction Temperarure
Value 80 40 70 0.0001 0.0013 0.0012 0.0010 0.0035 0.0034 0.0032 0.0022 0.0002
Item
unit W W W s s s s s s s s s
unit
P(PK) Peak Power (Watts)
100 80 60 40 20 0 t0 t1 t2 1 t3 2 t, Time (ms) 3 t4 t5 4 P1 P2 P3
T1
T2
P R(t1 ) 1
P R(t3 ) R(t3 t1 ) 1 P2 R(t3 t 2 )
[
]
T3 T1 T2
Value
T3
3 t4 t5 4
P R(t5 ) R(t5 t1 ) 1
[
] ]
0.0000 s 0.0003 s 0.0012 s
t0 t1 t2
1 t3 2 t, Time (ms)
P2 R(t5 t 2 ) R(t5 t3 ) P3 R(t5 t 4 )
[
158
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
How to read the curve
• For small time differences, and on a low resolution loglog plot, what can you do?
100
Transient thermal response for axial lead rectifier JunctiontoLead, varying lead length
10
1
0.1 0.0001
0.001
Item P1 P2 P3 t1 t3 t31 t32 t5 t51 0.01 t52 t53 t54
Value 80 40 70 0.0001 0.0013 0.0012 0.0010 0.0035 0.0034 0.0032 0.0022 0.0002
unit W W W s s s s s s 0.1 s s s
R(t)JL [°C/W]
R(t)1" R(t)3/8" R(t)1/32"
time [s]
1
10
100
159
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Assume power law (straight line on loglog plot)
R(t ) a t
n
t2
t1
t1 t
a R(t1 ) t1
n
R(t 2 ) t2
n
R(t
) R(t )
R(t
t t
n
n
R(t ) 1
log n
R(t 2 ) log t2
t
R(t1 ) t1
)
R(t )
R(t )n
t
160
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
How to read the curve #2
• For small time differences, and on a low resolution loglog plot, what can you do?
100
anchors R(.0001) 0.22 R(.02) 3.3
Transient thermal response for axial lead rectifier JunctiontoLead, varying lead length
10
1
0.1 0.0001
0.001
0.01
0.1
time [s]
1
interpolate R(t1) 0.22 R(t3) 0.82 R(t31) 0.79 R(t32) 0.72 R(t5) 1.36 R(t51) 1.34 10 R(t52) 1.30 R(t53) 1.08 R(t54) 0.32
R(t)JL [°C/W]
R(t)1" R(t)3/8" R(t)1/32"
100
161
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Results
P(PK) Peak Power (Watts)
100 80 60 40 20 0 t0 t1 t2 1 t3 2 t, Time (ms) 3 t4 t5 4 P1 P2 P3
T1 T2
P R(t1 ) 80 0.22 17 .6 C 1 P [ R(t3 ) R(t3 t1 )] P2 R(t3 t 2 ) 1 80 (0.82 0.79 ) 40 0.72 2.4 28 .8 31 .2 C
Junction Temperarure
T3 T1 T2
T3
P [ R(t5 ) R(t5 t1 )] 1 P2[ R(t5 t 2 ) R(t5 t3 )] P3 R(t5 t 4 ) 80 (1.36 1.34 ) 40 (1.30 1.08) 70 0.32 1.6 8.8 22 .4 32 .8 C
t0 t1 t2
1 t3 2 t, Time (ms)
3 t4
t5
4
162
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Another transient analysis example
power
18
25.2W
23
Here’s our power input scenario:
power 20 pulses 25.2W
repeat 20 times, then stop for 6800 microseconds
time
20 pulses
time 802 last of 20 pulses ends 7600
163
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
equivalence of nonzero average response with (constant average) + (zeroaverage disturbance)
Actual duty cycle and response
“peak” Tavg “valley” Tavg due to Qavg
Superposition of average and disturbance
Power shifted to average zero
Q
0
Qavg= d · Q
(1d) · Q
0
d·Q
164
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Another example, cont’
How do we analyze it? Over three time scales:
power
18
25.2W
instantaneous excursions based on “local” duty cycle
23
power
“average” Tj based on local avg power
average excursions based on “global” duty cycle
repeat 20 times
time
25.2W
burst
time
power 25.2W
7600
15200
“average” Tj based on global avg power
20 pulses time 802 last of 20 pulses ends 7600
165
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Endless Possibilities
166
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
A “ramp”
A finite ramp decomposed into several infinite steps
P
t a
P
is made up of
a
t
Temperature response constructed for finite ramp (constructed from superposition of many single pulse responses)
Q t a Q
results in this
a
t
167
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
An arbitrary pulse
An arbitrary pulse decomposed into several infinite steps
Q t a Q
is made up of
Temperature response constructed for this arbitrary pulse
Q Q
t
results in this
t
168
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Multiple TimeVarying Heat Sources
169
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Basically the same thetamatrix idea, only now everything is a function of time as well
junction temperature vector
T j1 (t ) T j 2 (t ) TxA (t ) TL1 A (t ) TBA (t )
one column for each heat source
(t ) 12 (t ) x1 (t ) L1 1 (t ) B1 (t ) (t ) JA 2 (t ) x 2 (t ) L1 2 (t ) B 2 (t )
12
power input vector
JA1
(t ) 23 (t ) x 3 (t ) L1 3 (t ) B 3 (t )
13
q1 (t ) q2 (t ) q3 (t )
one row for each heat source
one row for each temperature location of interest
170 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Each heat source needs to be independently heated.
Each temperature needs to be measured whether heating itself or being heated by another.
Time (sec) thermal system boundary
Measurement cycle(s)
Time (sec)
*Special thanks to Dave Billings for this slide
171 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Temperature (°C)
Power input (W)
but first, some words about …
Thermal RC networks
• Grounded (Cauer) vs.. nongrounded (Foster) • Orders of magnitude, rungs • Pulses and periodic waveforms • Shorttime behavior, limits, and limitations
172
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermal capacitance (grounded vs.. nongrounded)
• In electrical circuit, voltage on a capacitor is difference between its two terminals, and you have physical access to both • A “lumped parameter” thermal model is predicated on energy storage being determined entirely by one temperature • Therefore, in a thermal circuit, you only have access to one “terminal” of each capacitor, namely the one where you identify the temperature. The other terminal is “ground”.
173 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Models: grounded vs.. nongrounded
“Cauer” ladders “Foster” ladders
• Physical significance: if thermal capacitors are grounded, they bear some relationship to a physical system; not so for nongrounded C’s • Mathematical convenience: certain nongrounded networks are mathematically “trivial” • Interchangeability: singleinput thermal systems can be represented as either grounded or nongrounded; • multipleinput thermal systems are easy to model as groundedC networks; it can be done, though with some intricate bookkeeping, using nongroundedC networks
174 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Actual thermal RC networks
Nongrounded C
“Foster” ladder
1.7004 ° C/W 2.6627 3.9740 10.0255 11.7747 35.3008 33.1212 Tj
vs..
grounded C comparison
“Cauer” ladder
Tj
7.02E04 Wsec/°C 2.3207 ° C/W 4.41E03 5.89E02 1.55E01 6.03E01 1.46E+00 4.16E+00
2.7397 8.3551 14.6949 21.0538 39.637 9.7581 Ta
5.96E04 Ws ec/°C 4.20E03 3.67E02 1.01E01 4.13E01 9.20E01 1.14E+01
Ta
175 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
2rung Foster model
R1
C1
R1
1 sC1
1 sC1 1 R1
R1 R1C1 s 1
R2 R2 C2 s 1
R2
C2
R2
1 sC2
R2 R2 C2 s 1
176
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
2rung Cauer model
C1 R1
1 sC2
R1
1 sC1
R1
R2 R2 C2 s 1
1 sC1
R2
C2
R2
R1
R2 R2 C2 s 1
1 sC1
1 sC1 1 R1 R2 R2 C2 s 1
R1 R2 C2 s R1 R2 R1C1 R2 C2 s R1C1 R2 C1 R2 C2 s 1
2
177
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
2rung models compared
• • Time constants are roots of denominators "tau" is not RC product when capacitors are grounded !
nongrounded capacitors:
(Foster)
R1 R1C1 s 1 1 1 C1 s 1
1
2
groundedcapacitors:
R1 R2 C2 s R1 R2 R1C1 R2 C2 s R1C1 R2 C1 R2 C2 s 1
where time constants are 2 R1 C1
1
(Cauer)
s s R1 R2 R1 R2 C 2 1
1
can be written
1 C1
s
1
2
R2 R2 C2 s 1 1 1 C2 s 1
2
can be written
1
c r
c
1
c r
2
c
c 4 r
r 100 10 3 1 1
c 0.01 0.1 1 /3 0.1 1
1
2
R1C1
R2C2
2 R2 C2
2
where time constants are
1
r 1 c
r
r
r 1 c
R2 R1 c
2
r
C1 C2
4
r c
R1C1
2
R2 C2
and defining
0.99 0.91 0.73 0.90 0.38
1.01 1.10 1.36 1.11 2.62
178
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Interesting (and important) implications
“Foster” ladder
Rungs can be in any order and Tj has identical behavior! So where do you “split” it?
Tj
“Cauer” ladder
Order matters, so a “split” can make good physical sense. Tj
2.3207 ° C/W 5.96E04 Ws ec/°C 4.20E03 3.67E02 1.01E01 4.13E01 9.20E01 1.14E+01 Ta
1.7004 ° C/W 2.6627 3.9740 10.0255 11.7747 35.3008 33.1212
7.02E04 Wsec/°C 4.41E03 5.89E02 1.55E01 6.03E01 1.46E+00 4.16E+00
package
2.7397 8.3551
??
14.6949 21.0538 39.637 9.7581
environment
Ta
179
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Orders of magnitude and rungs
• It is a rare transient curve that cannot be followed very accurately with time constants no closer than about 1 order of magnitude apart. This means that you need only about one rung per decade of transient response.
180
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
For what it’s worth ...
Every Cauer (groundedcapacitor) ladder has a mathematical solution. By definition, its mathematical solution represents a corresponding Foster (nongroundedcapacitor) ladder. Every Foster (nongroundedcapacitor) ladder having distinct (i.e. nonrepeated) time constants, has a corresponding Cauer (groundedcapacitor) equivalent. (However, there is no a priori guarantee that all R’s and C’s will be positive!)
Ref: L. Weinberg, Network Analysis and Synthesis, McGraw Hill Book Company, Inc., 1962
181
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
RC network math
Cauer ladder continued fraction:
Z sC1 R1 sC 2 1 1 1 1 R2 ... 1 ...
with some algebra, both may be written:
Z Nn s Dn 1 s
Foster ladder, sum of simple fractions:
Z R1 R1C1s 1 R2 ... R2 C 2 s 1 Ri ... Ri Ci s 1
182
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Arbitrary repetitive pulse
Q
t p 2p
p – period of waveform
183
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Generalized periodic square wave
So consider a general rectangular pulse positioned arbitrarily within a
repetitive cycle of period p. We can derive expressions for the
infinite sum of responses in three regions :
Q t – arbitrary time of interest
m t a
R(t )
i 1
Ri 1 e Fi (t )
j 1
i
t a jp
Ri 1 e
m
i
t
R(t )
i 1
Ri 1 e
i
p b – pulse turns off a – pulse turns on
2p
t′ – measured from edge of power step
m t b
R(t )
i 1
Ri 1 e
t b jp
i
0 t
184
a
a t
b
b t
p
Fi (t )
j 1
Ri 1 e
i
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Infinite summation
Utilizing the identities
j 0
zj
1 1 z
and
j 1
z
j
1 z 1
and doing some algebra, we can show these results:
0 t
b t p
a
a t p
a t
i
b
b t p a t
i
b t
e
p
i i
p
b t a t
Fi (t )
Ri
e
i
e
p
i
Fi (t )
Ri 1
e
Fi (t )
Ri
e
i
e
p
i
i
1 e
1 e
1 e
m
In each domain, we then sum over all rungs of the Foster model:
F (t )
i 1
Fi (t )
185
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
RCEnd result for arbitrary periodic power Model for Multiple Square Waves in One Cycle
An important point of this example is that for complex power inputs, absolute peak temperatures do not always correspond to the ends of the highest power pulse!
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
186
Corporate R&D : Packaging Technology
Shorttime limits
• What do RC models do at shortest time? – roll off linearly with time
t t
R(t )
R1 1 e
1
R2 1 e t
1
2
t
2
R1 1 R1 t
1
1
t2 2 21
R2 1 R1 t
1
1
t2 2 2 2
t2 2 21
t2 2 21
R(t )
R1
1
R2
2
t
R1
2 1
R2
2 2
t2 2
So for small t: R(t ) (const) t
187
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
RC models and Surface Heating
• It turns out that if R’s and C’s grow at a uniform ratio, a sqrt(t) behavior ( b t ) is approached in the limit • If ratio is 1:3 (increasing with “distance” from junction node), rungtorung time constant ratio will be about
one order of magnitude, and the network will follow theoretical sqrt(t) within a few percent
188
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Rungs vs.. sqrt(t) behavior
1.0E+0
Exact sqrt(t) model
1.0E1
1.0E2
1 rung 1.0E3 2 rungs @ 3.0:1 3 rungs @ 3.0:1 4 rungs @ 3.0:1 1.0E4 1.0E8 5 rungs @ 3.0:1 1.0E7 1.0E6 1.0E5 1.0E4 1.0E3 1.0E2 1.0E1 1.0E+0 1.0E+1
189
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Where does that leave us with respect to the applicability of RC networks?
• We need to pay attention to silicon thickness vs.. device “technology”
– if surface heating is a good model, be sure your RC network has time constants short enough to land “within” the silicon – within the silicon timescale
L2
, if surface heating is a good
model, you have to “extend” the RC network into that region – on the other hand, if surface heating is not a good model (implying that volumetric heating is better), the fastest rung of your RC network can be set to the actual R&C corresponding to the volume being heated
190 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Multiple heatsource RC networks
• You need data (experiment or simulation) • Fit an RC network (or networks) to the data
– Foster networks, we’ll see an Excelbased approach – Cauer networks, beyond the scope of this tutorial
• Using SPICE: directly input your RC networks
– Foster networks can be complicated – Cauer networks are straightforward
• Using Excel:
– Foster networks, we’ll see an Excelbased approach – Cauer networks are not practical
191
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Typical 2input heating/response curves
250
q1 self avg q2 self avg interact
200
q1
R(t) [C/W]
150
q1 RCfit q2 RCfit q1<>q2 fit
100
q2
50
0 1E4
1E3
1E2
1E1
1E+0
1E+1
1E+2
1E+3
heating time [s]
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
192
Corporate R&D : Packaging Technology
Fitting Foster ladders in Excel
193
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Using RMSE as fitting parameter
A 1 B C D E F G H I J K L
RMSE test pow er [W] Time
1.9 1.00
1.3 1.00
0.5 1.00 TABLE 1: Foster model taus 1.0E4 1.0E3 1.0E2 1.0E1 1.0E+0 1.0E+1 1.0E+2 q1 R's 11.2 7.6 12.5 76.6 70.0 2.3 41.0 q2 R's 0.8 2.9 2.3 45.6 67.5 2.2 37.7 interact R's 0.1 0.3 0.7 2.3 38.4 4.0 34.1
2
3
q1 self q2 self q1 RC q2 RC q1<avg avg interact fit fit >q2 fit 8.48 9.68 10.59 11.82 14.14 14.64 15.37 15.87 16.37 16.98 18.45 18.58 20.29 1.16 1.32 1.50 1.67 1.84 2.02 2.21 2.42 2.65 2.89 3.14 3.17 3.76 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 9.18 10.51 11.67 12.63 13.45 14.14 14.80 15.46 16.16 16.86 17.60 18.39 19.21 1.01 1.20 1.40 1.59 1.79 1.98 2.19 2.41 2.67 2.92 3.20 3.50 3.80 0.03 0.03 0.03 0.03 0.02 0.01 0.00 0.01 0.02 0.03 0.04 0.04 0.05
4 5 6 7 8 9
1.26E4 1.64E4 2.09E4 2.60E4 3.18E4 3.82E4 4.58E4 5.48E4 6.57E4 7.78E4 9.19E4 1.09E3 1.28E3
10
11 12 13
{=SQRT(SUMSQ(E4:E100TABLE 5: results at arbitrary times B4:B100)/COUNT(E4:E100))} 120.5 100.7
0.00 25.0 {=$G$2*SUM(interact_Rs 25.0 *(1EXP(A4/taus)))} 30.0 0.01 59.6 0.05 90.0 39.1
14
15 16
194
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Excel’s “Solver”
Use Excel’s “solver” to minimize the error between the input data and the RC model.
1000 0.6 0.4 100 0.2 0
0.2 10 0.4 0.6 1 Simlation data RC model Fit Error 0.8 1 1.2 100 1000
0.1 1E06
1E05
1E 0.001 0.01 04
Simulation data
0.1
1
10
Time (sec)
After optimization
195
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Fit error (C/W)
R(t) (C/W)
Using SPICE to exercise the models
196
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Using Foster RC models in SPICE
Tself1 (volts) Tself2 (volts)
Q1 (amps)
+ + Tj1 Tpos1 Tj2 
Tpos2 time
Tneg2
Tneg1
Q2 (amps)
time
different networks of self heating elements
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
identical networks of positive interaction elements
Corporate R&D : Packaging Technology
identical networks of negative interaction elements
197
1/4th of a 4input Foster RC SPICE model
Piecewise linear current source for power input from each source generating heat.
Summing tool to add voltages from the separate interaction networks with the self heating network Thermal equivalent Foster RC networks (note that all these R’s are positive) The output port (OUT1) will be where you want to monitor the temperature response Each heat source will require a similar block in order to simulate the temperature response of the self heating effect as well as the interactions. Thermal ground – by adding a voltage potential to the ground point ambient temperature can be added.
*Special thanks to Dave Billings for this slide
198 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
A 2input Cauer network model
Tj1 C11 R11 R12 C12 C22 R22 C23 R23 C24 time R21 Tj2 Apply q1 to Tj1 node
C21
q1 (watts)
C13
R33 R13 C14 R14 C5 R44
R24
R5 C6 R6 q2 (watts) Apply q2 to Tj2 node
C7 R7 time
199
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Using Excel to exercise Foster models
200
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Setting up a 2input Excel model
I
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
J
K
L
M
N
ambient
O
25
P
Q
R
S
T
U
V
W
X
Y
Z
TABLE 1: Foster model taus 1.0E4 1.0E3 1.0E2 1.0E1 1.0E+0 1.0E+1 1.0E+2 q1 R's 11.2 7.6 12.5 76.6 70.0 2.3 41.0 q2 R's 0.8 2.9 2.3 45.6 67.5 2.2 37.7 interact R's 0.1 0.3 0.7 2.3 38.4 4.0 34.1
TABLE 2: pow er input vs time t0 change_at q1_pow er q2_pow er delta q1 delta q2 time 0.5 0.00 1.0 0.5 1.0 0.5 t1 0.10 1.0 0.0 0.0 0.5 t2 0.25 0.0 1.3 1.0 1.3 t3 0.40 0.8 0.9 0.8 0.4 t4 1.00 0.8 0.0 0.0 0.9 t5 1.10 0.0 0.0 0.8 0.0 t6 1.50 0.0 1.0 0.0 1.0 t7 2.00 0.0 0.0 0.0 1.0 t8 2.60 0.0 0.4 0.0 0.4 t9 3.00 0.6 0.4 0.6 0.0 t10 0.6 0.7 0.0 0.3 t11 0.0 0.7 0.6 0.0
5.00 10.00
TABLE 3: unit step relative to chosen time 0.50 141.5 51.7 0.40 0.25 0.10 68.4 15.1 0.00 0.0 0.0 0.00 0.0 0.0 0.00 0.0 0.0 0.00 0.0 0.0 0.00 0.0 0.0 0.00 0.0 0.0 0.00 0.0 0.0 1.4 0.00 0.0 0.0
TABLE 5: results at arbitrary times 120.5 0.00 0.01 0.05 0.10 0.25 0.26 0.26 0.27 0.29 25.0 59.6 90.0 112.2 143.6 116.5 110.0 101.0 89.2 100.7 25.0 30.0 39.1 47.4 37.1 46.3 50.3 56.9 67.4
TABLE 4: sum of terms T1_terms T2_terms 5.1 109.3 36.6 75.7
Tj1 =SUM(T2_terms)+ambient 180 Tj2 160 140 120 100 q1_pow er q2_pow er
{=IF(time>change_at,timechange_at,0)} {=SUM((S7*q1_Rs+S8*interact_Rs)* (1EXP(time/taus)))} 1.2
1
{=SUM((R7*interact_Rs+R8*q2_Rs)*(1EXP(time/taus)))}
{=TABLE(,time)}
0.8
Corporate R&D : Packaging Technology
=SUM(T1_terms)+ambient
perature [C]
201
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
80
Results from a 2input Excel Foster model
180.0 160.0 140.0 1.00 120.0 Tj1 Tj2 q1_pow er q2_pow er 1.20 1.40
temperature [C]
100.0 80.0 60.0
0.80
0.60
0.40 40.0 20.0 0.0 0 1 2 3 tim e [s] 4 5 6 0.20
0.00
202
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
power [W]
Organizing the sheet for transient solution
A cell for keeping track of the overall time progression of ALL blocks.
=IF(Master_Time>Row_Time,Master_timeRow_time,0)
Self heating column (each cell is a separate array formula) {=dPD#*SUM(R1:R10*(1EXP(dtime/Tau1:Tau10)))}
Interaction heated columns (each cell is a separate array formula) {=dPD#*SUM(R5:R10*(1EXP(dtime/Tau5:Tau10)))}
A section for power input to the heat sources
*Special thanks to Dave Billings for this slide
203 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
A section for Time changes
A section for power changes
A section for Temperature response calculation
Corporate R&D : Packaging Technology
Table for plotting temperature output
=SUM(D1:D1_by_D4) +T_ambient
Note!
Time in this column can be independent of the time values in the power input section
Next, Select this whole region Apply a Data > Table option
*Special thanks to Dave Billings for this slide
204 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Results from a 4input Excel Foster model
Temperature (C)
1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 0.05 0.1 0.15 Time (Sec) 0.2 0.05 0.1 0.15 Time (Sec) 0.2 0.05 0.05 0.1 0.15 Time (Sec) 0.2 D1
3 2.5 2 1.5 1 0.5 0 0 0.05 0.1 Time (Sec) 0.15 T_D1 T_D2 T_D3 T_D4
Power (W)
0.25 D2
0.2
0.25
0.3
Last power input
0.1 0.15 Time (Sec) 0.2
0.25 D3
Temperature (C)
3 2.5 2 1.5 1 0.5 0 0 0.05 0.1 Time (Sec) 0.15 0.2 0.25 0.3
Power (W)
0.25
Temperature (C)
3 2.5 2 1.5 1 0.5 0 0 0.05 0.1 Time (Sec) 0.15 0.2 0.25 0.3
Power (W)
0.25
Temperature (C)
D4
3 2.5 2 1.5 1 0.5 0 0 0.05 0.1
Power (W)
Time (Sec) 0.15
0.2
0.25
0.3
*Special thanks to Dave Billings for this slide
205 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Recap
• With the right tools, a thermal RC network can be generated from temperature data which is captured from measurements or finite element simulation. • Linear superposition provides a method for generating transient thermal models with several heat sources. • Foster networks can be used to simulate the thermal response of a system using a spreadsheet, whereas Cauer networks (which are closer to a physical lumped system) may require special tools (e.g. SPICE).
206
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Part III Thermal Runaway
207
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermal runaway
• Nonlinear power vs.. junction temperature device characteristic • System thermal resistance isn’t low enough to shed small perturbations
208
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
A linear thermal cooling system
TJ Q
Jx
Tx
junction temperature as function of power, theta, and ground … solving for power
Q
TJ
Tx
Jx
dQ dT
1
Jx
sensitivity (slope) of power with respect to temperature
209
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Effect of device line slope on system stability
system line power tendency to cool tendency to heat
Q
device line
Tx
TJ
junction temperature
210
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Operating points of thermal system when device line has negative second derivative
power the stable (that is, real) operating point an unachievable operating point system line tendency to cool device line
Q2
Q1
tendency to cool TJ1 Tx
power goes up with increasing temperature but rate of increase system falls with increase temperature tendency (negative second cannot be to heat system cannot derivative) maintained be successfully powered up
TJ2
junction temperature
211
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Operating points of thermal system when device line has positive second derivative
power an unachievable operating point even turning any the device on tendency perturbation destroyswill the stableit cause to cool (that is, real) runaway operating point tendency to heat system line
device line power goes up with increasing temperature, but rate of increase rises with increase (positive second derivative)
Q tendency to heat
Tx TJ
junction temperature
212
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Let’s see how it works
stable operating point device operating curve unstable operating point
2.0
1.6 Device Power Dissipation [W]
1.2
10°C/W system 25°C/W system 40°C/W system
213
0.8
NO operating point!
0.4
0.0 20 40 60 Junction Temperature [C] 80 100
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
A paradox
0.5 W Case A 100°C junction identical 50°C/W 75°C 50°C/W 75°C 0.5 W Case B 100°C junction
lead
100°C/W
lead
0.2°C/W
25°C
thermal ground
74.9°C
thermal ground
thermal runaway, based on Jx=150°C/W, calculated to be at 125°C
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
thermal runaway, based on Jx=50.2°C/W, calculated to be at 150°C
214
Corporate R&D : Packaging Technology
Paradox lost
raise the power by 0.1 W and see what happens
0.5 W0.1 +
0.5 W0.1 +
Case A
100 °C15 junction +
Case B
junction 100 °C5.02 + 50°C/W 75 °C0.02lead + 0.2°C/W
50°C/W 75 °C10 lead + 100°C/W
(fixed) 25°C
(fixed) 74.9°C
215
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Illustrating the paradox
Case B device line common nominal operating point 0.5 W
Case A
25°C
74.9°C 100°C
216
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Generic power law device and generic linear cooling system
power unstable operating point
Jx1
device line
system line A system line B 1
Jx1
Jx2
system line C
1
stable operating point
1
runaway point for original theta
runaway point for original thermal ground
Q
Tx
217
TJ Ty
TR2
TR1
junction temperature
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Don’t get confused by the terms!
a mathematical “power law” y ax an “exponential” power law (base is e)
device power
Q
V I
y
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
ex
218
Corporate R&D : Packaging Technology
Definition of power law device
rule of thumb for leakage; 2x increase for every 10°C
T 2 10
( ln 2 ) T 10 T 10 ln 2
for constant voltage, power does the same
T T
I
Io
Q
VRIo e
Qo e
I Io e
I
Io e
T
1st and 2nd derivatives
Io e
T1 T2 I ln 1 I2
defining:
dQ dT
Qo
T
e
d Q dT
2
2
Qo
2
T
e
both always positive
Corporate R&D : Packaging Technology
219
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
The mathematical essence
System line
Nondimensionalizing
Leads to:
(system)
Q
T Tx
Jx
q
kz
Tx JxQo
z
T Tx
temperature
where:
Power law device line
T
k
q 1 e Qo
Tx
e
Q
power
(power law device)
Q
Qo e
q ez
Eliminating q:
kz
ez
220
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Perfect runaway transformed
ez
at point of tangency, slope equals height
k=ez
k=ez
k=ez k=ez z0
1
zTz
z
0
T Tx
1 z 0
zT
1
z0
zT
1
zT
zT
221 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
z0
1
Corporate R&D : Packaging Technology
Transforming the nominal system
ez “operating” points nominal system line A k>e
(2 intersections)
k<e
(no intersections)
k=e 1
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
at point of tangency, slope equals height
222
Corporate R&D : Packaging Technology
Everything transformed
nondimensional power unstable, nonoperating point
device line
q=ez
system line A
q=k1z q=k1(zzx1)
system line B
1
stable operating point
k1
system line C
q=k2z
k2
e
runaway point for original theta
k2 1 zx1 zR2 1 zR1
k1
runaway point for original thermal ground
nondimensional temperature
z x1 ln(k1) 1 z R 2
223 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
1 zR1 ln(k1)
Corporate R&D : Packaging Technology
“Perfect runaway” results in original terms
runaway temperature based on original slope
runaway temperature based on original ambient
TR1
ln
Jx1Q o
TR2
Tx
max ambient that goes with it
system resistance that goes with it
Tx 1 Jx2
Tx1
ln
Jx1Q o
Qo
e
224
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
The “operating” points
ez “operating” points unstable
kzu e zu
kz
stable 1
zs
kzs
e zs
zu
225
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Newton’s method for the intersections
kz
zi
1
ez
z
zi
F(zi ) F (z i )
ln kz
zi
1
F(z) z ln kz
1 F ( z) 1 z
k ln z i e 1 1 zi
For k/e ranging between 1.01 and 1000, convergence is to a dozen significant digits in fewer than 10 iterations.
zo
1 k
1 this initial guess k converges to lower, e e stable point
this initial guess converges to upper, z o unstable point
Corporate R&D : Packaging Technology
ln k
1 ln
k e
226
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
And the intersection points come from …
find the nondimensional intersections first, then
Tstable
Tx
z stable
Tunstable
Tx
z unstable
227
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Real datasheet example
raw device data†
Vr [V]
Tmax [°C] Tref [°C] Itmax [A]
the device power curve parameters
@12V @40V
12
125 75 8.50E3
40
125 75 2.80E2
[°C]
17.9
9.4E5
17.8
1.02E3
Q o [W]
Itref [A]
5.20E4
1.70E3
T
I I0 e
Tmax Tref
Tmax Tref I ln max Iref
rule of thumb 10 gave us:
ln (2)
14.4
I0
It max e
Itref e
Q0
VRIo
† MBRS140T3
228
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Runaway analysis in nominal system
computed results
raw device data†
Vr [V] Tmax [°C] Tref [°C] Itmax [A] Itref [A] 12 125 75 8.50E3 5.20E4
Tx JxQo 1
@12V [°C] 17.9
9.4E5 10.6 117.2 135.1 1055 92.9
Jx1
@40V 17.8
1.02E3 0.97 74.4 92.2 96.6 92.8
@40V
40 125 75 2.80E2 1.70E3
Q o [W]
k (compare to unity) e Tx max [°C] given theta TR1[°C]
given ambient
Jx 2 max [°C/W]
1.609 83.5 101.3
k e
e
Tx
75
TR 2[°C]
100
† MBRS140T3
These translate into: a stable operating point at 80.6°C (and 0.09 W), an unstable point at 116.3°C (0.69 W)
Jx1
60
z z
0.312 2.315
229
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
HTRB example
• Bidirectional Thyristor in reliability stress test (High Temperature Reverse Bias) • Goal is life tests at elevated temperature (say 125°C) • Problem is, they don’t last very long, and if junction temperature is anything like the chamber temperature, they appear to fail way too early good!
*Special acknowledgements to Dave Billings and Geoff Garcia for their contributions to this project
230 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
MT1 + 640 V G MT2
40 kΩ
HTRB test circuit
HTRB DUTHTRB data  sockets without heatsinks test time tab temperature vs..
200
Heatsink (tab) temperature [degC]
180 160 140 120 100 80 60 40 20 0 0 2000 4000 6000 test time [s] 8000 10000 12000 DUT T1 DUT T2 DUT T3 DUT T4 DUT T5 DUT T6 DUT T7 DUT T8 DUT T9 DUT T10
231
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
HTRB DUT power vs.. test time
3 2.5 DUT W1 DUT W2 DUT W3 DUT W4 DUT W5 DUT W6 DUT W7 DUT W8 DUT W9 DUT W10
HTRB data  sockets without heatsinks
DUT Power [W]
2
1.5
1
0.5
0 0 2000 4000 6000 test time [s]
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
8000
10000
12000
232
Corporate R&D : Packaging Technology
HTRB example
233
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Quick calculations from datasheet
Pd I (640 40000 I )
TJ
or
MT1
Ta
THS
Pd
Pd
JA
+ 640 V 
G MT2
TJ
J HS
• At room temp, if IDRM is 5 uA, then Pd is about zero (≈3mW), and TJ should thus equal chamber set point. • At 85°C, IDRM is about 0.10.2mA, thus Pd is on the order of 0.1W, so depending on thetaJA, TJ could be several degrees hotter than chamber set point (note, however, that TJ will still be well within 1°C of heatsink temperature, THS) • HOWEVER, at 125°C, if IDRM is 2mA, then Pd will be in excess of 1W. Depending on thetaJA, TJ could be 3060°C above chamber set point (though still within a couple of degrees of heatsink temperature, if known).
234 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
40 kΩ
HTRB test circuit
Calculations based on actual measurements
Pd
I Vsense 1000
I (640 41000 I )
TJ
or
MT1
Ta
THS
Pd
Pd
JA
+ 640 V +
G MT2
TJ
J HS
• At room temp, IDRM (via Vsense) is 0.2uA, thus Pd is about zero (≈0.1mW), and TJ should thus equal chamber set point.
• At 85°C, IDRM is about 0.10.2mA, thus Pd is on the order of 0.1W, so depending on thetaJA, TJ could be several degrees hotter than chamber set point (note, however, that TJ will still be well within 1°C of heatsink temperature, THS) • At 125°C, IDRM is 23mA; Pd could be as high as 1.5W
40 kΩ
Vsense 
1 kΩ
• Max current observed was nearly 8mA (for Pd of 2.5W), and estimated TJ of 170°C just prior to device failure.
235 Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010 Corporate R&D : Packaging Technology
Modified HTRB test circuit
Actual “blocking current” data (time implicit)
1E2 blocking current when thetaJA=35°C/W (DUT's in unmodified HTRB board)
1E3 DUT I1 DUT I2 DUT I3 DUT I4 DUT I5 DUT I6 DUT I7 DUT I8 DUT I9 current
1E4
1E5
1E6
1E7 20 40 60 80 100 120 140 160
estimated junction temperature [°C]
236
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Actual “blocking Pd” data (time implicit)
1E+1 blocking Pd when thetaJA=35°C/W (DUT's in unmodified HTRB board)
1E+0
1E1
1E2
1E3
DUT W1 DUT W2 DUT W3 DUT W4 DUT W5 DUT W6 DUT W7 DUT W8 DUT W9 power
1E4 20 40 60 80 100 120 140 160
estimated junction temperature [°C]
237
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Power vs.. temperature (linear scales)
2.5 DUT W1 2 1.5 1 0.5 0 20 30 40 50 60 70 80 90 100 estimated junction temperature [°C] DUT W2 DUT W3 DUT W4 DUT W5 DUT W6 DUT W7 DUT W8 DUT W9
37°C/W system 10°C/ W system 4°C/W system
Pd [W]
110
120
130
238
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Proofofconcept modified HTRB fixture
239
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Pd vs.. temperature on better heatsinks
1E2 blocking current when thetaJA=10°C/W (using external 12°C/W heatsink)
1E3 DUT I1 DUT I2 DUT I3 DUT I4 DUT I5 DUT I6 DUT I7 DUT I8 DUT I9
1E4
1E5
1E6
1E7 20 40 60 80 100 120 140 160
estimated junction temperature [°C]
240
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
241
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
What if multiple devices on heatsink?
• Each device heats its neighbors to varying degrees, depending on distance • This adds background heat, that is, it raises the “effective ambient” of each device
242
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Graphically, background heat does this
2.0
Device Power Dissipation [W]
device operating curve
1.6
1.2
real runaway margin what you thought was your margin
0.8
25°C/W system
0.4
ij
Qi
60 Junction Temperature [C] 80
system with “background heating” of other devices
100
0.0 20
i j
40
243
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Math and Electrical References
1. M. Abramowitz, I. Stegun (eds), Handbook of Mathematical Functions, Dover Publications, Inc., 9th Printing, Dec. 1972 S.D. Senturia, B.D. Wedlock, Electronic Circuits and Applications, John Wiley & Sons, 1975 M.F. Gardner & J.L. Barnes, Transients in Linear Systems (Studied by the Laplace Transformation), Vol. I, John Wiley and Sons, 1942 R.S. Muller, T.I. Kamins, Device Electronics for Integrated Circuits, 2nd Ed., John Wiley & Sons, 1986 Ben Nobel, Applied Linear Algebra, Prentice Hall, 1969 H.H. Skilling, Electric Networks, John Wiley and Sons, 1974 L. Weinberg, Network Analysis and Synthesis, McGraw Hill Book Company, Inc., 1962 H. Wayland, Complex Variables Applied in Science and Engineering, Van Nostrand Reinhold Company
ThermalRelated Applications Notes available at http://www.onsemi.com/pub/Collateral/ANxxxxD.PDF
569: “Transient Thermal Resistance  General Data and its Use,” May 2003 1083: “Basic Thermal Management of Power Semiconductors,” October 2003 1570: “Basic Semiconductor Thermal Management,” January 2004 8044: “SingleChannel 1206A ChipFET™ Power MOSFET Recommended Pad Pattern and Thermal Performance,” December 2005 8072: “Thermal Analysis and Reliability of WIRE BONDED ECL,” April 2006 8080: “TSOP vs.. SC70 Leadless Package Thermal Performance,” January 2004 8199: “Thermal Stability of MOSFETs,” August 2005 8214: “General Thermal Transient RC Networks ,” April 2006 8215: “Semiconductor Package Thermal Characterization,” April 2006 8216: “Minimizing Scatter in Experimental Data Sets,” April 2006 8217: “What's Wrong with %Error in Junction Temperature,” April 2006 8218: “How to Extend a Thermal  RC  Network Model,” April 2006 8219: “How to Generate Square Wave, Constant Duty Cycle, Thermal Transient Response Curves,” April 2006 8220: “How To Use Thermal Data Found in Data Sheets,” April 2006 8221: “Thermal RC Ladder Networks,” April 2006 8222: “Predicting the Effect of Circuit Boards on Semiconductor Package Thermal Performance,” April 2006 8223: “Predicting Thermal Runaway,” April 2006 8402: “Thermal Considerations for the NCS5650”, June 2009 8432: “Thermal Consideration for a 4x4 mm QFN”, December 2009
2. 3.
4.
5. 6. 7. 8.
Thermal Textbooks & References
9. H.S. Carslaw & J.C. Jaeger, Conduction of Heat In Solids, Oxford Press, 1959 10. E.R.G. Eckert & R.M. Drake Jr., Heat and Mass Transfer, McGraw Hill, 1959 11. J.P. Holman, Heat Transfer, 3rd Ed., McGraw Hill, 1972 12. J. VanSant, Conduction Heat Transfer Solutions, Lawrence Livermore National Laboratory, Livermore, CA, 1980
244
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Thermal Test Standards
1. EIA/JEDEC Standard JESD512, Integrated Circuits Thermal Test Method Environmental Conditions  Natural Convection (Still Air), Electronic Industries Alliance, December 1995
2. EIA/JEDEC Standard JESD516, Integrated Circuit Thermal Test Method Environmental Conditions  Forced Convection (Moving Air), Electronic Industries Alliance, March 1999
3. EIA/JEDEC Standard JESD518, Integrated Circuit Thermal Test Method Environmental Conditions  JunctiontoBoard, Electronic Industries Alliance, October 1999 4. EIA/JEDEC Standard JESD5112, Guidelines for Reporting and Using Electronic Package Thermal Information  Electronic Industries Alliance, May 2005 5. JEDEC Standards No. 243, 244, 511, Electronic Industries Alliance, 1990 6. MILSTD883E, Method 1012.1, U.S. Department of Defense, 31 December 1996
245
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
Related Papers by Stout, et al
1. “TwoDimensional Axisymmetric ANSYS® Simulation for TwoParameter Thermal Models of Semiconductor Packages,” 7th International ANSYS Conference & Exhibition, May 1996, R.P. Stout & R.L. Coronado 2. “End User's Method for Estimating Junction Temperatures Due to Interactions of Other Dominant Heat Sources in Close Proximity to the Device in Question,” ITHERM, May 1996, D.T. Billings & R.P. Stout 3. “Evaluation of Isothermal and Isoflux Natural Convection Coefficient Correlations for Utilization in Electronic Package Level Thermal Analysis,” 13th Annual IEEE Semiconductor Thermal Measurement and Management Symposium, January 1997, B.A. Zahn and R.P. Stout 4. "Electrical Package Thermal Response Prediction to Power Surge", ITHERM, May 2000, Y.L. Xu, R.P. Stout, D.T. Billings 5. “Accuracy and Time Resolution in Thermal Transient Finite Element Analysis,” ANSYS 2002 Conference & Exhibition, April 2002, R.P. Stout & D.T. Billings 6. “Minimizing Scatter in Experimental Data Sets,” ITHERM 2002 (Eighth Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems), June 2002, R.P. Stout 7. “Combining Experiment and FEA into One, in Device Characterization,” ITHERM 2002 (Eighth Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems), June 2002, R.P. Stout, D.T. Billings 8. “A TwoPort Analytical Board Model,” ITHERM 2002 (Eighth Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems), June 2002, R.P. Stout 9. “On the Treatment of Circuit Boards as Thermal TwoPorts,” InterPack2003, July 2003, R.P. Stout 10.“A Conjugate NumericalRC Network Prediction of the Transient Thermal Response of a Power Amplifier Module in Handheld Telecommunication,” InterPACK 2005, July 2005, T.Y. Lee, V.A. Chiriac, R.P. Stout 11.“Part 1:Linear Superposition Speeds Thermal Modeling,” Power Electronics Technology, January 2007, R.P. Stout 12.“Part 2:Linear Superposition Speeds Thermal Modeling,” Power Electronics Technology, February 2007, R.P. Stout 13.“Power Electronics System Thermal Design: Linear Superposition,” IEEE Expert Now online tutorial <http://ieeexplore.ieee.org/articleSale/modulesabstract.jsp?mdnu mber=EW1054>, February 2007, Roger Stout 14.“Power Electronics System Thermal Design: Thermal Runaway,” IEEE Expert Now online tutorial <http://ieeexplore.ieee.org/articleSale/modulesabstract.jsp?mdnu mber=EW1055>, February 2007, Roger Stout 15.“Using Linear Superposition to Understand the True Meaning of ThetaJA,” 12th Annual Automotive Electronics Council Reliability Workshop, May 2007, R.P. Stout 16.“Using Linear Superposition to Solve Multiple Heat Source Transient Thermal Problems,” InterPACK 2007, July 2007, D.T. Billings, R.P. Stout 17.“Linear Superposition and the True Meaning of ThetaJA,” 1hr Seminar at 2007 Power Electronics Technology Exhibition and Conference, October 2007, R.P. Stout 18.“Thermal Performance of a Monolithic ThinShell Concrete Dome,” ASME Heat Transfer Conference, July 2007, R.P. Stout 19.“Beyond the Datasheet: Demystifying Thermal Runaway,” Power Electronics Technology, November 2007, R.P. Stout 20.“The Datasheet is Not Your Mother” (Executive Viewpoint article for) Power Electronics Technology, February 2008, R.P. Stout 21.“Psi or Theta: Which One Should You Choose?” Power Electronics Technology, March 2008, R.P. Stout 22.“Don’t Be Misled By Power Device Specs” Power Electronics Technology, May 2008, R.P. Stout 23.“Reliability of NLDMOS Transistors Subjected to Repetitive Power Pulses,” IEEE International Reliability Physics Symposium, April 2008, Poster Session paper, Chris Kendrick, Roger Stout, and Michael Cook
246
Semiconductor Device Thermal Characterization and System Analysis (RPS) April 2010
Corporate R&D : Packaging Technology
This action might not be possible to undo. Are you sure you want to continue?