Too Hot to Hold: Determining the Cooling Limits for

Transcription

Too Hot to Hold: Determining the
Cooling Limits for Handheld Devices
Guy R. Wagner
Electronic Cooling Solutions, Inc.
2915 Copper Road
Santa Clara, California, 95051
USA
gwagner@ecooling.com
William Maltz
Electronic Cooling Solutions, Inc.
2915 Copper Road
Santa Clara, California, 95051
USA
wmaltz@ecooling.com
Advancements in Thermal Management 2013
June 6-7, 2013 Denver, CO
Outline of Presentation
–
–
–
–
Handheld Device Background
Abstract
Surface Temperature Considerations
Tablet Construction
• Teardown of a Tablet Cooled by Forced Convection
• Teardown of a Tablet Cooled by Natural Convection
– Experimental Testing
• IR Imaging
• Thermocouple Measurements
• Mentor Graphics T3ster Measurements
– Experimental Results Compared to CFD Modeling
– Methods of Improving Natural Convection Power Dissipation
• Heat Spreaders
• Air Gaps
– Summary
– Conclusions
– References
2
Handheld Device Background
Handheld devices are increasingly capable of running applications that used to
require laptop and desktop computers. The requirement that these devices
provide simular performance with a smaller form factor presents significant
challenges, especially when one considers that passive cooling is almost a
requirement.
Thermal design of next generation handheld tablet devices will need to address
both a comfortable surface touch temperature and maximum temperature
limitations of internal critical components while also meeting aggressive
industrial design requirements.
3
Abstract
This study performs an analysis on thermal management techniques deployed
in tablets that exist in the market. This analysis is done by performing teardowns
and experimental measurements of several popular tablets available in the
market. The overall power dissipation is measured under various exercising
conditions.
IR measurements provide insight into the maximum hot spot temperatures.
Preliminary studies involving simplified representations of tablets using
simulations showing effects of orientation and ambient temperature will also be
presented.
The maximum possible heat transfer dissipation under ideal conditions is
calculated. Several thermal solutions are proposed and analyzed in order to
achieve higher heat transfer dissipation, while eliminating the hot spots, in order
to achieve uniform skin temperature.
The ergonomic temperature limit for handheld devices are also discussed.
4
Surface Temperature of a
Vertical Isothermal Tablet
Surface Temperature of a Horizontal
Isothermal Tablet on an Adiabatic Surface
Surface Temperature
Considerations
In a 25°C ambient condition, the maximum
total power dissipation is calculated with a
requirement that the surface temperature
does not exceed a touch temperature of
41°C. This is the maximum aluminum
enclosure comfort touch temperature as
presented by Berhe (2007). Use of low
conductivity case materials has the effect of
increasing the maximum comfortable touch
temperature by about 5C.
It can be seen that the theoretical
maximum total power dissipation is 13.9
watts when the device is suspended in
midair with heat transfer occurring at front
and back surfaces. When the device is
placed on a horizontal adiabatic surface,
heat transfer is occurring at the front screen
surface only and the maximum power
dissipation is reduced to 7.9 watts.
These calculations assume perfect heat
spreading. In actual practice, there will be
hot spots on the device which have the
effect of lowering the maximum allowable
power.
7
Tablet Construction
8
Teardown of a Tablet Cooled
By Forced Convection
Battery
EMI shield
Heat pipe over CPU
and graphics chips
Heat exchanger
Blower
Heat pipe
Heat pipe detail with EMI shield
removed
SSD
Tablet with back cover removed
DC-DC converter
9
Teardown of a Tablet Cooled
By Natural Convection
Speaker
Batteries
Camera
Cabling
The touch screen display has been removed
Main PCB with EMI Shields over ICs
10
Instrumentation of the Tablet’s
Internal Components
11
Experimental Testing
12
Thermocouple and IR
Measurements
13
Forced Convection Tablet
IR Imaging – Screen Surface
•
•
Results shown for screen
surface temperature while
playing video
Hottest surface temperature is
35oC over heat exchanger
exhaust port
14
Natural Convection Tablet
IR Imaging – Screen Surface
•
•
•
•
Results shown for screen
surface temperature while
playing demo of graphicintensive game
Hottest surface temperature is
45oC
Thermocouples were attached
at the corresponding IR test
points and were used to
determine the emissivity of the
surface.
External surface thermal test
results before disassembling
the device compared to the
results after re-assembling the
device within 0.8C
15
Thermocouple Measurements
Date
Tablet Orientation
Exerciser
Component
Test-01
3/20/11
Screen facing upward
(15 mm from table)
Windows Idle
Test-02
3/20/11
Screen at 30 degrees angle
Windows Idle
Test-01
Test-02
Test-03
3/20/11
Screen facing upward
(15 mm from table)
Windows with Youtube Video
(in Continuous Loop)
Test-03
Test-04
3/20/11
Screen at 30 degrees angle
Windows with Youtube Video
(in continuous loop)
Test-04
Measured
DT
Measured
DT
Measured
DT
Measured
DT
Ambient
1
2
3
4
5
6
7
8
9
9a
9b
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Battery-1
Battery-2
EMI_CAN
Exhaust_Air
Fan_Air_Inlet
HP-1
HP-2
LCD-1
LCD-2
MEM-1
MEM-2
PCB-1
PCB-2
SSD
Inlet_Ambient
21.3
32.6
29.8
28.1
31.7
30.2
30.2
24.5
24.3
26.9
24.4
30.0
28.1
31.1
25.2
30.2
25.3
24.5
31.0
32.7
32.0
28.0
25.4
29.1
30.5
26.0
27.2
22.8
23.6
30.0
24.8
38.1
29.5
31.8
39.9
38.3
31.6
26.8
35.9
36.5
33.7
32.6
33.7
25.3
0.0
11.3
8.5
6.8
10.4
8.9
8.9
3.2
3.0
5.6
3.1
8.7
6.8
9.8
3.9
8.9
4.0
3.2
9.7
11.4
10.7
6.7
4.1
7.8
9.2
4.7
5.9
1.5
2.3
8.7
3.5
16.8
8.2
10.5
18.6
17.0
10.3
5.5
14.6
15.2
12.4
11.3
12.4
4.0
21.2
31.6
27.7
27.4
30.3
28.9
28.7
24.2
23.9
26.7
23.9
27.6
28.4
30.6
25.2
29.7
25.2
24.8
30.4
33.0
32.6
28.1
25.4
29.0
30.6
26.2
27.3
22.8
23.4
29.1
24.4
38.0
30.2
30.2
39.8
38.0
31.2
26.7
35.6
36.3
33.4
32.4
33.2
25.2
0.0
10.4
6.5
6.2
9.1
7.7
7.5
3.0
2.7
5.5
2.7
6.4
7.2
9.4
4.0
8.5
4.0
3.6
9.2
11.8
11.4
6.9
4.2
7.8
9.4
5.0
6.1
1.6
2.2
7.9
3.2
16.8
9.0
9.0
18.6
16.8
10.0
5.5
14.4
15.1
12.2
11.2
12.0
4.0
17.0
35.4
29.2
24.3
31.4
29.2
29.1
23.7
23.7
24.9
23.8
32.2
23.7
29.0
23.3
29.2
24.8
24.2
32.0
33.7
34.5
27.9
24.8
25.6
26.8
22.6
24.8
22.4
23.1
28.3
24.0
38.4
31.9
32.5
40.5
39.9
30.6
26.2
40.7
41.0
32.5
32.8
29.5
21.0
0.0
18.4
12.2
7.3
14.4
12.2
12.1
6.7
6.7
7.9
6.8
15.2
6.7
12.0
6.3
12.2
7.8
7.2
15.0
16.7
17.5
10.9
7.8
8.6
9.8
5.6
7.8
5.4
6.1
11.3
7.0
21.4
14.9
15.5
23.5
22.9
13.6
9.2
23.7
24.0
15.5
15.8
12.5
4.0
16.8
34.0
28.0
23.9
30.1
28.3
28.0
23.3
23.2
24.9
23.2
30.4
23.6
28.6
23.5
28.9
24.5
24.0
31.3
33.3
33.7
27.9
24.6
25.3
26.7
22.5
24.6
22.1
22.8
27.7
23.5
38.0
30.9
31.5
39.7
39.1
30.3
25.9
40.6
40.9
32.0
32.6
28.9
20.8
0.0
17.2
11.2
7.1
13.3
11.5
11.2
6.5
6.4
8.1
6.4
13.6
6.8
11.8
6.7
12.1
7.7
7.2
14.5
16.5
16.9
11.1
7.8
8.5
9.9
5.7
7.8
5.3
6.0
10.9
6.7
21.2
14.1
14.7
22.9
22.3
13.5
9.1
23.8
24.1
15.2
15.8
12.1
4.0
CPU
ACTUAL CPU TEMP
57.0
47.0
33.0
59
49
25.7
AMD THERMNOW READING
56
46
24.8
16
60
50
32.2
Thermocouple Measurements
TC #
Idle
Despicable Me
Video
Riptide GP
Game
GL Benchmark
Exerciser
i1: A5X CPU
29.2
53.4
75.2
66.1
i2: Apple IC
29.8
58.3
71.8
66.4
i3: NAND Flash
27.8
44.5
54.3
50.2
i4: under WiFi
27.8
44.6
51.0
48.1
i5: Voltage Reg
28.4
50.3
58.7
55.3
i6: Logic temp NAND
27.6
42.1
48.7
45.9
i7: power connector
27.1
39.0
42.8
41.2
i8: WiFi/BlueTooth
27.9
45.2
51.9
49.0
i9: under Voltage
29.3
56.5
67.9
63.4
i10: under A5X
29.2
53.0
74.6
65.6
i11: RAM
28.7
50.1
67.3
60.3
i12: under NAND
27.7
43.2
52.0
48.3
i13: under Display
27.0
39.6
44.0
42.1
i14: case temp
27.1
39.7
44.4
42.4
i15: air temp
26.7
36.8
39.3
38.1
i16: batt left
26.7
37.5
40.9
39.4
i17: batt right
26.4
35.6
37.7
36.7
17
Thermal Characterization of CPU using
Mentor Graphics T3ster
18
What is T3ster?
When building a thermal model of a tablet, the thermal characteristics of the
processor are not always known with a high degree of accuracy. It is also true
that data sheets from the suppliers of thermal interface materials may not
accurately reflect the thermal resistance of the interface material and the
wetting properties of the material between the processor chip or lid and the
heat spreader.
To overcome this limitation and get an accurate thermal model of the processor,
a Mentor Graphics T3Ster® was used to determine the thermal resistance from
the processor IC to the lid or heat spreader and the PCB. The T3Ster® is able
to do a dynamic thermal characterization of the thermal resistance paths of a
packaged semiconductor device.
The transient temperature response of the die is recorded as a function of a
step input in power to the die and a structure function is derived from the
transient temperature response that characterizes the thermal resistance of all
the materials in the thermal path. The following slide shows the structure
function that was derived for a processor using the T3Ster®. Note that the
thermal resistance from junction to case is measured at 0.23 K/W using this
technique.
19
T3Ster Results
T3Ster Master: cumulative structure function(s)
10000
Fluid
Rth-JC = 0.23K/W
Copper
Cth [Ws/K]
100
TIM2*
Lid
1
Die-attach*
0.01
Silicon
Specific Heat=
Density=
Volume=
Cth=
Die
1e-4
0
0.1
0.2
0.3
0.4
0.5
Rth [K/W]
0.6
0.7 J/g*K
2.65 g·cm−3
0.1
4.90E-02 cm^3
0.090960 J/K
0.7
0.8
0.9
*Note: includes thermal contact resistance.
The package thermal resistance of 0.23K/W can now be put into the
FloTHERM model to yield accurate junction temperatures
20
Experimental Results Compared to
CFD Modeling
21
IR Imaging vs CFD Simulation
35.5
29.6
33.1
31.9
IR Camera Image, Emissivity = 0.94
Numerical Simulation
22
Internal Surface Temperatures
23
IR Imaging vs FloTHERM Simulation
IR Camera Image, Emissivity = 0.90
Numerical Simulation
24
Methods of Improving Natural
Convection Power Dissipation
25
Use of Heat Spreaders
Heat spreaders may be either internal or part of the case structure.
Through the use of high-conductivity heat spreaders, the maximum hot
spot temperature is reduced. Due to reducing hot spot temperature,
the average case temperature may be raised allowing increased power
dissipation while not exceeding the maximum surface temperature
requirement.
The following screen captures show the effect of increasing the thermal
conductivity of a 0.8 mm thick case from 0.2 W/mK (plastic) to 200
W/mK (aluminum).
This study assumes an internal power dissipation of 8.9 watts
26
Temperatures of Back Surface of Case
K = 0.2 W/mK
K = 2.0 W/mK
Temperatures of Back Surface of Case
K = 20 W/mK
K = 200 W/mK
Effect of Case Thermal Conductivity
k
Hot Spot
Case Center
CPU
W/mK
C
C
C
0.2
81.3
34.6
99.1
2
69.9
35.5
87.9
20
51.2
35.7
73.3
200
39.9
34.9
64.2
29
Effect of Case Thermal Conductivity
30
Use of Air Gaps
21C Ambient Air Temperature
Video Mode, Vertical Orientation, 1.2 mm thick Al heat spreader
Power Dissipation vs Air Gap Thickness
Air Gap Effect on Temperatures
Video Mode, Vertical Orientation
5.2
32
5.1
Back Power Dissipation (W)
Temperature Rise (C)
28
26
24
Hot Spot (C)
22
APU Core (C)
20
18
16
14
Power Dissipation (W)
30
Front Power Dissipation (W)
5
4.9
4.8
4.7
4.6
12
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
Air Gap Thickness between Heat Spreader and Rear Cover (mm)
4.5
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Air Gap Thickness between Heat Spreader and Rear Cover (mm)
Heat is transferred off all surfaces of the tablet by both convection and
radiation. The simulation results shown above take both modes of heat
transfer into account.
31
Summary
The maximum power dissipation of the internal components is not only governed by the size of
the tablet but is a strong function of how well that heat is spread internally to reduce hot-spot
temperatures. Few engineers realize the importance played by radiation in dissipating the heat
from the exposed surfaces of a tablet. It is not until precise calculations are made that the
importance of radiation is realized in the thermal design of the tablet. If the emissivities of the
various surfaces are high, over half of the heat transfer to the surroundings is due to radiation.
Overall heat transfer is maximized by reducing hot spot temperatures and spreading the heat
so that all surfaces are effectively providing maximum heat transfer through convection and
radiation.
In summary, building an accurate thermal model of the tablet allows the designer to rapidly test
the effect of design and material changes without incurring the high cost and schedule delays
of testing prototypes. A thermal model allows the thermal design engineer to investigate far
more alternatives than building prototypes. This results in a highly engineered tablet design that
better meets the expectations of the user while providing a competitive edge over the
competition. High quality thermal models speed time to market and lower development costs.
With the accuracy of the latest simulation software, the intermediate step of building and testing
thermal prototypes can be reduced or eliminated. The only need is final thermal verification of
production prototype samples.
32
Conclusions
Forced Convection
 The forced convection allows for about 50% more power than natural convection.
 Forced convection tablet suffers from audible noise from the small blower.
 Run time is decreased due to increased power consumption.
 The tablet may be placed on an insulating surface without overheating.
Natural Convection
 The natural convection tablet has the advantage of being perfectly silent.
 Run time is increased since there is no blower to consume additional power.
 It has the disadvantage of overheating when the unit is placed on a surface with low
thermal conductivity such as a blanket or pillow.
 Radiation accounts for about half of the heat dissipation.
 The back case touch temperature can be increased by about 5C if the material used for
the back case has a low thermal conductivity such as plastic.
 An internal heat spreader or a high conductivity case results in hot spot as well as
component temperature reduction.
 An air gap between the internal heat spreader and the case also results in hot spot
temperature reduction.
33
References
REFERENCES
[1]
Berhe, M.K., Ergonomic Temperature Limits for Handheld Electronic Devices,
Proceedings of ASME InterPACK’07, Paper No. IPACK2007-33873
[2]
Brown, L., Seshadri, H., Cool Hand Linux® - Handheld Thermal Extensions,
Proceedings of the Linux Symposium, Vol. 1, pp 75 – 80, 2007
[3]
Gurrum, S.P., Edwards, D.R., Marchand-Golder, T., Akiyama, J., Yokoya, S.,
Drouard, J.F., Dahan, F., Generic Thermal Analysis for Phone and Tablet Systems,
Proceedings of IEEE Electronic Components and Technology Conference, 2012
[4]
Huh, Y., Future Direction of Power Management in Mobile Devices, IEEE Asian
Solid-State Circuits Conference, 2011.
[5]
Lee, J., Gerlach, D.W., Joshi, Y.K., Parametric Thermal Modeling of Heat
Transfer in Handheld Electronic Devices, Proceedings of the 11th IEEE Intersociety
Conference on Thermal and Thermomechanical Phenomena in Electronic Systems, ITHERM, pp 604-609, 2008
[6]
Mongia, R., Bhattacharya, A., Pokharna, H., Skin Cooling and Other Challenges
in Future Mobile Form Factor Computing Devices, Microelectronics Journal, Vol. 39, pp
992 – 1000, 2008
34

Too Hot to Hold: Determining the Cooling Limits for

Transcription

Similar documents