optimizing pcb thermal performance for cree xlamp xq & …figure 2 shows simulated thermal...

TECHNICAL ARTICLE

Copyright © 2013-2020 Cree, Inc. All rights reserved. The information in this document is subject to change without notice. Cree®, the Cree logo and XLamp® are registered trademarks of Cree, Inc. This document is provided for informational purposes only and is not a warranty or a specification. For product specifications, please see the data sheets available at www.cree.com. For warranty information, please contact Cree Sales at [email protected].

Cree, Inc.4600 Silicon Drive

Durham, NC 27703USA Tel: +1.919.313.5300

WW

W.C

REE.

CO

M/X

LAM

PC

LD-A

P167 R

EV 0G

Optimizing PCB Thermal Performancefor Cree® XLamp® XQ & XH Family LEDs

INTRODUCTION

This application note outlines how to design a thermally

effective printed circuit board (PCB) to use with the XQ and XH

families of Cree XLamp® LEDs.

One of the most critical design parameters for an LED-based

illumination system is its ability to conduct heat away from

the LED junction. High operating temperatures at the LED

junction adversely affect the performance of LEDs, resulting in

decreased light output and lifetime.1 Specific practices should

be followed in the design, assembly and operation of LEDs in

lighting applications to properly manage this heat.

This application note provides guidelines for designing a PCB

layout that optimizes heat transfer from XQ and XH family LEDs.

Guidelines are provided for FR-4 and metal-core printed-circuit

boards (MCPCB). Cree encourages its customers to consider

these guidelines when evaluating the many LED thermal

management techniques available. For additional guidelines on

LED thermal management, refer to the Thermal Management

application note.

1 See the Long-Term Lumen Maintenance application note

TABLE OF CONTENTS

Introduction ...................................................................................1

Background ...................................................................................2

Thermal Simulation .......................................................................2

Thermal Spreading ........................................................................3

Thermal Crosstalk .........................................................................3

Copper Trace Size .........................................................................5

Usefulness of Thermal Vias .........................................................6

FR-4 Board Thickness ...................................................................8

FR-4 Trace Thickness ....................................................................8

MCPCB Trace Thickness ..............................................................9

MCPCB Dielectric Thermal Conductivity .....................................9

FR-4 vs. MCPCB ..........................................................................10

Measured vs. Simulated Results ................................................10

Trace Size Recommendations ...................................................11

FR-4 PCB with XQ Family LED .............................................11

MCPCB with XQ Family LED ...............................................12

FR-4 with XH Family LED .....................................................12

MCPCB with XH Family LED ...............................................12

Chemical Compatibility...............................................................12

http://www.cree.com/Xlamp

http://www.cree.com/~/media/Files/Cree/LED%20Components%20and%20Modules/XLamp/XLamp%20Application%20Notes/XLampThermalManagement.pdf

http://www.cree.com/~/media/Files/Cree/LED%20Components%20and%20Modules/XLamp/XLamp%20Application%20Notes/XLampThermalManagement.pdf

http://www.cree.com/~/media/Files/Cree/LED%20Components%20and%20Modules/XLamp/XLamp%20Application%20Notes/XLamp_lumen_maintenance.pdf


2

OPTIMIZING XQ & XH PCB THERMAL PERFORMANCE

BACKGROUND

Prior to the introduction of the XQ and XH families of LEDs, Cree XLamp LEDs were designed with an electrically isolated thermal path.

The neutral thermal pad simplifies the design of PCBs for thermal considerations. The pad provides a path for heat transfer away from the

LED chip junction to the thermal pad. Being electrically isolated from the anode and cathode of the LED means the pad can be soldered

or attached directly to the PCB or heat sink. With a neutral thermal pad, thermal vias can be placed directly under the thermal pad, but

because they have electrically conductive pads, this cannot be done for XQ and XH family LEDs. Figure 1 shows this difference.

Figure 1: LEDs with electrically isolated (top) and non-isolated (bottom) thermal pads mounted on a circuit board

THERMAL SIMULATION

The data presented here are the results of simulations conducted using ANSYS DesignSpace2 and Autodesk Simulation CFD3 software.

A standard off-the-shelf heat sink4 with a pre-attached thermal interface material (TIM) was used for all the simulations, with convection

set to 7 W/m2 °C and 1 W of total input for all scenarios.

Notes:

• Actual results may vary with different geometries and materials. Cree recommends performing actual verification testing to validate

the thermal management of any LED-based illumination system.

2 ANSYS, Inc., www.ansys.com/Products/Simulation+Technology/Structural+Mechanics/ANSYS+DesignSpace3 Autodesk Inc., www.autodesk.com/products/autodesk-simulation-family/features/simulation-cfd4 Model 374424B00035G, Aavid Thermalloy, www.aavid.com/products/bga/374424b00035g

3

+ -

Copper Plane

Copper Thermal Vias

Isolated

thermal pad

+ -

Copper Plane

Copper Thermal Vias

Non-isolated

Shorting together

http://www.ansys.com/Products/Simulation+Technology/Structural+Mechanics/ANSYS+DesignSpace

http://www.autodesk.com/products/autodesk-simulation-family/features/simulation-cfd

http://www.aavid.com/products/bga/374424b00035g


3


• Modeling and simulation are attempts to predict future performance based on assumptions and criteria that may differ from actual

device use and environment and slight modifications must sometimes be made to the design to enable the modeling and simulation

process. Actual results may differ from the modeling and simulation results due to these modifications and actual device use and

environment.

THERMAL SPREADING

We simulated thermal spreading to determine if it is proportional to the size of the contact pads of the XQ and XH family LEDs. If the

thermal spreading is not proportional, the traces for each contact pad must be sized to account for the disproportionality.

Figure 2 shows simulated thermal spreading for a single XQ-D LED on a 3.3 mm X 3.3 mm trace and a single XH-G LED on a 5.0 mm X 5.0 mm

trace, both mounted on an MCPCB with 2-oz thick copper and a dielectric thermal conductivity of 2.2 W/mK. The XQ-D LED contact pads

are of equal size and the proportion of simulated thermal flux conducted through each pad is within 0.3% of being equal. There is a 68.3%

to 31.7% size difference in the contact pads of the XH-G LED and the proportion of thermal flux conducted through each is within 3.7% of

being the same (64.6% to 35.4%). In each case, the proportion of thermal flux conducted through each contact pad is in direct correlation

with the relative sizes of the pads. Therefore it is logical to scale the sizes of the traces to be in proportion to the sizes of the contact pads.

Figure 2: Thermal spreading simulation for XQ-D (left) and XH-G (right) LEDs

THERMAL CROSSTALK

Thermal crosstalk occurs when the heat output from one LED affects an adjacent LED. We compared the thermal resistance, junction to

ambient temperature (Θj-a) of the center LED in a 3 X 3 pattern of nine LEDs mounted on a 50 mm X 50 mm PCB on an arbitrary sink with

a 3.3 mm X 3.3 mm trace to determine how closely together XQ family LEDs can be located without incurring a large thermal crosstalk

penalty. For XH family LEDs, we used a 5 mm X 5 mm trace.

Thermal spreading from XQ

• Single XQ-D on small trace on MCPCB

Copyright © 2013, Cree Inc. 4

49.7% of total heat flux through this pad

50.3% of total heat flux through this pad

Objective: Show how much heat flows through each contact pad Conclusion: Basically even thermal flux to both contact pads.

35.4% of total heat flux through small contact pad

64.6% of total heat flux through large contact pad

50.3% of total heat flux through this contact pad

49.7% of total heat flux through this contact pad

Thermal spreading from XH

• Single XH-G on small trace on MCPCB


64.6% of total heat flux through large contact pad

35.4% of total heat flux through small contact pad

Objective: Show how much heat flows through each contact pad Conclusion: More heat flows through larger pad, roughly proportional to the size differential Area: 4.23mm2 for large, 1.96mm2 for small (68.3/31.7%)


4


Figure 3 shows the inter-LED spacing measurements.

• 9x XQ-D on a 50x50mm PCB on an arbitrary heat sink with a small trace

• Comparison of the j-a of the center LED for each condition

XQ Thermal cross-talk


10mm 7mm 5mm 3mm 2mm

Objective: See how close we can pack XQ LEDs to each other without a huge penalty in thermal cross-talk Conclusion: Very minimal ‘cross-talk’ for MCPCB. More, but still small for FR4 even as close as 2mm

90.0%

95.0%

100.0%

105.0%

110.0%

0 2 4 6 8 10 12

%

_j-a

of

10m

m sp

acin

g

Distance between LEDs (mm)

MCPCB

FR4

• 9x XH-D on a 50x50mm PCB on an arbitrary heat sink with a 5mm trace

• Comparison of the j-a of the center LED for each condition

XH Thermal cross-talk


10mm 7mm 5mm 3mm 2.3mm

Objective: See how close we can pack XH LEDs to each other without a huge penalty in thermal cross-talk Conclusion: Very minimal ‘cross-talk’ for MCPCB. More, but still small for FR4 even as close as 2.3mm (which is about as close as I can do w/o traces touching)

90.0%

95.0%

100.0%

105.0%

110.0%

0 2 4 6 8 10 12

%

_j-a

of

10m

m sp

acin

g


MCPCB

FR4

Figure 3: Inter-LED spacing for thermal crosstalk simulation for XQ-D (left) and XH-G (right) LEDs

Figure 4 and Figure 5 show thermal simulation images for various inter-LED spacings, defined as the gap between the LEDs.

2-mm spacing 3-mm spacing 5-mm spacing 7-mm spacing 10-mm spacing

Figure 4: Simulation of XQ-D thermal crosstalk

2.3-mm spacing 3-mm spacing 5-mm spacing 7-mm spacing 10-mm spacing

Figure 5: Simulation of XH-G thermal crosstalk


5


Chart 1 and Chart 2 show the relative Θj-a of various inter-LED spacings compared to a 10-mm spacing, the largest spacing that was

simulated and where minimal cross-talk was observed. There was minimal thermal crosstalk for the LEDs mounted on an MCPCB. There

was more thermal crosstalk for the LEDs mounted on an FR-4 PCB, however the amount of crosstalk was small even for the smallest

spacings.XQ thermal cross-talk XQ_Thermal_AppNote_Data-mod.xlsx

90%

95%

100%

105%

110%

0 2 4 6 8 10 12

Qj-a

(°C/

W) o

f 10-

mm

spa

cing


MCPCB

FR4

XH thermal cross-talk XQ_Thermal_AppNote_Data_v2.xlsx

90%

95%

100%

105%

110%

0 2 4 6 8 10 12 Q

j-a (°

C/W

) of 1

0-m

m s

paci

ng


MCPCB

FR4

Chart 1: Thermal crosstalk simulation results for XQ family LEDs Chart 2: Thermal crosstalk simulation results for XH family LEDs

COPPER TRACE SIZE

We investigated the copper trace area needed below XQ and XH family LEDs to optimize thermal flow by comparing the Θj-a of a single

XQ-D and XH-G LED with varying copper trace sizes. The trace size was defined as the area of the trace, kept square and centered on the

board. Figure 6 and Figure 7 show images of the LEDs mounted on the various traces and the thermal simulation images.

3.3 mm X 3.3 mm 5.5 mm X 5.5 mm 7 mm X 7 mm 10 mm X 10 mm 15 mm X 15 mm

Figure 6: XQ-D LED mounted on various-size traces (top) and thermal simulation of XQ-D LED mounted on same various-size traces (bottom)


6


3.3 mm X 3.3 mm 5 mm X 5 mm 10 mm X 10 mm 15 mm X 15 mm 25 mm X 25 mm

Figure 7: XH-G LED mounted on various-size traces (top) and thermal simulation of XH-G LED mounted on same various-size traces (bottom)

Chart 3 and Chart 4 show a comparison of the relative Θj-a of a single LED for each copper trace size on an MCPCB and an FR-4 PCB. There

is a minimal difference in Θj-a for the various trace sizes on the MCPCB. There is a significant increase in Θj-a for small trace sizes and a

significant decrease in Θj-a for larger trace sizes on the FR-4 PCB. The Θj-a difference is due to the much better thermal conduction of the

MCPCB compared to the lesser thermal conduction of the FR-4 board.XQ copper trace size XQ_Thermal_AppNote_Data-mod.xlsx

0%

50%

100%

150%

200%

250%

300%

0 5 10 15 20 25 30

Qj-a

(°C/

W) o

f 15-

mm

trac

e

Copper trace size (mm)

MCPCB FR4

XH copper trace size XQ_Thermal_AppNote_Data-mod.xlsx

0%

50%

100%

150%

200%

250%

300%

0 5 10 15 20 25 30

qj-a

(°C/

W) o

f 25-

mm

trac

e

Copper trace size (mm)

MCPCB FR4

Chart 3: Copper trace size comparison for XQ family LEDs Chart 4: Copper trace size comparison for XH family LEDs

USEFULNESS OF THERMAL VIAS

An inexpensive way to improve thermal transfer for FR-4 PCBs is to add thermal vias - typically plated through-holes (PTH) between

conductive layers. Vias are created by drilling holes and copper plating them. Because the XQ and XH family LEDs do not have an

electrically isolated thermal pad, it is not possible to locate thermal vias directly underneath the LEDs, so we simulated locating thermal


7


vias outside the trace. Figure 8 shows an XQ and XH family LED mounted on traces of three sizes with thermal vias around the outside

of the trace.

3.3 mm X 3.3 mm trace 10 mm X 10 mm trace 15 mm X 15 mm trace

Figure 8: Thermal vias around the trace of XQ-D (top) and XH-G (bottom) LEDs

Chart 5 and Chart 6 show the Θj-a of plated and filled vias around various trace sizes compared to the Θj-a with no vias. Thermal vias can

be helpful with smaller trace sizes, but are of decreasing help as trace size increases. We added a second row of traces for some of the

scenarios, but the second row produced no difference in the Θj-a.


8


XQ thermal vias XQ_Thermal_AppNote_Data-mod.xlsx

80%

85%

90%

95%

100%

105%

110%

0 5 10 15 20 25 30

Θj-a

(°C/

W) o

f no

vias

Trace size (mm)

No Vias Plated Vias Filled Vias

XH thermal vias XQ_Thermal_AppNote_Data_v2.xlsx

80%

85%

90%

95%

100%

105%

110%

0 5 10 15 20 25 30

qj-a

(°C/

W) o

f no

vias

Trace size (mm)

No Vias Plated Vias Filled Vias

Chart 5: Thermal via results for XQ family LEDs Chart 6: Thermal via results for XH family LEDs

FR-4 BOARD THICKNESS

Chart 7 compares the relative Θj-a of two standard thicknesses of FR-4 PCBs with various copper trace sizes to that of an 0.8-mm thick

PCB with a 10 mm X 10 mm trace. The thinner FR-4 PCB had a lower Θj-a and is recommended for thermally demanding applications.

Chart 7: FR-4 PCB thickness comparison

FR-4 TRACE THICKNESS

To determine how useful thicker traces are for dissipating heat on an FR-4 PCB, Chart 8 and Chart 9 compare the relative Θj-a of three

thicknesses of FR-4 traces with various copper trace sizes to that of a 2-oz copper trace. The thicker traces have better Θj-a than the

thinnest trace, but increasing from 2 oz to 3 oz is a less significant improvement than increasing from 1 oz to 2 oz.

XQ FR4 board thickness XQ_Thermal_AppNote_Data-mod.xlsx

0%

50%

100%

150%

200%

250%

300%

0 5 10 15 20 25 30

Trace size (mm)

0.8 mm thick, 2 oz trace 1.6 mm thick, 2 oz trace

Θj-a

(°C/

W) o

f 0.8

mm

thic

k w

ith 1

0-m

m tr

ace


9


XQ FR4 trace thickness XQ_Thermal_AppNote_Data-mod.xlsx

90%

95%

100%

105%

110%

115%

120%

125%

0 5 10 15 20 25 30

qj-a

(°C/

W) o

f 2-o

z co

pper

Trace area (mm)

1 oz trace 2 oz trace 3 oz trace

XH FR4 trace thickness XQ_Thermal_AppNote_Data-mod.xlsx

90%

95%

100%

105%

110%

115%

120%

125%

0 5 10 15 20 25 30

qj-a

(°C/

W) o

f 2-o

z co

pper

Trace size (mm)


Chart 8: FR-4 trace thickness comparison for XQ family LEDs Chart 9: FR-4 trace thickness comparison for XH family LEDs

MCPCB TRACE THICKNESS

To determine how useful thicker traces are for dissipating heat on an MCPCB, Chart 10 and Chart 11 compare the relative Θj-a of three

thicknesses of MCPCB traces with various copper trace sizes to that of a 2-oz copper trace. The thicker traces have better Θj-a than the

thinnest trace but increasing from 2 oz to 3 oz is a less significant improvement.XQ MCPCB trace thickness XQ_Thermal_AppNote_Data-mod.xlsx

90%

95%

100%

105%

110%

115%

120%

125%

0 5 10 15 20 25 30

qj-a

(°C/

W) o

f 2-o

z tr

ace

Trace size (mm)


XH MCPCB trace thickness XQ_Thermal_AppNote_Data-mod.xlsx

90%

95%

100%

105%

110%

115%

120%

125%

0 5 10 15 20 25 30

qj-a

(°C/

W) o

f 2-o

z tr

ace

Trace size (mm)


Chart 10: MCPCB trace thickness comparison for XQ family LEDs Chart 11: MCPCB trace thickness comparison for XH family LEDs

MCPCB DIELECTRIC THERMAL CONDUCTIVITY

Chart 12 shows the simulated Θj-a with varying thermal conductivity of the dielectric of an MCPCB. Although here is some improvement in

performance when the dielectric conductivity increases above 2 W/mK, it is not particularly significant. However there is a large decrease

in thermal performance below 2 W/mK.


10


Chart 12: MCPCB dielectric thermal conductivity

FR-4 VS. MCPCB

Chart 13 and Chart 14 compare the relative Θj-a of four FR-4 PCBs with various copper trace sizes to that of an MCPCB. For smaller traces,

there is a very large thermal conductivity penalty for using an FR-4 PCB, but the penalty decreases as the trace size increases.XQ FR-4 vs. MCPCB XQ_Thermal_AppNote_Data_v3.xlsx

0%

50%

100%

150%

200%

250%

300%

350%

400%

0 5 10 15 20 25 30

qj-a

(°C/

W) o

f FR-

4 to

MCP

CB

Trace size (mm)

FR4: 1 oz trace FR4: 2 oz trace FR4: 2 oz trace with vias FR4: 3 oz trace MCPCB

XH FR-4 vs. MCPCB XQ_Thermal_AppNote_Data_v3.xlsx

0%

50%

100%

150%

200%

250%

300%

350%

400%

0 5 10 15 20 25 30

qj-a

(°C/

W) o

f FR-

4 to

MCP

CB

Trace size (mm)

FR4: 1 oz trace FR4: 2 oz trace FR4: 2 oz trace with vias FR4: 3 oz trace MCPCB

Chart 13: FR-4 vs. MCPCB comparison for XQ family LEDs Chart 14: FR-4 vs. MCPCB comparison for XH family LEDs

MEASURED VS. SIMULATED RESULTS

To verify the validity of the simulations herein, we measured and simulated the Θj-a of an XLamp XQ-D and an XH-G LED, each mounted

on a 25 mm X 25 mm FR-4 PCB on a heat sink, as shown in Figure 9. We measured and simulated FR-4 PCBs of 0.8 mm thickness with

various trace sizes.

XQ MCPCB dielectric thermal conductivity XQ_Thermal_AppNote_Data-mod.xlsx

0%

50%

100%

150%

200%

250%

0 1 2 3 4 5 6 7 8 9

qj-a

(°C/

W) o

f 2.2

W/m

K

Dielectric conductivity (W/mK)


11


Figure 9: XQ-D (left) and XH-G (right) thermal measurement configurations

Chart 15 shows that the measured data very closely matches the simulation data.

Chart 15: Comparison of measured and simulated Θj-a for XQ (left) and XH (right) 0.8 mm FR-4 PCB

TRACE SIZE RECOMMENDATIONS

Cree recommends the following trace sizes for optimized heat transfer without using vias or an overly large trace.

FR-4 PCB with XQ Family LED

• 2 oz copper

• 0.8 mm thick

• 10 mm X 10 mm trace

XH measured vs. simulated - .8 mm XQ_Thermal_AppNote_Data_v4.xlsx

0

20

40

60

80

100

120

0 5 10 15 20 25 30

qj-a

(°C

/W)

Trace size (mm)

Simulated 2 oz trace Measured 2 oz trace

XQ measured vs. simulated - .8 mm XQ_Thermal_AppNote_Data-mod.xlsx

0

20

40

60

80

100

120

0 5 10 15 20 25 30

qj-a

(°C/

W)

Trace size (mm)

Simulated 2 oz trace Measured 2 oz trace


12


MCPCB with XQ Family LED

• 2 oz copper

• 1.0 mm thick

• 5 mm X 5 mm trace

• thin (75 µm) dielectric with thermal conductivity ≥ 2.0 W/mK

FR-4 with XH Family LED

• ≥ 2 oz copper

• ≤ 0.8 mm thick

• ≥ 15 mm X 15 mm trace

MCPCB with XH Family LED

• ≥ 2 oz copper

• ≤ 1.0 mm thick

• ≥ 5 mm X 5 mm trace

• thin (75 µm) dielectric with thermal conductivity ≥ 2.0 W/mK

CHEMICAL COMPATIBILITY

As with any LED-based illumination system, it is important to verify chemical compatibility when selecting thermal interface materials,

as well as other materials to which the LEDs can be exposed. Certain materials can outgas and react adversely with the materials in

the LED package, especially at high temperatures when a non-vented secondary optic is used. This interaction can cause performance

degradation and product failure. Consult Cree’s Chemical Compatibility application note for compounds and products safe for use with

XLamp LEDs. Consult your PCB manufacturer to determine which materials it uses.

http://www.cree.com/~/media/Files/Cree/LED%20Components%20and%20Modules/XLamp/XLamp%20Application%20Notes/XLamp_Chemical_Comp.pdf

optimizing pcb thermal performance for cree xlamp xq & …figure 2 shows simulated thermal...

Documents