Essential Thermal Mechanical Concepts Needed in Today s Microwave Circuit Designs. John Coonrod, Nov. 13 th, 2014

Transcription

1 Essential Thermal Mechanical Concepts Needed in Today s Microwave Circuit Designs John Coonrod, Nov. 13 th,

2 Outline Page Basic overview of heat flow for PCB s (Printed Circuit Board) Understanding the two basic types of PCB thermal issues Laminate properties critical to thermal management Results of heat flow experiments with thermal images Aging characteristics of thermoset circuit materials 2

3 Basic overview of heat flow for PCB s Page Three general types of heat transfer: Radiation Convection Conduction PCB applications typically are most concerned with Conduction, however: some PCB designs can have concerns with radiation convection can influence PCB thermal issues Heat conduction is a transfer of energy arising from temperature difference between adjacent parts of a body 3

4 Basic overview of heat flow for PCB s Page Assuming a thermal steady state, heat flow (H) is defined as: or A is area between the two fixed temperature plates k is thermal conductivity is temperature gradient, which is a change in temperature, with a change in distance 4

5 Basic overview of heat flow for PCB s Page The simple heat flow example related to a PCB 3D view of microstrip circuit without heat sink Assuming a microstrip circuit (double sided PCB) with a heat sink attached (TECA) Cross-sectional view of microstrip circuit with heat sink Assumptions for this example: Heat is generated at the signal plane The ground plane and heat sink are the same temperature TECA (Thermally and Electrically Conductive Adhesive) 5

6 Basic overview of heat flow for PCB s Page In general, for thermal management issues, a higher H (heat flow) is desired Higher heat flow allows the heat generated in the circuit to be efficiently moved to the heat sink, allowing a cooler circuit High heat flow allows increased power capability due to cooler circuit Power capability is limited by the Maximum Operating Temperature (MOT) of the circuit 6

7 Basic overview of heat flow for PCB s Page Important heat flow relationships to consider Increased thermal conductivity (k), yields increased heat flow (H) A thinner circuit (smaller L) will increase heat flow Wider signal conductor will increase A, which increases heat flow Baseline circuit Thinner circuit with improved heat flow 7

8 Page Understanding the two basic types of PCB thermal issues Device mounted on PCB and generating heat RF power applied to PCB conductors, generating heat 8

9 Device heating Page When a device is mounted on a circuit, it is often the heat source A via farm is sometimes used under the heat generating device Copper plated via s act as very good thermal conductors active device 9

10 Conductor heating due to applied RF Power Page Typically via farms can not be used where RF conductor heating is an issue because the via s will short the signal to ground Circuit material properties are far more critical for the RF heating scenario A via fence may be used to improve thermal management; this is done by using a Grounded Coplanar Waveguide (GCPW) 10

11 Page Laminate properties critical to thermal management Thermal conductivity Low dissipation factor Smooth copper surface profile Low dielectric constant, not critical, but can be helpful 11

12 Laminate Page properties critical to thermal management Power capability of PCB is limited by Maximum Operating Temperature (MOT) MOT is maximum temperature that should not be exceeded for long term thermal exposure of a PCB A MOT rating is achieved by a circuit fabricator submitting circuits to UL and the rating is based on the circuit construction and material properties If MOT is unknown, a good general rule of thumb is a MOT of 85 C, or in other words, the circuit should not be exposed to 85 C for long periods of time 12

13 Laminate Page properties critical to thermal management Thermal conductivity (TC) Unit is W/m/K Most common PCB circuit materials have low thermal conductivity Some high frequency PCB laminates have higher thermal conductivity General rule of thumb is TC of 0.50 W/m/K or greater is considered good Copper is an excellent thermal conductor; TC 400 W/m/K Typical TC values High Tg FR-4 Woven glass PTFE laminate RO3003 TM laminate RO4350B TM laminate RO4360G2 TM laminate RT/duroid 6035HTC laminate 0.30 W/m/K 0.25 W/m/K 0.50 W/m/K 0.62 W/m/K 0.80 W/m/K 1.44 W/m/K 13

14 Laminate Page properties critical to thermal management Dissipation Factor (Df) Unit-less Df affects dielectric losses A smaller Df number is better Typically a lower dielectric loss translates to lower insertion loss Lower insertion loss will generate less heat for RF conductor heating Typical Df values High Tg FR Woven glass PTFE laminate RO3003 TM laminate RO4350B TM Laminate RO4360G2 TM laminate RT/duroid 6035HTC laminate

15 Laminate Page properties critical to thermal management Copper surface profile The surface of concern, is at the copper-substrate interface of a laminate Unit is microns and describes the surface RMS (root mean square) Copper surface roughness affects conductor losses Smoother copper yields lower conductor loss, which translates to lower insertion loss [1] Lower insertion loss will generate less heat for RF conductor heating Copper surface roughness impact on insertion loss is frequency and substrate thickness dependent: Thinner substrates are more sensitive to copper surface roughness Higher frequencies have more conductor loss with rougher copper 15

16 Laminate Page properties critical to thermal management Low dielectric constant Generally dielectric constant is not directly related to thermal issues Low dielectric constant allows for a wider conductor A wider conductor will have less conductor losses Less conductor losses means less insertion loss and less heat generated Additionally a wider conductor increases the area (A) between the two circuit copper planes and the formula shows an increase in heat flow Narrow signal conductor Wide signal conductor 16

17 Page Results of heat flow experiments with thermal images Overall thermal testing test set-up Results from device heating study Results from applied RF power conductor heating study 17

18 Overall thermal testing set-up Page Overall thermal test set-up Device heating experiments: used a 100 ohm termination resistor and circuit as shown Used DC power to cause device heating Conductor heating experiments: Used same configuration as shown but the circuit was a microstrip or GCPW transmission line circuit RF power was at 3.3 GHz and power levels varied according to experiment conditions Mounting plate with circuit laminated to it Water cooled heat sink All circuits were laminated to the mounting plate with COOLSPAN TECA film 18

19 Overall thermal testing set-up Page Overall thermal test set-up, Device heating Heating the resistor with DC power eliminates the high frequency variables Heating a discrete component simplifies the thermal models Two simple microstrip circuit designs were used One design with via farms (top circuit) and one without via farms All circuits in this study used the same substrate thickness (20mils) and the same copper thickness (1.4mils) Top view of circuit patterns shown 19

20 Results from device heating study Page 229ºF 109ºC Woven Glass PTFE 84ºC rise above ambient 180ºF 82ºC RO4350B TM laminate Circuits with No Via Farm 56ºC rise above ambient 143ºF 62ºC RT/duroid 6035HTC laminate 36ºC rise above ambient 20

21 Results from device heating study Page Device heating experiment results RO4360G2 TM Circuits with No Via Farm 21

22 Results from device heating study Page Device heating experiment results 202ºF 109ºF Circuit using RO4350B laminate with No via 5 Watts Circuit using RO4350B laminate with via 5 Watts Comparison: Same circuit with No Via Farm and with Via farm 22

23 Results from device heating study Page Device heating experiment results Even though via farms improve the thermal management significantly, there is still benefit of materials with high TC Circuits with Via Farm 23

24 Page Results from applied RF power conductor heating study RF applied power was at 3.3 GHz For the best thermal images and most accurate measurements, everything in the camera view should have the same emissivity All circuits were painted flat black, with some noted exceptions The black paint increased the insertion loss of the circuit What is shown in this study is a worst case scenario because the insertion loss is higher than it would normally be without the paint Differences in insertion loss due to the black paint are noted The temperatures reported in this study are the highest temperatures recorded on the circuit during testing 24

25 Page Results from applied RF power conductor heating study Circuits with different dielectric constant (Dk) had slightly different locations for highest temperatures It is assumed that at 3.3 GHz there will be wavelength related high / low temperature areas and these will vary with materials of different Dk Since the copper conductor is extremely good for thermal conductivity, the wavelength related high / low temperatures are spread across some length Picture to the right is the top view and along the length of a microstrip transmission line circuit made on 24mil RO4360G2 TM laminate with an applied power of 50 watts at 3.3 GHz Most pictures in this study are not showing the length, but are a close up view of the hottest area with a top view across the conductor width as shown on the bottom right picture on the following page 25

26 Page Results from applied RF power conductor heating study 24mil RO4360G2 TM laminate with 50 watts applied power at 3.3 GHz In the circuit shown below it can be seen there is little heat spreading in the substrate Most of the heat is in the center of the conductor and the heat flow is down, into the page and into the heat sink below Rotate 90 and zoom 26

27 Page Results from applied RF power conductor heating study Comparison of very similar microstrip transmission line circuits Main difference between these circuits is dissipation factor Applied power is 50 watts at 3.3 GHz 24mil RO4360G2 TM laminate Max. temp. 111 F Max. temp F 25mil RO3006 TM laminate Dk = 6.4 Df = TC = 0.75 W/m/K Dk = 6.5 Df = TC = 0.79 W/m/K 27

28 Page Results from applied RF power conductor heating study Just for reference, shown is the impact on circuit heating with and without a heat sink attached The applied RF power was 85 watts at 3.3 GHz This is the same 50 ohm microstrip transmission line circuit: Without heat sink shown in the left picture With heat sink shown in right picture The circuit was made on 10mil RO4350B TM laminate 28

29 Page Results from applied RF power conductor heating study Microstrip transmission line study using different materials Comparing ID: 1 to 2 is showing heat rise difference due to circuit thickness difference 2 to 3 is showing thermal difference due to copper surface roughness ID 4 is a worst case example regarding circuit material properties ID 5 is a best case scenario Heat rise is based on the maximum circuit temperature after reaching thermal equilibrium Reference temperature for the heat rise is the temperature of the water cooled heat sink 29

30 Page Results from applied RF power conductor heating study Transmission line comparisons, microstrip vs. GCPW (via fence) It was found to be a bad comparison to use black paint for this test The paint increased the IL much more for the GCPW than the microstrip transmission line and the IL difference would bias the thermal comparison Instead of black paint covering the circuits, black bands were used The black bands were paint across the width of the circuit, less than 1/10 of the wavelength at 3.3 GHz for coverage and appeared to make no significant difference on insertion loss The black bands appeared to give relatively accurate thermal measurements when the thermal camera was zoomed in so only the black bands were in its view. 30

31 Page Results from applied RF power conductor heating study Microstrip transmission line compared to GCPW (via fence) The lowest temperature rise was the loosely coupled GCPW This benefits from the via fence, however the loose coupling does not increase the insertion loss much when compared to microstrip The tightly coupled GCPW is the highest temperature rise because it has the most insertion loss and the closer via fence is not enough thermal benefit to offset the loss 31

32 Page Results from applied RF power conductor heating study Thermal image of a microstrip bandpass filter, 30 watts at 2 GHz applied power The filter has 6 db loss in the passband and this is a partial reason why there is diminishing heat in the filter segments. Also at different frequencies within the passband, there are different current density patterns. When power is applied in the stopband, there is still some heat generated in the first elements 32

33 Page Results from applied RF power conductor heating study Microstrip edge coupled bandpass filter, material comparisons Tested at center frequency of the passband (3.34 GHz) with 50 watts 20mil RO4003C TM laminate 20mil RT/duroid 6035HTC laminate Dk = 3.55 Df = TC = 0.64 W/m/K Max. temp. = 231 F Dk = 3.6 Df = TC = 1.44 W/m/K Max. temp. = 141 F 33

34 Page Results from applied RF power conductor heating study Microstrip hairpin bandpass filter, Bare copper vs. ENIG Tested at center frequency of the passband (3.32 GHz) with 25 watts 20mil RO4003C laminate, bare copper 20mil RO4003C laminate, ENIG plated finish Dk = 3.55 Df = TC = 0.64 W/m/K Max. temp. = 152 F Dk = 3.55 Df = TC = 0.64 W/m/K Max. temp. = 262 F 34

35 Page Results from applied RF power conductor heating study Microstrip hairpin bandpass filter, Bare copper vs. ENIG Reference: screen shots of hairpin filters electrical performance 20mil RO4003C laminate, bare copper 20mil RO4003C laminate, ENIG plated finish 35

36 Page Aging of thermoset circuit materials Reality check.just about everything on Earth changes as it ages Printed circuit board materials are no different Some materials age more gracefully than others In general, thermoplastic circuit materials age better than thermoset materials Just for fun, this is a thermal image of the steering wheel in my car and the dash board behind it, on a warm summer day in the Arizona desert; it should not be a surprise that after a few years of exposure, the plastic dash had cracks. 36

37 Aging of thermoset circuit materials Page Summary of aging on circuit materials: All thermoset materials oxidize as they age, provided oxygen is present Elevated temperatures increase the speed of the aging effect Given enough time, thermoset materials will reach an equilibrium where the oxidation rate of the substrate is reduced to near zero Different thermoset resins and their additives can cause the material aging properties to vary from one material formulation to another The aging effect on high frequency circuitry is very design dependent 37

38 Aging of thermoset circuit materials Page Typical aging performance differences related to some common design types (assuming long periods of time at elevated temperatures which do not violate the circuits MOT rating and oxygen is present): Stripline has minor or no perceivable electrical performance difference Microstrip circuitry can have more electrical change than stripline Transmission lines, stubs, stepped-impedance features can have minimal differences in performance Edge coupled features may have more performance difference Copper features protect the substrate and reduce the oxidation effect Worst case scenario, which is usually not a design type, is when all of the copper is removed and the substrate is completely exposed 38

39 Aging of thermoset circuit materials Page Worst case scenario testing with completely exposed substrate 39

40 Aging of thermoset circuit materials Page Rogers high frequency circuit materials have been used successfully in some of the most demanding applications It is extremely rare that oxidation results in significant performance issues However, since the oxidation effect is highly dependent on: design application environment it is strongly recommended that customers evaluate each material and design combination to determine fitness of use over the life of the end product 40

41 Become a Member of Rogers Technology Support Hub for Additional Technical Tools & Information. Sign up to Receive Updates to be Kept up to Date on Rogers Recently Released Products. Microwave Impedance Calculator ROG Mobile App Electrical & Thermal Calculators Engineering Support Technical Papers 541

42 Page Thank Page You References: [1] J.W. Reynolds, P.A. LaFrance, J.C. Rautio & A.F. Horn III, Effect of conductor profile on the insertion loss, propagation constant, and dispersion in thin high frequency transmission lines, DesignCon RO3003, RO3006, RO4003C, RO4350B, RO4835, RO4360G2, LoPro, COOLSPAN and RT/duroid are licensed trademarks of Rogers Corporation The world runs better with Rogers. and the Rogers' logo are licensed trademarks of Rogers Corporation The information in this presentation is intended to assist you in working with Rogers' High-Frequency Materials. It is not intended to and does not create any warranties, express or implied, including any warranty of merchantability or fitness for a particular purpose or that any results show in this presentation will be achieved by a user for a particular purpose. The user is responsible for determining the suitability of Rogers' High Frequency Materials for each application. Prolonged exposure in an oxidative environment may cause changes to the dielectric properties of hydrocarbon based materials. The rate of change increases at higher temperatures and is highly dependent on the circuit design. Although Rogers high frequency materials have been used successfully in innumerable applications and reports of oxidation resulting in performance problems are extremely rare, Rogers recommends that the customer evaluate each material and design combination to determine fitness for use over the entire life of the end product. This presentation may not be reproduced, copied, published, broadcast, transmitted or otherwise distributed without the written approval of Rogers Corporation Rogers Corporation. All rights reserved 42