Thermal Management Units, Measures and Models Steven Curtis Chief - - PowerPoint PPT Presentation

Thermal Management – Units, Measures and Models

Steven Curtis – Chief Engineering Officer, Cambridge Nanotherm Ltd LED Light for You – April 2015

Thermal Management Knowledge

Tier 1 LED Manufacturers & Mega-corps

• Invested in strong

engineering teams

• Electrical, mechanical,

simulation and test resources

• Good facilities with well

kitted labs and suites Mid-to-Large Corporations

(250+ employees)

• Varying skill sets
• Thermal Eng

generally

• utsourced

Small business

Limited exposure to thermal management tool kit

Thermal Management – An art? A science? Pure Black Magic?

Units and measures

• W/mK – thermal conductivity (k), the ability of a material to transfer

heat through it, a bulk property of a material

• K/W – thermal resistance (Rth), the resistance to heat flow exhibited

by a material, a property relating to a given physical form of a material When looking at a data sheet, which is the correct figure to work with? A lot of decisions still based on the headline figure - ‘the 5W/mK is better than the 3W/mK material’, but this is flawed….

Which material?

• 10W/mK MCPCB – Chinese MCPCB highly filled epoxy dielectric
• 3W/mK MCPCB – Leading MCPCB type for HB LED (filled epoxy)
• 1W/mK MCPCB – standard entry-level MCPCB (filled epoxy)
• 0.3W/mK MCPCB - Steve’s super-special MCPCB (B&Q own brand carpet

tile adhesive) “Looking at these figures, it’s a no-brainer, I’ll take the 10W/mK please”

(Caveat: assuming price, mechanical and electrical properties are the same for all)

Let’s just think that one through first….

Which material? (2)

• Let’s calculate the thermal resistance of each of these material
• To do that we need to know the crucial missing piece of data -

dielectric thickness;

• 10W/mK material – 200 micron
• 3W/mK material – 50 micron
• 1W/mK material – 100 micron
• 0.3W/mK material – 3 micron

Thermal resistance

• Let’s calculate the thermal resistance* of each of these material.
• To do that we need to know the crucial missing piece of data -

dielectric thickness (t);

• 10W/mK material – 200 micron

Rth = 0.29K/W

• 3W/mK material – 50 micron

Rth = 0.21K/W

• 1W/mK material – 100 micron

Rth = 1.1K/W

• 0.3W/mK material – 3 micron

Rth = 0.18K/W

*Assumes 1oz Cu and 1.5mm Al base material. Calculated on 1cm2 substrate area (A) Rth = t / kA

Thermal resistance of a system

• When you start putting a luminaire together, the

thermal resistance of the system is the sum of all individual component thermal resistances ….represented here by their electrical counterpart.

• Don’t forget that it is not just the layers which have

a thermal resistance, but the interfaces between them! Even down to the interface between the MCPCB Cu layer and its dielectric

• In the most basic of models, this then allows you to

calculate the LED die junction temperature (Tj) - given that you know the thermal resistance, ambient temperature and electrical power;

• So, Tj = Ta+(Rth-total x Pd)

MCPCB Rth-die Rth-MCPCB Rth-TIM Rth-HS Pd

So that’s it – job done?!

Well, not quite…

• Thermal resistance is only part of the

equation, the other half is thermal capacity.

• Each material layer in the system has a

capacity, an ability, to accept and store heat.

• If we expand our simple thermal model we

now have our thermal capacitors in the string. Now the maths starts to get more complex!

• Thermal capacity, as well as depending upon

a materials thermal conductivity, also depends upon it’s size/mass.

• An Al heat sink has a much higher thermal

capacity than a MCPCB dielectric.

MCPCB Rth-die Rth-MCPCB Rth-TIM Rth-HS Pd Cth-die Cth-MCPCB Cth-TIM Cth-HS

How can we visualise this easily?

• Luckily, those clever maths guys

have figured out a simple graph for us.

• It’s called the cumulative

structure function, to the ley person, it is a graph of thermal capacity v. thermal resistance.

Understanding the C v R graph

• Every time you encounter a

thermal resistance the graph moves horizontally, every time you encounter a thermal capacitor the graph moves vertically.

• We know from our simple thermal

model showing the string of Thermal Resistors and Thermal Capacitors that all materials posses an element of both, therefore, all elements are lines of various gradients.

Rth Cth Aluminium 3W/mK Dielectric 2oz Copper AuSn Die Attach Semiconductor TIM Total Rth

How do I optimise my luminaire design using the C v R method?

• A hierarchy exists in these simple models – the importance of good

thermal conductivity/low thermal resistance is higher at the die end

• f the chain.
• Consider the hose pipe analogy!
• A design can fall into three categories;
• Materials have been under-specified and high thermal resistances mean that

you have your components running hot and component life reduced.

• Materials have been over-specified and you have ice cold components, but a

luminaire that costs a fortune.

• Materials optimised for the required power dissipation - well done!

The optimised graph

• In over-specifying your die end

materials, you’ve done a great job thermally but you may have blown your budget for the project.

• In under-specifying your die end

materials – you’re now playing catch-up with your Thermal interface material (TIM) and heat sink.

• The optimal path lies on a diagonal

line between these, whose gradient is dependent upon the power that needs to be dissipated.

Rth Cth

Under spec – epoxy die attach, 1W/mK MCPCB Over spec – nanoAg die attach, 5W/mK MCPCB

Rth-OS Rth-Opt Rth-US

The optimised graph

• In over-specifying your die end

materials, you’ve done a great job thermally but you may have blown your budget for the project.

• In under-specifying your die end

materials – you’re now playing catch-up with your Thermal interface material (TIM) and heat sink.

• The optimal path lies on a diagonal

line between these, whose gradient is dependent upon the power that needs to be dissipated.

Rth Cth

Under spec – epoxy die attach, 1W/mK MCPCB Over spec – nanoAg die attach, 5W/mK MCPCB

Rth-OS Rth-Opt Rth-US

Need a much better MCPCB dielectric!!!

Watch out for…

• Signs of under specifying
• Hot components!
• Falling into the trap of having to increase Cu layer on MCPCB to gain capacity

and heat-spreading in this layer.

• Increasing the MCPCB Aluminium thickness.
• Improving your TIM material and enlarging your heat sink doesn’t have as big

an effect as expected.

• Signs of over specifying
• Huge reductions in Tj, beyond what is reasonable required.
• The marketing guy is telling you that your competitors are half the price.

All very well in theory, but how do I get my Tj?

• Your options;
• Use the thermal resistance data from data sheets to calculate Rth-total and then Tj.
• Dependent upon accuracy (aka honesty!) of data sheet figures, requires no experimental set-

up.

• Measure using thermocouples on solder points
• quick, easy, but you’ll not get to the actual junction temperature.
• Use an IR camera to look at the LED
• Again, fairly easy to do, but beware that all materials in front of the die (phosphor,

encapsulation) have different emissivities and this can skew the readings.

• Use a source-and-measure approach (JEDEC JESD51) to measure Forward Voltage

(Vf) which is directly proportional to the junction temperature (Tj)

• Requires high specification equipment and controlled test rig, but gives accurate results.
• And also can be used to generate the C v R curves and understand what is happening in the

layers!

One final thing, and thanks

Don’t forget, the more power (heat) that you are looking to dissipate, the higher performance your materials in the stack will need to be. That’s where Nanotherm’s materials come into their own.

Nanotherm MCPCB and Packaging materials are based on our unique ceramic dielectric system. (*1.5mm Al MCPCB substrate with 10micron nano-ceramic and 1oz Cu)

www.camnano.com | tel: +44 (0)1440 765 520 | email: info@camnano.com Thank you and enjoy the rest of the conference!

Nanotherm LC Substrate Nanotherm DM Substrate Thermal Conductivity (k) 4.8 W/mK (dielectric) 7 W/mK (dielectric) Rth(dielectric) 0.1 K.cm2/W, 0.02 K.cm2/W Rth(substrate) 0.25 K.cm2/W 0.1 K.cm2/W * *