Thermal Management – Units, Measures and Models Steven Curtis – Chief Engineering Officer, Cambridge Nanotherm Ltd LED Light for You – April 2015
Thermal Management Knowledge Thermal Management – An art? Tier 1 LED A science? Pure Black Magic? Manufacturers & Mega-corps - Invested in strong engineering teams Mid-to-Large Limited exposure to Corporations - Electrical, mechanical, thermal management (250+ employees) simulation and test tool kit resources - Varying skill sets - Thermal Eng Small generally - Good facilities with well business outsourced kitted labs and suites
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 (R th ), 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 R th = 0.29K/W • 3W/mK material – 50 micron R th = 0.21K/W R th = t / kA • 1W/mK material – 100 micron R th = 1.1K/W • 0.3W/mK material – 3 micron R th = 0.18K/W *Assumes 1oz Cu and 1.5mm Al base material. Calculated on 1cm 2 substrate area (A)
Thermal resistance of a system • When you start putting a luminaire together, the P d thermal resistance of the system is the sum of all individual component thermal resistances ….represented here by their electrical counterpart. R th-die MCPCB • Don’t forget that it is not just the layers which have R th-MCPCB a thermal resistance, but the interfaces between them! Even down to the interface between the MCPCB Cu layer and its dielectric R th-TIM • In the most basic of models, this then allows you to R th-HS calculate the LED die junction temperature (T j ) - given that you know the thermal resistance, ambient temperature and electrical power; • So, T j = T a +(R th-total x P d )
So that’s it – job done?! Well, not quite… P d • Thermal resistance is only part of the equation, the other half is thermal capacity. C th-die R th-die MCPCB • Each material layer in the system has a capacity, an ability, to accept and store heat. C th-MCPCB R th-MCPCB • If we expand our simple thermal model we now have our thermal capacitors in the string. C th-TIM R th-TIM Now the maths starts to get more complex! • Thermal capacity, as well as depending upon C th-HS R th-HS 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.
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 TIM Cth moves horizontally , every time you Aluminium encounter a thermal capacitor the 3W/mK Dielectric graph moves vertically . • We know from our simple thermal 2oz Copper model showing the string of AuSn Die Attach Thermal Resistors and Thermal Capacitors that all materials posses Semiconductor an element of both, therefore, all elements are lines of various Rth gradients. 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 of 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 Cth thermally but you may have blown Rth-OS your budget for the project. • In under-specifying your die end Rth-Opt Rth-US materials – you’re now playing catch-up with your Thermal interface material (TIM) and heat sink. • The optimal path lies on a diagonal Over spec – nanoAg die attach, 5W/mK MCPCB line between these, whose gradient is dependent upon the Under spec – epoxy die attach, 1W/mK MCPCB power that needs to be dissipated. Rth
The optimised graph • In over-specifying your die end materials, you’ve done a great job Cth thermally but you may have blown Rth-OS your budget for the project. • In under-specifying your die end Rth-Opt Rth-US materials – you’re now playing catch-up with your Thermal interface material (TIM) and heat Need a much better sink. MCPCB dielectric!!! • The optimal path lies on a diagonal Over spec – nanoAg die attach, 5W/mK MCPCB line between these, whose gradient is dependent upon the Under spec – epoxy die attach, 1W/mK MCPCB power that needs to be dissipated. Rth
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 T j ? • Your options; • Use the thermal resistance data from data sheets to calculate R th-total and then T j . • 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 (V f ) which is directly proportional to the junction temperature (T j ) • 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 LC Substrate Nanotherm DM Substrate Thermal Conductivity (k) 4.8 W/mK (dielectric) 7 W/mK (dielectric) 0.1 K.cm 2 /W, 0.02 K.cm 2 /W R th(dielectric) * 0.25 K.cm 2 /W 0.1 K.cm 2 /W R th(substrate) * 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!
Recommend
More recommend
