Dr Giles Humpston, Applications Manager
A variation of this question vexed Prince Hamlet while the modern version unsettles many a LED PCB designer. It’s all about copper thickness. When designing a PCB for LEDs, what weight of copper should be specified?
The first thing to decide is why a design needs thick or thin copper. Usually there are two considerations. These are the ability of the copper tracking to conduct electricity efficiently to the LEDs and secondly, remove the heat from the LEDs by spreading it over a large area for easy dissipation.
High power LEDs and LEDs in densely packed arrays produce too much heat to be attached to traditional FR4 PCBs. Instead, metal-clad PCBs (MCPCB) are used. Basically, these comprise an aluminium plate and a copper wiring trace, separated by a dielectric that is very thin and thermally conductive. By selecting a dielectric with these attributes, the thermal resistance of the dielectric can be as low as 0.01 C.cm2/W, making it very easy for any heat in the copper to reach the aluminium plate and be conducted away to a heat sink.
The current carrying ability of copper tracking on a PCB is limited by temperature rise. The passage of excessive current causes resistance heating, this increases the resistance of the tracks causing additional heating and eventually the track will fuse. Learned folk generally consider a 10°C temperature rise to be safe and calculators exist to predict the current that tracks on an FR4 board can safely carry. However, the thermal conductivity of an MCPCB is orders of magnitude greater. This means the tracks on an MCPCB will stay cool even when carrying fifty times the permissible current on an FR4 PCB. Thus, there is seldom a need for thick copper tracks on an MCPCB.
Copper is an excellent thermal conductor (400 W/mK), so it is often assumed that by having a thick area of copper connected to an LED it will keep the device cool. Alas (poor Yorick) this is not so. Consider a copper track 1mm wide, 50 um thick, 5 mm long. The end-to-end thermal impedance is 259°C/W. Comparing that to the thermal impedance of the dielectric and it is easy to see that virtually all the heat will be sucked instantaneously downward into the aluminium plate and essentially none will spread radially. Changing the copper between 1 and 3 oz makes not one jot of difference. Having a copper pad underneath an LED is better than not having one, but anything larger than 3.5mm diameter is well beyond the point of diminishing returns in terms of improving heat conduction by radial spreading.
In fact, selecting thick copper actually incurs some penalties. The first is that the ability to make fine tracks and gaps diminishes with increasing weight. For the new chip size package LEDs, some of which have fiendishly small gaps between the anode and cathode pads, just forming the wiring trace in heavy copper can be a challenge that few PCB shops are prepared to accept. The second and potentially more bruising consideration is the wrath of the buyer on receipt of the quotation for heavy copper.
So, the next time you hear Guilderstern say to Hamlet: “If it will please you to make me a sensible answer…”, you’ll know the correct response should have been “one ounce copper”, (before the stage direction continues on to the bloodbath at Elsinore).
About the Author
Dr. Giles Humpston is a metallurgist by profession and has a doctorate in alloy phase equilibria. He is a cited inventor on more than 250 patents and has co-authored over 150 papers as well as several text books. Dr Humpston currently works as the Field Applications Manager for Cambridge Nanotherm on thermal substrate technologies.