⏩ TL;DR
In a sealed enclosure every watt has to leave by conduction: out of the component, through its pad and vias, along the copper, and into the case. The board runs hot wherever that chain is weakest, so find the weak link before layout, set copper weight in the stack-up before the first prototype, and test the finished board inside its own enclosure at full load. The four basic techniques (via arrays, heavier copper, placement, thermal pads) are summarised in our PCB miniaturisation guide. What follows is the next level down.
Table of Contents
Where does the heat path break?
Think of the route out as a chain of links: junction to package, package to pad, pad through the via array, into inner copper, then across to the case through a thermal pad or a metal mounting point. Heat in a sealed board settles at the highest-resistance link, and that link is often not the one you would guess from the schematic.
Three places are worth checking before routing starts:
- Under the part: how many vias sit beneath the exposed pad, and do they reach an inner plane large enough to spread the heat?
- Across the board: is there a copper path from that plane to somewhere that can shed heat, or does it dead-end in the middle of the board?
- Out to the case: does anything physically touch the housing? Copper-plated mounting holes that land on a metal boss or chassis point add a conduction path to the case, on top of any thermal pad.
What copper weight should you specify?
1oz copper (around 35µm) is the standard for outer and inner layers wherever possible, and it is where most boards should start.
Very high density, many-layer boards often have to thin their inner layers to 0.5oz (around 17µm) simply to fit the layer count. 2oz copper (around 70µm) is the exception.
It is most often chosen to carry high current, because thicker copper has lower resistance and the board itself heats less. It also conducts more heat sideways, but it makes the board thicker. In a sealed enclosure with no airflow it can earn its place.
Decide it in the layer stack at the start, because a routed board cannot take a heavier layer without a respin.
Can component choice cut the heat at source?
Yes, and it is a decision worth making before layout gives you a hotspot to chase.
On a small industrial motor-drive board project, ByteSnap Design’s engineers used an H-bridge: four transistors, two N-channel and two P-channel, that drive a motor forwards, backwards and to a stop. The first transistors chosen ran very hot, so they were changed more than once. The main factor was Rds(on), the transistor’s resistance when it is switched on. The other was the thermal resistance of the package. Our engineers moved to parts with a low Rds(on) in packages with low thermal resistance and added large heat dissipation planes on the bottom of the board.
Does firmware affect board temperature?
Yes. On the Cocoon smart home security board development, our engineers extracted every power-saving feature the system offered and advised on writing application code for better heat management, as per our winning British Engineering Excellence Awards entry.
A processor on a tighter power budget has less heat to move in the first place. That makes thermal a joint job for hardware and firmware. Sleep states, clock scaling, peripheral duty cycling and wake-up timing all change the heat the board has to shed, so agree them with the thermal design early, before a hotspot turns up on the bench weeks later.
How do you test a thermal design before it ships?
A thermal camera at bring-up tells you more in five minutes than a datasheet will. Validation goes further than finding one hotspot:
- Compare candidates side by side where you can. If pin-compatible, equivalent parts are available, test each under identical conditions and see which runs coolest, because a datasheet only shows the best case.
- Test in the real enclosure. Run the finished board inside its own housing, in a representative environment, at the highest performance level the product will ever see. Leave it there for an extended period and watch where the temperature settles.
- Test across the full operating range in an environmental chamber or test house. Some products spend their lives in an air-conditioned office, others in desert sun or arctic cold. Simulating those conditions catches issues before they appear in the field. And this matters most on industrial products.
Our guide to circuit board bring-up issues covers the wider bring-up process.
Where this leaves your enclosure design
If your board is also dense enough to need HDI routing, HDI microvias, signal integrity and EMC covers the layout side. For a second pair of eyes on the conduction path before layout locks in, ByteSnap Design’s custom PCB design team can review it.
Get your heat path reviewed before layout locks in
Trapped heat gets expensive to fix once a board is routed. ByteSnap Design's engineers plan the heat path, via strategy and component placement before layout locks in, alongside your firmware team where power budget and thermal load are one decision.
PCB Miniaturisation FAQs
How do you read a thermal camera image of a hot board?
A hotspot with no single component beneath it usually points to a broken or missing conduction path in the copper. A component well above its datasheet junction temperature at idle needs its placement checked against nearby heat sources. A lightly loaded part running warm often has a stalled or misconfigured power state, so check the firmware before the hardware.
Can a heatsink fix a sealed board that runs hot?
Sometimes. A heatsink fitted to a component package conducts heat away from it, as on a PC processor. In a sealed enclosure the heat still has to leave, so the heatsink needs a route onwards to the case, such as a thermal pad. Without one, it only moves heat from the part into the trapped air inside.
Dunstan is a chartered electronics engineer who has been providing embedded systems design, production and consultancy to businesses around the world for over 30 years.
Dunstan graduated from Cambridge University with a degree in electronics engineering in 1992. After working in the industry for several years, he co-founded multi-award-winning electronics engineering consultancy ByteSnap Design in 2008. He then went on to launch international EV charging design consultancy Versinetic during the 2020 global lockdown.
An experienced conference speaker domestically and internationally, Dunstan covers several areas of electronics product development, including IoT, integrated software design and complex project management.
In his spare time, Dunstan enjoys hiking and astronomy.
