Most engineers reach for a heatsink the moment a component runs warm. The reflex is understandable, but a finned sink is not free cooling. It buys a few degrees with board space, bill of materials cost, and a new point of failure: the thermal interface material that bonds sink to board dries, cracks, or was simply never mounted right. For a surprising number of designs, the aluminum substrate under the component already moves enough heat to skip the sink entirely.
What the Sink Was Actually Doing
A heatsink works by increasing the surface area exposed to air. Heat leaves the component, crosses the board, crosses a slab of thermal interface material, climbs into the fin, and finally convects into still air. Every one of those crossings has a resistance, and the fin is only as good as the weakest link beneath it. An aluminum core board collapses the first three steps into one: a copper circuit layer, a thin dielectric, and a 200 W/m·K aluminum base that already spans most of the enclosure. The board is the spreader. Bolting a fin on top of a board that already spreads heat well often just moves the bottleneck somewhere you cannot see.
The thermal math behind that spread is worth a look before you spec a sink. Our guide to how an aluminum base moves heat at the component level walks through dielectric conductivity and the 0.5 to 3.0 °C/W you can expect from the stack.
The Natural Convection Ceiling
Once heat reaches the board surface, it still has to leave into air, and still air is a poor conductor. Natural convection delivers a heat transfer coefficient of roughly 5 to 25 W/m²·K, about one to two orders of magnitude below a forced fan. That ceiling is why spread area matters so much: with a low convection coefficient, the only lever left is giving the heat more board to escape from. A 2 W device dumped into a 4 cm² patch will cook; the same 2 W spread across 40 cm² of aluminum base may never exceed what the enclosure can shed passively.
Spread the Heat Before You Sink It
The single highest-leverage move is area. A thicker aluminum base, 2.0 to 3.0 mm, carries heat farther from the hot component before it ever meets the convection limit, because the base itself is the path, not a bolt-on afterthought. Thin 1.0 mm boards work for low loads but pinch the spread, forcing the heat back toward the component. When the case is sealed and silent, thickness is the cheapest insurance you have.
Let the Copper Do More
Heavier copper, 2 to 4 oz, shortens the in-plane path from component to base and lowers resistance along the way. A power stage that would need a wide pour on 1 oz FR-4 often fits comfortably on 2 oz aluminum, because the base underneath is already pulling heat down. Going heavier than the current demands just adds cost, so match copper weight to the actual amp load rather than to habit.
Pick the Dielectric for the Job
The dielectric is the one layer that both insulates and conducts, so its conductivity sets the whole stack’s resistance. A 2 to 3 W/m·K dielectric shaves the path versus a 1 W/m·K part, but it costs more and is not always needed. The three-layer aluminum substrate structure stays the same either way; only the dielectric grade changes. For LED and small consumer loads, 1 W/m·K is often enough, while a 3 W/m·K part earns its price when the device sits inches from its thermal limit.
Use the Enclosure as the Radiator
The board spreads; the case radiates. Mounting the aluminum base against a metal enclosure wall turns the whole housing into the fin you did not have to buy. A plastic shell kills this option, which is why sealed plastic enclosures are the hardest passive-cooling cases, and metal ones are the easiest. When the wall is already aluminum or steel, you have effectively added surface area for free.
When a Fin Is Still the Answer
Honesty matters here. Above roughly 3 to 5 W/cm² in a sealed small case, no amount of spread area replaces a real sink or a fan. The convection ceiling does not move just because the board is good. At that point a fin, or forced air, is the correct call, and pretending otherwise ships a board that fails in the field. The point of skipping the sink is not to refuse it on principle, but to stop adding one where the substrate already does the job.
If the device is sealed, silent, or too thin for a fin, design the board to spread first. Get an aluminum PCB quote sized for passive cooling, or send us the thermal load and we will suggest a stack-up that skips the heatsink.
