DEV Community

Asher Hu
Asher Hu

Posted on Originally published at hezidesign.com

Product Thermal Design: Passive Heatsinks, TIMs, and Ventilation — Practical Notes

Product Thermal Design Handbook: Passive Heatsinks, Thermal Interface Materials, and Ventilation Design — Practical Notes from a Structure Designer

A few years ago I took on a smart speaker project. The client wanted 30W of amplifier power in a sealed enclosure — no fan holes allowed. Our team figured passive cooling would be enough. The first prototype hit 62°C case temperature after half an hour of continuous playback. The client said "it can fry eggs." After that we did a full post-mortem and rebuilt our thermal design playbook — passive heatsinks, thermal interface materials, and ventilation layout — and haven't burned another project since.

Product structure design: thermal management seems straightforward, but it's where many projects fail. This article covers the key points from a practitioner's perspective.

Passive vs. Active Cooling: When to Choose Which

The first step in thermal design isn't crunching numbers — it's picking the approach.

Passive cooling relies on natural convection and radiation — no moving parts, low cost, high reliability, zero noise. It works for products under 10W: smart-home sensors, remote controls, low-power LEDs. We built a temperature/humidity sensor that dissipates just 0.5W — the plastic housing alone handles it, no heatsink needed.

Active cooling covers fans, heat pipes, vapor chambers, and liquid loops. Fan-based designs dissipate more heat but add noise and reliability risk (fan lifespan is typically 30,000–50,000 hours). Heat pipes and vapor chambers are passive conductors but active spreaders — heat pipes reach 5000 W/mK+, 25× that of pure aluminum, ideal for moving heat quickly from source to a radiating surface. On a projector thermal design we used a single 8mm heat pipe to channel LED heat to the housing's cooling fins — the temperature rise stayed under 15°C.

The rule of thumb is two thresholds: if heat flux exceeds 0.2 W/cm², or if the surface temperature limit is below 45°C, passive cooling alone probably won't cut it — go active or hybrid.

Heatsink Design: Aluminum or Copper? How Tight Should the Fins Be?

The heatsink is the most basic thermal component, but newcomers trip here often.

Material: Aluminum (ADC12 die-cast or 6063 extruded) conducts about 200 W/mK; copper does ~400 W/mK. Copper conducts twice as well but costs 4× more, weighs 3× more, and is harder to machine — fins require CNC or welding. In practice, 80% of projects are fine with aluminum. Copper is reserved for high heat-flux, space-constrained cases like high-power LED modules.

Fin pitch is where many designs go wrong. Too tight and air can't flow, dropping efficiency. Under natural convection, aim for 6–10mm between fins. For forced-air (with a fan), tight pitch down to 2.5–4mm is okay. Fin thickness should be 0.8–1.5mm — too thin and they deform; too thick and you waste material and weight.

On a router thermal design we used the die-cast ADC12 housing itself as the heatsink: 3mm base, 15mm-tall fins at 5mm pitch. Under full CPU load for 4 hours, the housing stabilized at 52°C — 8°C below the client's 60°C limit. The beauty is that the housing is the heatsink, saving a separate component and an assembly step.

Thermal Interface Materials: Grease, Pad, or Graphite?

The interface between die and heatsink is microscopically rough, and air fills the gaps — at just 0.026 W/mK. You need a thermal interface material (TIM) with conductivity of 1–15 W/mK to bridge it. Three common types each have their niche.

Thermal grease: highest conductivity (3–12 W/mK), lowest thermal resistance, but messy to apply — must be spread evenly to 0.1–0.2mm thickness. Not ideal for large areas or rework. On consumer electronics we use grease for CPU and GPU hotspots, typically 5 W/mK+.

Thermal pad: 3–8 W/mK, compressible, tolerates larger gaps, easy to apply, automation-friendly. Higher resistance than grease, suited to gaps of 0.3–3mm. On a power-supply design, the 1.5mm gap between MOS and housing was filled with a pad — first-pass OK.

Graphite sheet: exceptional in-plane conductivity (300–1500 W/mK), ultra-thin (0.025–0.1mm), used in phones and tablets where every micron counts. But it's anisotropic — through-plane conductivity is only 5–10 W/mK — so it's for heat spreading, not conduction. We tore down a branded tablet; the CPU heat was spread across the entire backshell via graphite sheet, paired with a metal mid-frame.

Ventilation Design: Position and Size

For naturally convected products, vent position and size decide efficiency. The basic rule: inlets at the bottom or low on the sides, outlets at the top or high on the sides — letting hot air rise creates natural flow. Vent area should be at least 2–3× the heat-source area. Hole diameters of 6–10mm work well — bigger risks EMI and dust ingress, smaller adds flow resistance. On a set-top box, we cut a row of slots (8×80mm) in the bottom shell and matched outlets on top — the chip ran 12°C cooler than the unvented version.

Real Case: Smart Speaker Passive Cooling — from Scorched to Ship

That smart speaker from the opener — how did it end? Three changes.

First, an aluminum heatsink on the amp chip: 45×45×15mm, 4mm base, 6mm fin pitch (natural convection). Second, the housing was switched to die-cast aluminum with a thermal pad on the interior contact. Third, an annular inlet ring at the bottom and an annular outlet ring at the top created a natural convection chimney. The revised prototype held 48°C after an hour of playback — 14°C cooler than the original, and the client signed off.

The lesson: passive cooling isn't "just don't use a fan." It's a four-dimensional exercise — heatsink, TIM, housing material, and airflow path — and every dimension has to be right.

FAQ: Common Thermal Questions

Q: How do you cool a product with a plastic housing?

A: Plastic conducts poorly (~0.2 W/mK), so it can't be the heatsink. Three options: (1) add a metal heatsink inside touching the heat source; (2) use thermally conductive plastic (2–5 W/mK) at higher cost; (3) add vents and a fan. Low-power products handle passive convection and radiation; above 10W, add a heatsink.

Q: Die-cast vs. extruded aluminum for heatsinks?

A: Depends on shape. Die-cast (ADC12) handles complex shapes and can integrate housing and heatsink, but conducts ~96 W/mK — lower than pure aluminum. Extruded (6063) hits ~200 W/mK but only does straight fins. High volume → die-cast; low volume or high performance → extruded.

Q: How often should thermal grease be replaced?

A: Consumer products typically never — grease lasts 5–8 years. In high-temp environments (70°C+) or under constant vibration, it may dry out or pump out; check at 3–5 years. Pads last 10+ years with no maintenance.

Q: Can you cool a waterproof (IPX5) product?

A: Yes. Options: metal housing as heatsink with sealing gaskets; potting the whole board with thermally conductive epoxy; running a heat pipe to a dedicated cooling zone on the waterproof housing. We built an IPX6 outdoor camera that used a thermal pad to conduct heat to an aluminum housing with large cooling fins — passed.


Originally published on the Hezi Industrial Design blog: Product Thermal Structure Design Guide

Top comments (0)