CNC Personalization of Thermal Dissipators: How Machining Precision Shapes Heat Paths

• 1
  CNC personalization of thermal dissipators starts at the heat path
• 2
  When machined fin geometry beats extruded or cast profiles
• 3
  Flatness, pitch and interface finish that hold a real joint
• 4
  Material and finish choices that change the numbers
• 5
  How we verify a dissipator before it ships
• 6
  Machined vs extruded vs cast dissipators

Heat leaves a device in three stages. It conducts through the package into the base, spreads sideways through the base material, then crosses into the air at the fin surfaces. Each stage has its own bottleneck. Machining changes all three, but not equally. Most of the gain comes from the first two stages, which is why base thickness and flatness usually beat fin count when you are chasing a few extra degrees.

Take a 40 × 40 mm aluminium base under a 30 W load. A rough sawn surface with Ra 3.2 μm and 0.05 mm of flatness error can add contact resistance equal to several degrees at the junction. Machining the base to Ra 0.8–1.6 μm and holding flatness inside 0.01 mm removes most of that penalty before any fin is touched.

Fin geometry matters next. Thin, closely spaced fins add surface area, but they also add boundary-layer interference and pressure drop if air cannot move through the channel. A 0.8 mm fin at 2.0 mm pitch behaves very differently from a 1.5 mm fin at 4.0 mm pitch. The right gap depends on whether you cool by natural convection, a low-speed fan, or a cold plate.

This is the part extrusion cannot do. An extruded profile is locked to one cross-section along its length. When the base needs a step, a pocket for a heat pipe, or a mounting boss at a specific height, the die cannot follow it. Machining can.

• 1
  Conduction firstBase flatness and thickness set the floor for everything downstream.
• 2
  Spreading secondA thicker base spreads heat to the outer fins instead of cooking the center.
• 3
  Convection lastFin pitch and height only help if air actually reaches the channel.

Extruded heatsinks are cheap and fast when the profile runs straight and the fin ratio stays inside what the die can support. Typical limits sit around a 1.2 mm fin with a 0.9 mm gap, and the aspect ratio rarely exceeds about 10:1. Push thinner and the die wears or the fins tear during handling.

Castings allow more complex shapes but bring porosity into the base. Porosity reduces the effective cross-section for conduction and can vary from part to part. For a 20 W device that may be acceptable. For a 200 W IGBT module, a pore under the die footprint can raise the junction temperature by more than the design margin allows.

CNC personalization of thermal dissipators solves both limits. On our 5-axis centers we routinely cut fins down to 0.5 mm with a 0.6 mm gap, hold wall thickness within ±0.02 mm, and keep the base solid. Skived and machined fins also have a slightly rougher surface than extruded ones, which in natural convection can add 3–8% to the effective area through early boundary-layer transition.

The trade is cost per part. Machining from solid removes a lot of material and takes time. It pays off when the part count is low, the geometry is complex, or the thermal margin is thin. Above roughly 10,000 identical simple parts, extrusion or casting usually wins on unit price.

• 1
  Choose machiningComplex base features, thin fins, or a first article that must meet the thermal target.
• 2
  Choose extrusionStraight profile, generous fin ratio, high volume, loose flatness requirement.
• 3
  Choose castingOrganic shapes and mounting bosses, when porosity under the die is controlled.

A heatsink only performs as well as the joint under it. Two flat surfaces still trap air in the microscopic valleys, and air conducts poorly. Thermal interface material fills those valleys, but only if the gap is small enough. The usual target is flatness within 0.02 mm across the die footprint and Ra 0.8–1.6 μm on the base.

Going smoother is not always better. Below about Ra 0.4 μm, some pastes and pads struggle to wet the surface and the joint can get worse. For hardcoat anodized bases, a light bead blast before anodizing often gives a more stable bond than a mirror polish.

Fin pitch tolerance affects airflow more than it affects conduction. If pitch varies by ±0.15 mm across the array, the tight channels starve while the wide ones bypass. Holding pitch within ±0.05 mm keeps the flow distribution even and makes fan curves predictable.

Mounting holes and bosses carry their own tolerance stack. A hole pattern held within ±0.05 mm avoids preloading the base when the screws are torqued, which would otherwise bow the part and open the joint. On our 3-axis and 4-axis mills we hold this on the same setup as the base face whenever the part size allows.

• 1
  Flatness target0.02 mm across the die footprint, measured on the base face.
• 2
  Finish targetRa 0.8–1.6 μm for paste; Ra 1.6–3.2 μm for thick pads.
• 3
  Pitch tolerance±0.05 mm keeps channel flow even and fan curves stable.

Aluminium 6061-T6 is the default for machined dissipators. It machines cleanly, anodizes well, and its thermal conductivity sits around 167 W/m·K. For higher heat flux, 2024 or 7075 give more strength but slightly lower conductivity. Copper C110 moves roughly twice the heat of 6061, at about three times the weight and a much higher cut time.

Anodizing adds a thin oxide layer that is electrically insulating and mildly thermally resistive. A clear or black anodize at 10–15 μm is usually harmless. Hardcoat at 50 μm can add a measurable interface resistance if it sits between the die and the base. In that case, mask the die footprint and anodize only the fins.

For high-power modules, nickel or silver plating on the base gives a stable, solderable surface without the oxide layer. Electroless nickel at 5–10 μm is common on copper cold plates that will be soldered or reflowed.

Surface finish on the fins affects convection, not conduction. Bead blasting or a light brush finish raises the effective area slightly and cuts glare inside enclosures. Polish looks good in a photo but adds cost without a thermal return on the fin side.

• 1
  6061-T6Default choice for most machined heatsinks and cold plates.
• 2
  C110 copperUse when heat flux is high and weight is not the limit.
• 3
  Mask the die areaKeep hardcoat and thick anodize away from the thermal joint.

Verification starts with the raw material. We check the alloy and temper against the certificate before the first cut, because a 6061-T6 bar that is actually 6061-T4 will bend during machining and lose flatness after anodizing.

In-process monitoring watches the base face and the fin array as they are cut. We measure flatness on the machine while the part is still clamped, so any drift is corrected before the setup is broken. The final inspection covers flatness, pitch, wall thickness and hole position on a CMM.

We inspect 100% of parts before shipment. Reports are available on request. For parts that will see thermal cycling, we can also check the base after a simulated anodize bake, since some alloys move a few microns when they see 180 °C.

This is the difference between a heatsink that meets a drawing and one that meets a thermal target. The drawing is the contract. The measurement is what the customer actually receives.

• 1
  Material checkAlloy and temper verified against the mill certificate.
• 2
  In-processFlatness and pitch measured while the part is still on the machine.
• 3
  FinalCMM report on flatness, pitch, wall thickness and hole position.