Pin Fin vs Straight Fin vs Skived: Heat Sink Geometry Compared

Published 2026-11-02 · 9 min read

For forced-air electronics up to roughly 300 W, straight fins give the lowest cost per degree of cooling, pin fins win when airflow direction is unknown or the sink sees multi-directional flow, and skived fins win when fin density per millimetre of base is the constraint. SkillsProto machines all three geometries from solid aluminium or copper with no tooling charge, MOQ 1, at ±0.005 mm on critical features and 3-7 days lead time, so geometry can be chosen on thermal merit rather than on tooling budget.

Why geometry decides the answer before material does

A heat sink is a surface-area problem dressed up as a metal part. The job of the geometry is to put as much surface as possible into the air stream without choking that stream, and to move heat from a small die footprint into a large base fast enough that the fins are not starved.

Three variables set the ceiling on what any geometry can achieve: the base spreading resistance between the heat source and the fin field, the convective coefficient the airflow can sustain through the fin gaps, and the fin efficiency — how much of each fin actually reaches air temperature before it ends. Geometry sets all three. Material only scales them.

That is why two heat sinks with identical footprints and identical alloys can differ by 40% in thermal resistance, and why the cheapest geometry is often the one with the best performance per euro at prototype quantity.

The three geometries, in one line each

  • Straight fins. Parallel plates, cut or milled, open at both ends. Airflow must run along the channels. Highest fin efficiency per gram, lowest machining time, cheapest to make in small numbers.
  • Pin fins. An array of round, square or elliptical posts. Airflow can arrive from any direction and the array mixes flow between pins, which helps where the fan is not ducted. Pressure drop is higher and machining time is 2-4x straight fins.
  • Skived fins. A skiving tool peels a thin layer of metal up from a solid block, folding it into a fin. Produces the thinnest fins at the highest density, usually in copper, with the fin and base as one continuous crystal. Tooling is geometry-specific, so it is a volume process, not a prototype one.

The comparison that matters at prototype quantity

GeometryAirflow requirementTypical fin thicknessFin densityTooling costLead time at 50 pcsRelative part cost
Straight fin, machinedDirectional, ducted0.8-1.5 mm4-8 fins per 10 mmNone3-7 days1.0x
Pin fin, machinedAny direction1.0-2.0 mm posts5-9 pins per 10 mmNone4-8 days1.8-3.0x
Skived finDirectional, tight duct0.3-0.6 mm10-20 fins per 10 mmSkiving die3-5 weeks0.4-0.7x at 5,000+
Extruded finDirectional, ducted1.0-2.0 mm5-10 fins per 10 mmExtrusion die4-6 weeks0.3-0.6x at 2,000+
Cast finTightly ducted1.5-3.0 mm3-6 fins per 10 mmCasting tool5-8 weeks0.5-0.8x at 10,000+

The relative cost column is the one that decides most projects. At 50 pieces a machined straight-fin sink is the cheapest thing that works. At 5,000 pieces the same sink is the most expensive option on the list, by a factor of two or more against skived or extruded.

Straight fins: the default that is hard to beat

Straight fins are what a machinist would cut with a slitting saw and what a thermal engineer would draw first, and for good reason. The fin is a flat plate in a channel, which is the geometry with the best understood convective behaviour, the lowest pressure drop per unit of surface, and the highest fin efficiency for a given thickness.

Machining is straightforward: a stack of thin slitting saws or a small end mill cuts the channels in one pass over the base, so machining time scales with base length, not with fin count. That is why a 60 x 60 mm sink with 12 fins costs roughly the same to machine as one with 20 fins above the same base — the passes get closer together but the tool path length is unchanged.

The constraint is direction. If the fan is not ducted, or the enclosure lets air wander, a straight-fin sink can lose a third of its rated performance because the flow crosses the fins instead of running down them. In that case the fix is a shroud or a duct, not a different geometry.

There is also a manufacturing floor worth knowing: below about 0.8 mm fin thickness in aluminium, the fin starts to deflect under cutting pressure, and the machinist has to take lighter passes, which raises cost faster than the extra surface lowers thermal resistance. SkillsProto quotes fin thickness down to 0.5 mm, but flags anything under 0.8 mm as a cost step rather than a performance bargain.

Pin fins: buying immunity to bad airflow

A pin fin array is what you specify when you cannot control where the air comes from. Because each pin is a discrete obstacle rather than a wall, flow arriving at 45 degrees still scrubs surface, and the wakes behind the pins mix the boundary layer, which raises the convective coefficient between pins.

The trade is pressure drop and machining time. A pin array of the same footprint as a straight-fin sink might have 30-60% more wetted area, but it also presents far more resistance to the fan, so the fan curve matters as much as the sink. On a low-static-pressure axial fan the extra area can be wasted: the fan simply moves less air and the sink runs hotter than the model predicted.

Machining pin fins from solid is a pocketing job. A small end mill has to clear the field between pins, so tool path length scales with pin count squared in the worst case, and the tool has to be long enough to reach the base — which limits how closely pins can be spaced before the tool shank fouls its neighbours.

Where pin fins genuinely pay: natural convection with no fan at all, enclosures with multiple heat sources and chaotic internal flow, and any design where the mechanical team cannot guarantee a duct. Those are exactly the cases where a straight-fin sink's datasheet number evaporates.

Skived fins: the copper density play

Skiving is not machining in the usual sense. A tool with a controlled rake angle shears a continuous chip up from a solid block without removing it, so the fin stays attached to the base with no interface, no solder layer and no filler metal. Because there is no bond line, the fin-to-base joint has no thermal resistance of its own — the one failure mode that quietly costs machined-and-bonded assemblies 10-20% of their performance.

The process produces the highest fin density available in a one-piece part: 0.3-0.6 mm fins on 1.0-2.0 mm pitch are routine in copper. It is also the process with the least geometry freedom, because the skiving tool is cut for one fin height, one pitch and one material. Change the fin count and you need a new tool.

For prototypes, skiving is the wrong answer almost every time. For a 5,000-piece production run in copper, it is usually the cheapest way to reach a thermal resistance that no machined sink can hit, because the fin density is simply not achievable any other way.

What CNC can and cannot do to a fin

The honest limits, in numbers a designer can put on a drawing:

FeaturePractical machined limitWhat happens beyond it
Fin thickness, aluminium0.5 mmFin deflects, needs light passes, cost steps up
Fin thickness, copper0.5 mmChatter and burr, needs support or a second op
Fin height to thickness ratio8:1Tall thin fins ring and taper, tolerance drifts
Slot width between fins1.0 mmTool breakage risk, depth limited to 3-4x diameter
Base flatness over 50 mm0.05 mmNeeds a finishing pass on a clean setup
Mounting face finishRa 1.6 µmBelow Ra 1.6 needs lapping or a separate op
Pin array pitch1.5 mmTool shank fouls neighbours on deep pockets

Two of these decide most quotes. The first is fin height to thickness ratio: at 8:1 a 0.8 mm fin can be 6.4 mm tall before it becomes a machining problem, which covers most electronics cooling. Beyond that the fin has to be roughed with stock left on, then finished in a second operation, and the price reflects the second setup rather than the metal.

The second is the thermal interface itself. Fins are the visible part of the design and rarely the reason a sink underperforms; a base that is 0.05 mm out of flat over the die footprint adds contact resistance that no amount of fin area recovers. SkillsProto holds 0.05 mm flatness and Ra 1.6 µm on mounting faces as standard, and will go to Ra 0.4 µm on request, because that is where the thermal budget is actually spent.

Cost and lead time: what the quote is actually pricing

Four things move a machined heat sink quote, in this order:

1. Material grade and stock size. A 60 x 60 x 40 mm block of 6061 is a few euros of metal; the same block in copper is 4-6x more and much slower to cut.

2. Fin count at fixed envelope. More fins means more tool path in the same area. A 12-fin and a 24-fin sink differ mainly in machining minutes.

3. Fin aspect ratio. Beyond 8:1 height-to-thickness, the shop adds a roughing and finishing sequence, which is a step change, not a slope.

4. Interface requirements. Flatness, Ra and hole position on the mounting face are inspection time. A ±0.005 mm bore for a press-fit insert costs more than the fins.

Lead time is dominated by material availability, not machining. Standard 6061 plate is usually in stock and a machined sink ships in 3-7 days, with a 48-hour expedite possible on simple geometry. Copper plate is often a mill order, which is where a 2-week quote turns into a 4-week one.

How to choose in ten minutes

Work through four questions in order and stop at the first one that answers itself:

1. Is the airflow ducted and directional? If yes, straight fins. If no, pin fins or add a duct.

2. Is the production volume above 2,000 pieces? If yes, evaluate extrusion or skiving, and use a machined sink only for the prototype.

3. Is thermal resistance already at the limit of what aluminium can do? If yes, move to copper before changing geometry — conductivity goes from 167 to 401 W/m·K, which is a bigger lever than fin shape.

4. Is the base footprint smaller than the heat source needs? If yes, the problem is spreading resistance, and a thicker base or a vapour chamber does more than any fin pattern.

Most projects that arrive at SkillsProto asking for "the best fin geometry" turn out to have a spreading or interface problem, and the machined sink that solves it is a straight-fin part with a 6 mm base and a flat mounting face.

Tolerances and finish on the thermal path

The fin field is forgiving; the thermal path is not. Three features carry the performance:

  • Base flatness across the die footprint, 0.05 mm as standard. A concave base touches the die at the edges only.
  • Mounting face finish, Ra 1.6 µm or better. Surface roughness pits trap air, and air is a worse conductor than any thermal compound.
  • Hole position for the retention screws, since a sink that rocks on two screws has an air gap under the third.

SkillsProto inspects these three on every heat sink order and reports the measured values with the parts, rather than leaving the customer to assume them.

Getting a machined heat sink quoted

Send a STEP file with the fin count, fin thickness, base thickness and mounting hole pattern, plus the heat load, the available airflow and the space envelope. DFM feedback on fin aspect ratio and manufacturability comes back within 24 hours, with a priced quote against the same file.

SkillsProto machines heat sinks in 6061, 6063 and copper from MOQ 1, ships DDP to the UK, EU and North America in 3-5 days, and holds ±0.005 mm on critical features. Start at /quote/ with a model and the thermal numbers, and the geometry question is usually settled in the first reply.

Frequently Asked Questions

Can CNC machining produce pin fin heat sinks without tooling?

Yes. A pin fin array is a pocketing operation, so the pins are what remains after the field between them is cleared. There is no tooling charge and MOQ is one piece, which makes CNC the standard route for pin fin prototypes. The practical limits are pitch — about 1.5 mm minimum before the tool shank fouls neighbouring pins — and fin height to thickness ratio of about 8:1.

Which fin geometry has the lowest thermal resistance?

Per unit of base area, skived copper fins reach the lowest thermal resistance because they achieve 10-20 fins per 10 mm with no bond line between fin and base. Among machined geometries, pin fins usually beat straight fins in unducted airflow, and straight fins win in a ducted channel because they carry less pressure drop. The ranking changes with airflow, so it should be checked against the fan curve rather than assumed.

How thin can a machined aluminium fin be?

0.5 mm is achievable in aluminium, but 0.8 mm is the practical floor before cost rises faster than performance. Below 0.8 mm the fin deflects under cutting pressure, the shop has to take lighter passes, and the extra machining time costs more than the additional surface area saves.

Is copper worth it over aluminium for a machined heat sink?

Copper conducts 401 W/m·K against 167 W/m·K for 6061, which is 2.4x, but it is 3-4x denser and 4-6x more expensive as stock, and it machines more slowly. Copper pays when the base is small relative to the heat load and spreading resistance dominates. If the base is large and the fins are the limiting factor, aluminium with more fin area is usually cheaper per degree.

What is the minimum order quantity for a CNC machined heat sink?

One piece. SkillsProto has no tooling to amortise on a machined sink, so a single prototype and a 500-piece run are quoted from the same setup. That is the main structural difference against skived or extruded sinks, where the die cost has to be spread over volume.

How long does a machined heat sink take to deliver?

3-7 days for standard aluminium grades, with a 48-hour expedite available on simple geometry, plus DDP shipping in 3-5 days. Copper is the variable: if the plate is a mill order rather than stock, add 1-2 weeks. Quotations are issued within 24 hours of receiving a STEP file and the thermal requirements.

Related capabilities

Related guides

Have parts to quote?

Upload your files — DFM-checked quotation within 24 hours.

获取报价