Aluminum Base PCB With 3oz Copper: 6 Extra RFQ Inputs | PCB168

The path to this point is usually the same. A power device runs hot, so you pick an aluminum base. Then current or temperature rise pushes the copper up, and the RFQ goes out as a single line: "Aluminum base PCB, 3oz copper."

That line often cannot be quoted as written. Not because it is too simple, but because it lands on two product families at once, and PCB168 publishes their specs separately. The aluminum PCB page lists 1–2 layers, thermal conductivity 1.0–3.0 W/mK, aluminum thickness 0.8–3.0mm, copper weight 1–10oz, board size up to 1200×600mm. The heavy copper page lists copper weight 3–20oz, 1–16 layers, board thickness 0.8–6.0mm, minimum line width 6mil.

What is buildable is the intersection of those two pages, not the union. Copper weight 3–10oz sits inside the intersection. The heavy copper page's 1–16 layers does not carry over to an aluminum base, which is published as 1–2 layers. The two board size figures are not the same number either. That is why one line comes back as a clarification email.

Boundaries first. How to declare a general aluminum thermal path in an RFQ — who owns the conductivity figure, how to state aluminum thickness, which numbers cannot be given — is already covered in aluminum PCB RFQ inputs: declaring the thermal path. That article is upstream of this one and is not repeated here. How copper weight itself gets chosen from current and temperature rise is covered in current and thermal budget to copper weight. This article assumes both are done and answers a narrower question: what your RFQ has to add when heavy copper lands on an aluminum base.

1. Correct one directional assumption first: heavy copper on aluminum is not better cooling

This is the most common misread, and every input below follows from getting it right.

On standard FR-4, copper carries a meaningful share of lateral heat spreading, so "thicker copper helps cooling" is partly true. On a metal core, the dominant path turns vertical: device → pad and copper → dielectric → aluminum → external thermal interface.

The narrowest link in that chain is normally the dielectric. The published 1.0–3.0 W/mK belongs to the dielectric layer, not the copper and not the aluminum — both metals conduct one to two orders of magnitude better. Going from 2oz to 3oz or 4oz improves lateral spreading and current capacity; it does very little to widen the narrowest vertical opening.

So the defensible reasons to specify heavy copper on aluminum are these two, not "better heat dissipation":

• Current capacity: high-current traces, bus-bar style pours, multiple output rails merging.
• Lateral spreading: taking a small concentrated heat source and spreading it over more area before it enters the dielectric, which effectively enlarges the area conducting through that layer.

The second is a real gain, but its ceiling is set by the dielectric and by how the aluminum back mounts to something else. That gives us the first required input.

2. Input set one: a heat-source list, not a total wattage

"30W total" cannot be used to judge whether heavy copper helps, because the value of lateral spreading depends entirely on whether the heat is concentrated or distributed. What the RFQ needs is a list:

• Package, location and individual dissipation for each major heat source (a rough placement sketch helps, even a schematic one).
• Which sources run at full load simultaneously — staggered loads change the whole thermal picture.
• Maximum ambient temperature and whether forced air is present.
• Any target or limit on case or junction temperature. If you have one, give it. If you do not, say so explicitly rather than leaving a blank to be guessed at.

Those four determine two things: whether spreading via heavy copper is worth paying for, and which end of the 1.0–3.0 W/mK range the dielectric needs to come from. They are not a design outsourcing request; they are the minimum for engineering to judge feasibility.

When aluminum is the wrong answer altogether — more than two layers needed, high-frequency material required, or heat flux beyond what a metal core can take — is already tabulated in the upstream article and is not repeated here.

3. Input set two: how the aluminum back gets mounted

This set usually does not exist in an FR-4 RFQ, which is exactly why it gets skipped.

The point of a metal core is to hand heat off to something external: an enclosure, a heat sink, a chassis floor. If that handoff interface is never stated, the aluminum has only moved the heat to the back of the board and stopped there. It also determines real machining content. State:

• Whether the aluminum face mates directly to a heat sink or enclosure, and whether a thermal pad, grease or insulating sheet sits in between.
• Mounting holes, countersinks and screw locations, and whether they must be flush with the aluminum face.
• Any flatness or surface-condition requirement on the aluminum face (no protective film residue, deburring required, and so on).
• Whether the aluminum face needs insulation or anodizing — mandatory to state when dielectric strength or safety compliance is involved.
• Whether the aluminum needs slots, routing or local removal.

These are machinability questions. They move both price and lead time, and most of them cannot be inferred from Gerber data — Gerber describes copper and solder mask, not the mechanical and assembly intent of the back face.

On dielectric strength: that property belongs to the dielectric layer, and PCB168 does not publish a dielectric strength figure for its aluminum construction. So if your product carries a safety isolation or withstand-voltage requirement, state it as an input and ask engineering to confirm rather than assuming a value. Heavy copper does not help here. Thicker copper does not make the dielectric hold off more volts.

4. Input set three: above 3oz, line width and spacing have to open up

This is the item in the aluminum-plus-heavy-copper combination that most often surfaces after production has started.

The thicker the copper, the harder it is to hold vertical sidewalls through etch, and the wider the achievable minimum line width and spacing become. PCB168's heavy copper page publishes a minimum line width of 6mil, while the capabilities page publishes a general minimum of 2mil/2mil. These do not contradict each other — they describe different process conditions. 2mil is the fine-line limit; 6mil is the heavy copper figure. The minimum line width for aluminum construction specifically is not published separately, so it has to be confirmed with engineering.

Your job is not to guess a value. It is to state the demand side in the RFQ:

• The minimum line width and spacing you actually need, and which nets they appear on.
• Whether those narrow traces are signal or power — signal can often give way, power cannot.
• Whether any nets need controlled impedance. If so, note that impedance control generally requires a multilayer reference structure, which collides head-on with the 1–2 layer limit on aluminum (see the next section).
• The concession you can accept: widen the traces, reduce the copper weight, or change the construction.

That last point matters most. Write down which direction you can give ground in, and a supplier can return an executable proposal in one round. Leave it out, and they have to ask, which costs you a round trip.

Whether "3oz" means finished copper thickness or base copper is a separate high-frequency ambiguity, covered in finished copper thickness vs base copper on the PO. Do resolve it in the RFQ though — it changes the actual etch condition and the achievable line width.

5. Input set four: layers and total thickness, where a hard boundary sits

The aluminum page publishes 1–2 layers. That boundary is firmer than most buyers expect, and combining it with heavy copper makes the budget tighter:

• If routing needs three or more layers, aluminum construction largely stops being applicable and the structure has to be reopened — conventional multilayer with a mechanical thermal solution, or another metal core or ceramic option (the capabilities page lists metal core as Aluminum/Copper and ceramic as Al₂O₃/AlN).
• Total thickness has to be budgeted together with the aluminum. Aluminum thickness of 0.8–3.0mm consumes most of the stack; add 3oz-plus copper and the dielectric, and the finished board runs noticeably thicker than a comparable FR-4 part.
• If your mechanical envelope caps total thickness, that cap belongs in the RFQ. Otherwise it will likely conflict with the aluminum thickness you want, and the conflict surfaces only after prototypes exist.

How layer count and board thickness constrain each other on thick copper multilayers is its own topic, covered in what limits layer count on thick copper multilayers. Here we only use the 1–2 layer boundary and do not recompute multilayer ceilings.

Board size deserves the same care. The aluminum page says up to 1200×600mm; the capabilities page says 600×1100mm. Both are published, under different conditions. Do not pick the larger one yourself. Give your actual outline dimensions and let engineering confirm which figure applies.

6. Input set five: assembly conditions, reflow above all

A heavy copper board is a heat sink during reflow: more copper means more energy to bring up to temperature, and a slower ramp. An aluminum base amplifies that — the whole point of the metal core is to move heat away, and it does that just as diligently inside a reflow oven.

Stacked together, those two properties genuinely shift the soldering window. So the RFQ needs:

• Whether you assemble it or need PCBA/turnkey. If the latter, the board's thermal mass has to travel with it as an assembly input; scope boundaries are covered in PCB assembly contract manufacturing scope.
• Lead-free or leaded process, and expected reflow cycles including rework.
• Whether wave solder, hand solder or press-fit operations are involved.
• Whether large exposed copper areas or oversized pads need specific mask or opening treatment.

Surface finish selection (the aluminum page publishes HASL/ENIG/OSP) is fundamentally an assembly decision rather than a cosmetic one, argued in full in the upstream article and not repeated here. One reminder only: if you plan large pads on heavy copper with HASL, raise the coplanarity question in the RFQ rather than discovering it at first article.

7. Input set six: fix the acceptance basis at quotation, not at receipt

The most common dispute on aluminum-plus-heavy-copper parts is that the boards look fine but the thermal performance falls short. The cause is usually the dielectric or the interface, not a visible cosmetic attribute.

So push acceptance onto items that are measurable and judgeable, at quotation time:

• What first article (FAI) must include: cross-section (copper thickness, dielectric thickness, layer structure), outline and mounting hole dimensions, aluminum face condition, 100% electrical test report. The capabilities page publishes AOI on every layer, 100% electrical test (flying probe or fixture), TDR impedance, cross-section, X-ray and ionic contamination testing — name the ones you need. Do not write "per your standard."
• What routine lots must include: usually electrical test and visual. Cross-section normally belongs to first article or sampling rather than every lot — but write that down instead of assuming it.
• Which specification and revision the criteria come from: IPC-A-600 for visual acceptance, IPC-6012 for performance and structural criteria, IPC-2221 for general design basis, IPC-2152 as the industry basis for current-carrying and temperature rise. All of these are industry specifications, not PCB168 internal control values, and the revision referenced on your drawing governs.
• What does not constitute rejection: normal etch sidewall profile on heavy copper, mask coverage behaviour at heavy copper edges. Agree on these at quotation, or they become an incoming-inspection argument.

The general eight-step framework for a heavy copper purchase specification is its own article and is not rewritten here — see heavy copper PCB purchase specification and acceptance. To connect this evidence to the receiving dock, see incoming inspection and acceptance planning. And because contradictions between drawing, specification and Gerber are the second-largest source of clarification emails, run the package through the fabrication notes conflict checklist before sending.

8. A copyable RFQ block: aluminum base plus 3oz and above

The six sets compressed into something you can paste into an email. Fill the brackets and delete no lines — writing "none" beats leaving a blank, because a blank reads as unknown and triggers a clarification.

```

Construction: aluminum base (metal core), [1 / 2] layers

Copper weight: [3 / 4 / 6 …] oz, as [finished copper / base copper] - please confirm achievable range

Aluminum thickness: [target within 0.8-3.0mm]; total finished thickness cap: [mm]

Outline: [L x W mm] - please confirm which published board size limit applies

Heat sources: [device 1 package/location/dissipation; device 2 ...]; simultaneous full load: [yes/no]

Environment: max ambient [C]; forced air: [yes/no]; case or junction temp target: [value / none]

Dielectric: conductivity needed toward [higher end / standard]; dielectric strength requirement:

[value / none] - engineering to confirm

Aluminum-face mounting: to heat sink [direct / via thermal pad / none]; mounting holes: [yes/no, see drawing];

special requirements: [flatness / insulation / anodize / routing / none]

Line width: minimum width/spacing actually needed [mil], on [power / signal] nets;

acceptable concession: [widen traces / reduce copper weight / change construction]

Surface finish: [HASL / ENIG / OSP]; assembly: [lead-free / leaded], reflow cycles [n] incl. rework

Assembly scope: [bare board only / with PCBA]

Acceptance: FAI requires [cross-section (copper + dielectric thickness) / outline and holes /

aluminum face condition / 100% E-test]; routine lots require [E-test + visual];

criteria per [IPC-A-600 / IPC-6012 / IPC-2221 / IPC-2152] revision [ ]

Change control: re-quote and written confirmation required for any change to copper weight,

aluminum thickness, dielectric, outline, layer count or surface finish

```

9. One sentence

Heavy copper on an aluminum base buys current capacity and lateral spreading, not vertical cooling — so what your RFQ has to add is not a larger copper figure but six input sets: the heat-source list, the aluminum-face mounting interface, your acceptable line-width concession, the total thickness budget, the assembly thermal conditions, and named acceptance items. The intersection of two spec pages is the buildable window; anything outside it needs the construction reopened at quotation, not after prototypes.

Once those inputs exist, send them with your Gerber through the quote request, or check the published ranges first on the aluminum PCB page.

FAQ

Q1: Can an aluminum base PCB be built with 3oz copper or more?

Yes. PCB168's aluminum PCB page publishes a copper weight range of 1–10oz, so 3oz, 4oz and 6oz all sit inside the published window. Watch the layer boundary though: aluminum construction is published as 1–2 layers, and the heavy copper page's 1–16 layers does not apply to it. The minimum line width for aluminum construction specifically is not published and has to be confirmed with engineering.

Q2: Does thicker copper on an aluminum PCB improve heat dissipation?

Separate the directions. On a metal core the dominant path is vertical (device → copper → dielectric → aluminum → external interface), and the narrowest link is normally the dielectric. Thicker copper improves current capacity and lateral spreading — taking a concentrated source and spreading it so more area conducts through the dielectric. It does not widen the vertical bottleneck. Buying copper weight for "better cooling" usually buys the wrong thing.

Q3: Is 1.0–3.0 W/mK the thermal conductivity of the aluminum?

No. That is the published figure for the dielectric layer. Copper and aluminum both conduct one to two orders of magnitude better, which is precisely why the dielectric is the narrowest link in the chain. It is also why the RFQ should say which end of that range you need.

Q4: Why does "aluminum base PCB, 3oz copper" come back as a clarification request?

Because that line lands on two separately published product families, and the buildable window is their intersection, not their union — layer count is bound by the aluminum side's 1–2 layers, and the two pages publish different board size figures. It also omits inputs only you can supply: heat-source distribution, the aluminum-face mounting interface, dielectric strength requirements, the line width you actually need, the total thickness cap, and the assembly process. Without those, engineering cannot judge feasibility and has to ask.

Q5: What is the dielectric strength of PCB168's aluminum construction?

It is not published, and this article does not supply a number. Dielectric strength belongs to the dielectric layer, and thicker copper does not raise it. If your product carries a safety isolation or withstand-voltage requirement, state it as an RFQ input and ask engineering to confirm achievability in writing.

Q6: If total thickness is over budget, do I reduce aluminum thickness or copper weight first?

It depends on which side your bottleneck is on, which is exactly why the RFQ should state an acceptable concession. Roughly: if heat leaves mainly through the vertical path, protect the aluminum and dielectric first; if the bottleneck is current, protect the copper and consider widening traces. That trade needs the heat-source list to decide, so do not leave the supplier guessing — write both candidate options and let engineering answer which is feasible.