Aluminum PCB RFQ Inputs: Declaring the Thermal Path So It Can Be Quoted

1. Why "good heat dissipation" cannot be quoted

A quotation is built from process steps and material selections. Each step needs a value.

"Dissipates heat well" maps to no step. Worse, it hides a decision the buyer has already

made implicitly and not communicated: whether the heat leaves through the board thickness

into a chassis, or spreads sideways through copper into a larger area.

Those two routes buy different things and cost different money. A metal-core construction is

a vertical-path solution: heat crosses a thin dielectric into an aluminum carrier and then

into whatever the aluminum is bolted to. Copper on the surface is a lateral-path and

current-carrying solution. If the RFQ does not say which route the design depends on, the

supplier has to guess, and a guess becomes either an overbuilt price or a board that does

not perform.

So the first job of the RFQ is not to specify heat. It is to specify **construction plus four

numbers**, and to state what the board is bolted to.

2. The four inputs you can put a number on

PCB168 publishes these as aluminum PCB capability ranges on the

Aluminum PCB product page. Every value below is quoted from

that page, and the range is the boundary within which you are free to specify.

Four numbers and one selection. That is a quotable RFQ. Note what is not in the table:

there is no thermal resistance figure, no dielectric layer thickness, no bond strength and no

warpage limit, because PCB168 does not publish those as capability values. Asking a supplier

to commit to an unpublished number at quotation stage produces either a refusal or a number

nobody can hold them to. Those belong in engineering review, covered in section 8.

Also worth stating plainly in the enquiry: the mounting condition. An aluminum carrier that

is not thermally coupled to a chassis, heatsink or enclosure is a heat store, not a heat

path. This is a design input, not a fabrication parameter, but it changes which of the four

numbers actually matter, and a supplier reviewing your files will ask about it anyway.

3. The most common misreading: 1.0–3.0 W/mK belongs to the dielectric

This deserves its own section because it causes real requoting.

The 1.0–3.0 W/mK range published on the aluminum PCB page is the **thermal conductivity of

the dielectric layer** that sits between the copper circuit and the aluminum carrier. It is

not the conductivity of the aluminum, and it is not a thermal resistance for the finished

board.

Three consequences follow, and all three belong in your RFQ thinking:

The dielectric is the bottleneck, so its specification matters most. Aluminum conducts

heat far better than any laminate dielectric. The thin insulating layer is therefore the

restricting element in the vertical path, which is exactly why it is the number you declare.

Moving from the low end toward the high end of the published range buys you improvement at

the restricting element.

A W/mK figure is a material property, not a board-level result. Board-level performance

depends on the dielectric thickness, the heat source footprint, the copper area, the aluminum

thickness and the mounting condition. Requesting "3.0 W/mK" does not entitle you to a

thermal resistance figure. If your design needs a validated board-level number, that is a

thermal test or simulation activity to agree separately, with the test method and boundary

conditions written down — not a line item you can obtain from a fabrication quote.

Do not specify a dielectric thickness you have not confirmed. The dielectric layer

thickness is not published as a PCB168 capability range, and it interacts with both thermal

performance and dielectric withstand voltage. State the conductivity you need and the

isolation requirement your application imposes, and mark thickness as subject to engineering

review.

4. Copper weight on a metal-core board buys current, not vertical cooling

The GSC queries that led to this article are explicit about copper: buyers are searching

aluminum base pcb 3oz copper, high thermal aluminum pcb 2oz copper and

3oz copper based power pcb. So the copper number is already on their minds. What it does

is often misread.

On a metal-core board, surface copper mainly does two things:

1. Carries current. Cross-sectional area sets the conductor's own heating. Public sizing

guidance for conductor width against current lives in IPC-2152; cite the standard as the

basis for your calculation rather than inventing a rule of thumb.

1. Spreads heat sideways. Thicker copper widens the effective area feeding the dielectric,

which helps a small hot component present a larger footprint to the vertical path.

What it does not do is bypass the dielectric. Vertical heat flow still crosses the

insulating layer, so adding copper weight does not substitute for declaring the dielectric

conductivity. Two boards with identical copper and different dielectric grades behave

differently.

Copper also has a cost in geometry. Heavier copper needs wider conductors and larger spacing

than light copper. If your layout already has fine routing, specifying high copper weight can

make the layout unmanufacturable as drawn — you may get the copper and lose the routing. Ask

the question in the enquiry rather than discovering it in DFM feedback.

The boundary against heavy copper FR-4. PCB168's

Heavy Copper PCB page publishes 3oz–20oz copper on 1–16

layers, 0.8mm–6.0mm board thickness and 6mil minimum line width. The aluminum PCB page

publishes 1oz–10oz copper on 1–2 layers. These are two different answers to two different

problems:

• If the requirement is high current in the copper and you need more than two circuit

layers, heavy copper FR-4 is the construction to price.

• If the requirement is getting heat out of the board into a chassis and the circuit fits

in one or two layers, the metal-core build is the construction to price.

Boards needing both high current and many layers are a design conversation, not an RFQ line.

Send both requirements and ask which construction the supplier would review, rather than

specifying a hybrid you have not confirmed exists.

5. Two published size figures, and why you must ask which applies

This is a documentation trap worth naming, because it is visible on the site itself.

• The Aluminum PCB page states board size up to 1200×600mm.
• The Manufacturing Capabilities page states maximum board size

600mm × 1100mm for standard PCB capability.

Those are two different statements on two different pages, with different scopes: one is the

aluminum product entry, the other is the general standard-PCB capability table. They are not

interchangeable, and a buyer should not resolve the difference by picking whichever is more

convenient.

The correct RFQ behaviour is to state your actual panel or board dimensions and ask for

written confirmation of the buildable size for your specific construction. Actual

achievable size depends on the construction, the aluminum thickness, tooling and handling, and

must be confirmed in engineering review. If a supplier confirms a size verbally, get it into

the quotation text — size is one of the cheapest things to confirm before tooling and one of

the most expensive to discover afterwards.

The same discipline applies to copper weight, incidentally. The capabilities page publishes

0.5oz–20oz across standard PCB work; the aluminum page publishes 1oz–10oz. For a metal-core

enquiry, the aluminum page range is the one to work from, and anything beyond it is a

feasibility question rather than a specification.

6. Surface finish is an assembly decision, not a cosmetic one

The aluminum PCB page publishes HASL, ENIG and OSP. All three are available, so this is a

selection you make rather than a range you specify — and the basis for the selection sits on

the assembly side, not the fabrication side.

What to weigh, in RFQ terms:

• Component mix and pitch. Fine-pitch and area-array parts are sensitive to surface

planarity. If your placement includes fine-pitch devices, say so in the enquiry so the

finish choice is reviewed against the actual footprint set rather than assumed.

• Assembly process and number of thermal cycles. How many reflow passes the board sees,

and whether there is hand soldering or rework, affects which finish behaves predictably.

• Storage and handling window. Finishes differ in how long a board can sit before

assembly. If your build is staged weeks after fabrication, that belongs in the enquiry.

• Who assembles the board. If PCB168 also quotes assembly, the finish should be reviewed

as part of one package. The

PCB Assembly page states that quotation input is BOM plus

fabrication and placement data, and that assembly scope is confirmed by engineering review —

so send all three file sets together rather than fabricating first and asking later.

For a wider comparison of finish behaviour, the published

PCB surface finish comparison covers the

trade-offs. Choose from that basis, then declare one finish in the RFQ rather than writing

"HASL or ENIG" and leaving the supplier to pick.

7. When aluminum is not the answer

Three published limits do the work here, and all three come from the ranges already quoted.

You need more than two circuit layers. The aluminum PCB page publishes 1–2 layers. If your

routing genuinely needs internal layers, a metal-core build is not the construction. The

Manufacturing Capabilities page publishes 1–40 layers for standard PCB

work; a multilayer FR-4 board with a deliberate thermal strategy is the alternative to price.

You need controlled impedance. Controlled impedance requires a defined dielectric stackup

with reference planes. The capabilities page publishes ±5% impedance control with TDR

verification for standard PCB construction. A 1–2 layer metal-core board does not give you

the stackup that specification depends on. If the design carries impedance-sensitive nets,

you are looking at a different construction, or at splitting the function across two boards.

Your heat is spread rather than concentrated, and nothing is bolted to a chassis. The

vertical path only pays off if the aluminum couples into a larger thermal mass. Where the

mechanical design has no such coupling, the money may be better spent on design-side measures.

The published guide on

PCB thermal management design covers those

design-side options; use it to decide before writing the enquiry, not after.

A fourth, softer case: cost. If the thermal load is modest and the design is comfortable on

standard laminate with copper area and via structures, that is usually the cheaper route.

Metal-core is a specific solution to a specific problem, not a general upgrade.

8. The RFQ input list you can copy

Two blocks. The first is filled in with numbers by you; the second is deliberately left open

for engineering review. Sending both is what separates a quotable enquiry from a question

generator.

Block A — declare a value (within the published ranges above):

1. Construction: aluminum / metal-core, single or double sided
2. Layer count: 1 or 2
3. Dielectric thermal conductivity required, within 1.0–3.0 W/mK
4. Aluminum thickness, within 0.8mm–3.0mm
5. Copper weight, within 1oz–10oz — state whether it is driven by current or by spreading
6. Board outline dimensions, plus panel preference if you have one
7. Surface finish: one of HASL / ENIG / OSP
8. Quantity for this build, and the next expected quantity if there is one
9. Isolation / withstand voltage requirement your application imposes
10. Mounting condition: what the aluminum face is fixed to
11. Acceptance basis you intend to apply, by standard name (for example IPC-A-600 class, or

IPC-6012 for performance requirements) — state the class you require

Block B — mark as "to be confirmed in engineering review", do not invent numbers:

1. Dielectric layer thickness
2. Board-level thermal resistance, if you need one validated — with test method to be agreed
3. Flatness and warpage acceptance
4. Buildable maximum size for this specific construction (see the two published size figures

in section 5)

1. Feasibility of any value outside the published ranges
2. Depaneling or routing method, if the outline is non-trivial
3. Whether assembly is in scope, and if so the inspection and test plan

Send Block A as values, Block B as questions. Then attach the fabrication data. When you are

ready, the quote request form is the intake, and

engineering contact is where the Block B items get resolved.

One habit worth adopting: keep Block B in the file even after the answers arrive, with the

answers written next to the questions. That record is what protects you at the second order,

when the person who ran the first enquiry has moved on.

9. What "reviewed" means on the supplier side

Worth knowing so you can judge the reply you get. The

Manufacturing Capabilities page lists AOI on every layer, 100% electrical

testing by flying probe or fixture, microsection analysis, X-ray inspection and ionic

contamination testing as quality control activities. Those are the verification steps

available to be written into a project's inspection plan.

What that means for your enquiry: the inspection plan is something you can ask to have stated,

rather than assumed. For a metal-core board the questions worth asking explicitly are which

electrical test method applies to your board, and what the acceptance basis is for the

dielectric isolation. Both are project-specific rather than published defaults. The published

quality inspection methods guide explains what

each method can and cannot detect, which helps you ask for the right one instead of asking

for all of them.

FAQ

Is the 1.0–3.0 W/mK figure the conductivity of the aluminum?

No. It is the published thermal conductivity range of the dielectric layer between the copper

circuit and the aluminum carrier, from the aluminum PCB product page. Aluminum itself conducts

far better; the dielectric is the restricting element in the vertical path, which is why it is

the value you declare in the RFQ.

How many copper ounces should I specify on an aluminum PCB?

Specify what your current and lateral spreading needs require, within the published 1oz–10oz

range. Size conductor width against current using IPC-2152 as the basis rather than a rule of

thumb, and be aware that heavier copper needs wider conductors and larger spacing, which can

conflict with dense routing. If both high current and more than two layers are needed, ask

whether heavy copper FR-4 (3oz–20oz on 1–16 layers, per the heavy copper page) is the better

construction.

Can I get a thermal resistance figure in the quotation?

Not as a fabrication line item. Thermal resistance is a board-level result that depends on

dielectric thickness, heat source footprint, copper area, aluminum thickness and mounting

condition. If your design needs a validated figure, agree it as a separate thermal test or

simulation activity with the method and boundary conditions written down.

Which maximum board size applies to my aluminum board — 1200×600mm or 600×1100mm?

Those are two separate published statements: 1200×600mm on the aluminum PCB page, and

600mm × 1100mm in the standard PCB capability table on the capabilities page. Do not resolve

the difference yourself. State your actual dimensions and ask for written confirmation of the

buildable size for your specific construction; actual achievable size must be confirmed in

engineering review.

Can an aluminum PCB have four layers?

The aluminum PCB page publishes 1–2 layers. If the design needs internal routing layers, the

metal-core construction is not the right one to price — look at multilayer FR-4, where the

capabilities page publishes 1–40 layers, and treat the thermal requirement as a design-side

problem instead.

What do I send with the enquiry?

Fabrication data plus the Block A values in section 8, plus Block B as explicit open

questions. If assembly is also in scope, the assembly page states that quotation input is BOM

plus fabrication and placement data, with scope confirmed by engineering review — so send all

three file sets in aligned revisions rather than in sequence.