Heavy Copper PCB Via Current Capacity: Three RFQ Inputs Copper Weight Does Not Cover

You specified 6 oz finished copper. The fabricator quoted it, accepted it, and built it. Then

the board runs warmer than the thermal model said it would, and the warm spot is not on a trace.

It is at a layer transition.

Copper weight is an in-plane specification. It describes how much copper sits on the surface

of a layer, so it governs how much current a trace on that layer can carry. The moment current

has to move between layers, it stops using that copper entirely and starts using a different

conductor: the thin plated barrel on the inside of a hole. That barrel is not made by the copper

weight you ordered. It is made by a plating process whose thickness, hole count, and fill state

were decided by the fabricator's defaults, because nothing in your RFQ said otherwise.

That is the gap this article closes. Not "how many amps does a via carry" — that answer belongs

to your electrical engineer and your acceptance standard. The question here is procurement's:

**which inputs must appear in the RFQ so that the layer transition is a specified, quotable,

verifiable item rather than a default.**

Why "copper weight equals current capacity" stops being true at a layer change

A heavy copper order is usually written as a single number plus a unit: copper weight, in ounces

or microns, sometimes with a note about whether it means base copper or finished copper. That

single number does real work. It sets the in-plane cross-section, it drives the etch and plating

effort, it drives the laminate and press decisions, and it is the number both sides argue about

when a quote looks high.

What it does not do is follow the current through the board.

Think of the current path in a power board as three segments in series:

1. In-plane segment on the entry layer. Governed by your copper weight and trace geometry.
2. The transition segment. Current leaves the plane and travels through plated holes — the

barrel wall, plus whatever the hole is filled with, if anything.

1. In-plane segment on the exit layer. Governed by copper weight again.

Segments 1 and 3 are specified by your order. Segment 2 is specified by nobody, unless you do it.

And segment 2 is the narrowest of the three by a wide margin: a barrel wall is a plated film on

a cylinder, not a slab of rolled copper. A series circuit is limited by its worst segment. You can

order the heaviest copper in the catalogue for segments 1 and 3 and still have the board's real

limit sitting in an unspecified segment 2.

This is why thermal surprises in heavy copper boards tend to show up at transitions rather than

on traces. The traces are built to a number you chose. The transitions are built to a number

somebody else chose on your behalf.

The three inputs that are missing

Input 1: Barrel plating thickness, declared independently of copper weight

The plating that forms the hole wall is applied in a process step that also adds copper to the

outer surfaces. That coupling is the source of most of the confusion: buyers assume that because

plating adds to the surface, specifying the surface total must also pin down the barrel. It does not,

in a way that is usable as a contract term. A surface copper figure can be reached by different

combinations of starting foil and plating, and those combinations do not all leave the same

thickness on a vertical hole wall deep inside the board.

So barrel plating thickness has to be stated as its own line item. Three things make a declaration

usable rather than decorative:

• It names the barrel, not the surface. "Minimum copper thickness in the plated hole wall" is

a different quantity from "finished copper on outer layers." If your RFQ only carries the second,

you have not specified the first.

• It is a minimum, not a target. Plating distributes unevenly along the depth of a hole. A

nominal value with no stated minimum is not auditable, because any single measurement can be

defended as being off-nominal. A minimum is a pass/fail line.

• It says where the minimum applies. Hole wall thickness varies with depth and with position

on the panel. A minimum with no location rule leaves the measurement point to whoever is

reporting it.

Where does the number itself come from? Not from this article, and not from your supplier's

marketing page. It comes from one of two places: your electrical/thermal analysis, which produces

a required conductor cross-section at the transition, or the acceptance standard and class your

program is built to, which specifies minimum plated hole copper for that class. Procurement's job

is not to invent the figure. It is to make sure the figure your engineering team already relies on

is written into the document the fabricator quotes against — and to notice when it is absent, which

is the common case.

Input 2: Via count, spacing, and arrangement — and who owns that decision

The transition rarely happens through one hole. It happens through a group of them, and the group's

behaviour is not simply "one hole times N."

Three properties of the group matter commercially, because each one is a thing the fabricator either

follows from your data or decides for you:

• Count. How many holes carry the transition. This is a design output: it falls out of the

required cross-section and the per-hole contribution. It belongs in your data, not in the

fabricator's judgment.

• Spacing and pattern. Holes packed close together share heat with each other and with the

copper between them. Spacing also interacts with manufacturability: a dense group in heavy copper

has different drilling and plating behaviour than the same group in thin copper. This is where

design intent and process capability meet, and where a DFM comment is legitimate.

• Connection to the surrounding copper. Whether each hole ties directly into the plane or

through a restricted connection changes how much copper actually feeds the transition, and it

changes assembly behaviour as well. In heavy copper this is not a detail; the surrounding copper

is massive, so how you connect to it dominates local thermal behaviour.

The boundary that keeps projects out of trouble is simple to state and routinely violated:

> **Count and electrical connection intent are buyer-owned design inputs. Spacing and pattern are

> buyer-proposed and fabricator-reviewed. Nothing in this group is fabricator-owned by default.**

When a buyer leaves the via group under-specified and the fabricator "optimizes" it — merges holes,

changes a connection style, adjusts a pattern to improve plating uniformity — the change can be

entirely reasonable from a manufacturing standpoint and still break the current path the designer

sized. That is why this group needs to travel with an explicit instruction about whether deviation

is permitted at all, and if so, through what approval. A fabrication note that says "via count and

connection style in the power transition areas are functional; no substitution without written

approval" costs nothing to add and removes the entire category of silent change.

Input 3: Fill state, declared as a function not a finish

"Filled vias" is a term that covers several different operations with different purposes, and the

purposes matter more than the names when you are writing an RFQ.

A hole may be left open. It may be filled with a non-conductive material and capped. It may be

plated to a greater or lesser degree of closure. These are not grades of the same thing on a single

quality scale — they serve different objectives, and the objective is what the fabricator needs

from you:

• If the objective is assembly-related — keeping solder from escaping, permitting a pad to sit

over the hole, keeping flux out — then what matters is the closure and the surface condition, and

the specification lives in the assembly and surface requirements.

• If the objective is current or heat at the transition, the relevant question is whether the

fill contributes conductively and whether it is being relied upon for that contribution.

• If the objective is mechanical or cosmetic, say so, because it changes nothing about the

conductor and should not be allowed to imply it does.

The failure mode in heavy copper RFQs is to carry "filled vias" as an unqualified line and let the

fabricator infer the purpose. If the fabricator infers "assembly" and your designer assumed

"conductive contribution," you get a board that passes every stated requirement and still does not

have the transition that was modelled. Declare the function and let the process follow.

What is actually provable at acceptance, and what is not

Buyers often try to solve segment 2 with an inspection clause. That works only for part of it, and

knowing which part saves a lot of argument later.

Measurable on the delivered product, destructively: barrel wall thickness is read from a cross

section. That means a coupon or a sacrificial board, which means it has to be in the order from the

start — you cannot ask for it after the fact without building more boards. A cross section tells you

what that hole, on that panel, in that position, looked like. It is strong evidence and it is a

sample, not a census.

Measurable non-destructively, indirectly: electrical continuity and resistance testing on the

finished board will catch gross plating failures and opens. It will not distinguish a barrel at the

low end of acceptable from one comfortably above it. Treat it as a screen, not a thickness

measurement.

Not provable from the delivered boards at all: that every hole on every panel in the lot met

the minimum. This is where buyers over-reach, and it is worth being blunt about it, because an

unachievable clause gets quietly ignored by both sides and then nobody is covered. Lot-wide

confidence does not come from inspecting boards. It comes from process evidence: plating process

controls, the fabricator's own coupon records for the lot, and the traceability that links the

coupon to your panels. If the transition is functionally critical, the thing to negotiate is

access to that evidence as a deliverable, not a stricter inspection clause on the boards.

So the acceptance package for a current-carrying transition has three parts, and they are different

in kind:

If a supplier resists the third row, that is useful information at quote stage rather than at

first article.

Turning this into RFQ fields

The point of all of the above is to produce a short list of things that go into the document,

not a longer conversation. Six fields cover it:

Two notes on using the table. First, every one of these is quotable — which means leaving them blank

does not make the quote simpler, it makes it unpriced. The work still happens; you just do not know

what was assumed. Second, the "who owns it" column is the part that prevents the expensive version

of this problem. A fabricator who changes a buyer-owned item without approval has made an error. A

fabricator who chose a default for an item the buyer never specified has not. Deciding ownership in

the RFQ is what converts a future dispute into a future change order.

Three questions to ask when the quote comes back

When you receive a heavy copper quote and want to know whether segment 2 was actually priced,

these three questions separate suppliers who read the requirement from suppliers who filled in a

template:

1. "What minimum barrel copper thickness did you quote to, and where is it measured?" A supplier

who quoted your stated minimum will repeat it back with a location. A supplier who quoted a

default will answer with a class name or a general statement. Both answers are informative.

1. **"If your DFM review wants to change via count, spacing, or connection style in the power areas,

what is the approval path?"** You are testing whether change control exists before you need it.

1. **"What evidence for the plated hole copper comes with the shipment, and what does it cost to add

a coupon cross section?"** If this is priced as an extra, you want it priced now, at quote stage,

not negotiated after a first article discussion.

None of these require the supplier to reveal process detail. They establish whether the transition

was a specified item in the quote or an assumption inside it.

Where this sits relative to the rest of your copper specification

This article deliberately covers one segment of the current path. Three adjacent decisions are

specified elsewhere, and keeping them separate keeps each one answerable:

• Whether your copper figure means base or finished copper is an in-plane question about the

number itself, and it is the most common source of mismatched expectations in heavy copper

purchase orders. See

finished copper thickness vs base copper in the PO.

• Whether the 4 oz class your design sits in carries process consequences is about the copper

weight tier, not the transition. See

the 4 oz copper standard and the heavy copper threshold.

• Whether all of this is contractually binding depends on it living in a document the quote

references. A specification sheet does that; an email thread does not. See

how a fabrication specification sheet binds the quote.

For the quantity and panel economics of a heavy copper order, see

order quantity and panel price breaks.

Capability context for this product family is on the

heavy copper PCB page, and when the six fields above are filled in

you can submit them through the quote form.

The short version

Copper weight buys you in-plane conductor. It buys you nothing at the layer transition, where the

conductor is a plated barrel produced by a process you have not specified. Three inputs fix that:

a minimum barrel thickness with a measurement location, a via group declared as functional with a

stated deviation path, and a fill state declared by function rather than by name. One more input

makes it checkable: the evidence you require, named and priced in the quote rather than requested

after the first article.

Fill those in and the layer transition becomes something you ordered. Leave them blank and it

remains something you received.

FAQ

Does specifying 6 oz copper also specify the copper inside the holes?

No. Copper weight is an in-plane specification describing copper on the surface of a layer. The

copper inside a hole is produced by a plating step and is a separate quantity. A surface figure can

be reached by different combinations of starting foil and plating, and those combinations do not

leave the same thickness on a vertical hole wall. If your document carries only the copper weight,

the barrel is running on the fabricator's default.

Who is supposed to decide how many vias carry a current transition — us or the fabricator?

You. Via count follows from the required conductor cross-section at the transition, which is an

electrical design output. Spacing and pattern are reasonable things for a fabricator to comment on

during DFM review, because they interact with drilling and plating behaviour. But nothing in the

via group should be fabricator-owned by default. Flag those vias as functional in the fabrication

notes and state whether deviation requires written approval.

Can we just write "filled vias" and let the supplier choose the method?

Only if you do not care what the fill does. "Filled" covers open-and-capped arrangements,

non-conductive fill, and various degrees of plated closure, and they serve different objectives —

assembly protection, conductive or thermal contribution, or mechanical and cosmetic needs. Declare

the function. If a designer assumed the fill contributes to the transition and the fabricator read

the line as an assembly requirement, the board can meet every stated requirement and still not be

the board that was modelled.

How do we verify barrel copper thickness on boards we receive?

Thickness itself is read from a cross section, which requires a coupon or a sacrificial board

declared in the order from the start. Electrical test on 100% of boards screens for gross plating

defects and opens but cannot distinguish a barrel at the low end of acceptable from one well above

it. Lot-wide confidence comes from process evidence — plating controls, the fabricator's coupon

records for the lot, and traceability linking those records to your panels — not from inspecting

more boards.

Is it enough to reference an acceptance class instead of stating a minimum?

Referencing a class is better than silence, and for many programs it is the right basis — but it is

only complete if both sides agree on which class, which revision, and which performance level

applies to the holes in question. If the transition is functionally critical, state the minimum

value explicitly alongside the class reference, and state where the minimum is measured. A class

name with no measurement location still leaves the measurement point to whoever is reporting it.

What should we do if the supplier will not quote a coupon cross section?

Treat it as information rather than a dealbreaker, and get it at quote stage rather than after a

first article. Some fabricators price coupons as an extra, some include them by class, some do not

offer them on small orders. What matters is knowing which before the PO, because adding a coupon

after the fact means building more boards. If the transition is critical and no evidence is

available at any price, that is a supplier-selection input, not a specification problem.

---