16-Layer Thick Copper PCB: Why Two Quotes Are Not Comparable Until the Stackup Is Locked

The price spread you are looking at may not be a price spread

You asked three suppliers for a 16-layer thick copper board. The quotes came back and the highest is far above the lowest. The instinct is to treat that gap as information about the suppliers: one is competitive, one is expensive, one is somewhere in between.

Before accepting that reading, check what the three of them were actually asked. If the request said "16 layers, thick copper, this outline, this quantity", then the three quotes are answers to three different questions. Each supplier had to decide what "thick copper on a 16-layer board" means in its own plant, and each decided in the direction that made the board easiest to build there.

That is not dishonesty. A quotation has to be based on a buildable stackup, and if the buyer did not supply one, the supplier constructs one. The construction is silent, it is not usually attached to the quote, and it is the single largest reason that two thick copper multilayer quotes for the same layer count land far apart.

So the gap in front of you is some mixture of two things: real differences in cost structure, and differences in the assumptions each supplier filled in. Until the second part is removed, you cannot see the first part at all. Choosing the lowest number at this stage is not choosing the cheapest supplier; it is choosing whichever supplier assumed the least demanding build.

This page is about removing that second part. It covers which variables are still open inside the phrase "16-layer thick copper", what each one does to the build, where each one is declared in the RFQ, and how to work backwards from quotes you have already received to find out what was assumed.

Why the layer count is the least informative thing you told them

Layer count feels like the main specification because it is the number that appears in the part description and in the search box. For cost purposes it is close to the least informative figure in the request.

A layer count tells the supplier how many conductive layers exist. It says nothing about:

• what thickness of copper each of those layers carries
• whether those thicknesses are the same on every layer
• what sits between the layers, and how thick the finished board becomes
• how many separate lamination operations are needed to reach that arrangement
• how much copper area each layer actually holds

Every item on that list changes the process route. Several of them change the number of passes through the most capacity-constrained operations in the plant. Two boards that are both "16-layer thick copper" can therefore sit in genuinely different cost brackets without either supplier having done anything unusual.

This is the structural reason thick copper multilayer sourcing behaves differently from standard multilayer sourcing. On a conventional board, the stackup is largely implied once you give layer count, finished thickness and material class, because the copper is thin enough that the remaining choices are small. Add thick copper and the copper itself starts to dominate the geometry: it raises the height to be filled between layers, it affects what the lamination has to accomplish, and it interacts with how the layer is patterned. The stackup stops being an implementation detail and becomes the thing being priced.

Related background on where the layer-count ceiling itself comes from is covered separately in thick copper multilayer layer count limits; this page assumes the layer count is already settled and deals only with what is still open after that.

Variable one: which layers carry the thick copper

"Thick copper" in a request is usually read by the buyer as a property of the board and by the supplier as a property of specific layers. Those are different statements.

If every layer in the stack carries heavy copper, the board is one thing. If only the two outer layers do, or only a pair of internal power planes, it is another thing entirely — different lamination behaviour, different patterning difficulty, different finished thickness. A mixed-weight stack, where some layers are heavy and others are ordinary, is extremely common in power designs and is also the arrangement most often left unstated.

So the first thing to lock is not a thickness value but a map: a layer-by-layer assignment of which copper weight belongs to which layer. Without the map, a supplier reading "16 layers, thick copper" has three reasonable readings available, and the cheapest of the three is the one it will quote.

The decision test for this variable: *if the supplier returned a board where only the outer layers were heavy, would your paperwork say that was wrong?* If not, you have not specified this.

The mechanics of writing that map into a request, and what else has to travel with it, are covered in heavy copper multilayer mixed-weight stackup RFQ inputs. For this page the point is narrower: an unmapped stack makes every quote against it non-comparable, no matter how carefully the rest of the request was written.

Variable two: finished copper or base copper

This one produces more unmatched quotes than any other single item, because both readings are legitimate and the words used in a request rarely distinguish them.

Copper on a finished board is not the copper that arrived on the laminate. Processing adds material in some places and the two figures are not interchangeable. A request that states a copper weight without saying which figure it refers to has given the supplier a choice, and the two choices do not describe the same board.

The consequence is asymmetric, which is what makes it dangerous. If you meant finished and the supplier assumed base, the board arrives lighter than intended and the quote was for less work than you wanted. If you meant base and the supplier assumed finished, the quote is for more work than you needed and you may reject a supplier for being expensive when it was simply reading you correctly.

The decision test: *does the request contain the word "finished" or "base" anywhere next to the copper figure?* If not, this variable is open, and it is open on every quote you have received.

How to phrase it so that it survives into the purchase order rather than living only in an email is covered in finished copper thickness vs base copper in the PO declaration.

Variable three: dielectric arrangement and finished board thickness

Heavy copper changes what the dielectric has to do. Thicker conductors leave more height to be filled, and the material between the layers has to accommodate that while still arriving at whatever overall thickness the board needs.

That creates a chain: copper map → dielectric arrangement → finished thickness. Fix any two and the third is constrained. State only one and the supplier fixes the others.

Buyers often supply the finished thickness and nothing else, assuming it pins the stackup down. It does not — many internal arrangements reach the same overall thickness, and they are not equal in cost or in electrical behaviour. Conversely, buyers who supply a copper map but no thickness target leave the supplier free to arrive wherever its standard materials land, which may not fit the connector or the enclosure.

The decision test: *given only what you sent, could two people draw materially different internal arrangements and both be complying?* If yes, this variable is open.

Note what this page is not doing. It is not about how finished thickness is measured, where the gauge is placed, or what tolerance applies — that is a separate acceptance question. Here, finished thickness matters only as one of the inputs that makes two quotes refer to the same board. If your concern is the tolerance and how it gets written into the order, that is handled in finished thickness tolerance on high-layer thick copper boards in the PO.

Variable four: how many lamination operations the arrangement implies

A multilayer board is not pressed in one event unless its arrangement allows it. Certain combinations of layer grouping, copper distribution and via requirements force the stack to be built in stages, with a separate press operation for each stage.

From a cost perspective this is one of the heaviest variables in the list, because each additional operation consumes capacity on equipment that is usually the bottleneck, adds handling, and adds another point where the panel can be lost. From a buyer's perspective it is almost invisible, because nothing in the request mentions it. It is a consequence of the other variables rather than an independent choice.

That is precisely why it belongs on the list. A supplier that reads your request in a way requiring fewer stages will quote lower, and the difference will look like efficiency. It is not efficiency; it is a different board.

The decision test here is indirect: *have I constrained the inputs enough that the number of stages is determined, rather than chosen?* If the copper map, the dielectric arrangement and the via requirements are all declared, the staging follows and every supplier computes the same one. If any of the three is open, staging is open too, and you have no way to tell from the quote which path was assumed.

Variable five: copper area distribution across the layers

Two layers can carry the same copper weight and behave very differently depending on how much of the layer is actually copper and how it is distributed. A layer that is nearly a solid plane and a layer with the same weight but sparse routing are not equivalent to process.

This affects what has to happen during patterning and during pressing, and it can affect whether the arrangement is balanced enough to come out flat. Suppliers assess it from the data files, which means — unlike the earlier variables — this one is not something you declare so much as something you must *send*. A request that arrives without the actual layer data forces the supplier to assume a typical distribution, and typical is the one that prices best.

The decision test: *did the quote request include the layer data, or only a description of the board?* A description-only request cannot produce a quote that survives first article, because the supplier has priced a board it has not seen.

Where trace geometry interacts with copper weight is a different question, treated in trace width and spacing limits at a given copper weight. The relevance here is only that distribution is part of what makes a quote real.

Working backwards: recovering the assumption behind a quote you already hold

Most readers arrive at this page after the quotes are already in hand. Re-issuing the whole RFQ is one option, but it costs time and goodwill, and it is often unnecessary. The alternative is to recover each supplier's assumption from what it already sent you.

Three things to look for, in order:

First, look for a proposed stackup. Suppliers that quoted carefully usually attached or offered one. If it is there, the assumption is not hidden — read it against your five variables and you will see immediately whether two quotes describe the same board. If it is not there, ask for it before asking about price. The request is routine and reveals nothing about your budget.

Second, read the quote's own qualifiers. Phrases describing the copper as a nominal or standard value, notes that the price is subject to engineering review, or statements that the stackup will be confirmed after order are all declarations that the supplier filled a gap. They are telling you the quote is conditional. A quote with such a qualifier and a quote without one are not at the same level of commitment, independent of their numbers.

Third, compare the structure of the quotes rather than the totals. If one supplier separated tooling from unit price and another folded everything together, the totals cannot be compared directly even if the stackup assumptions happened to match. The components of a thick copper quote and what each responds to are broken down in heavy copper PCB quote comparison and cost drivers.

If the assumptions turn out to differ, the decision is practical: re-issue when the differences touch the copper map or the finished thickness, query item by item when they do not. Those two variables change the board's identity, so a corrected quote is a new quote and everyone should price the same corrected version. The rest — staging consequences, how distribution was read, how tooling was split out — can usually be closed with a specific question to each supplier without restarting the round. A general warning sign that a low quote rests on a favourable assumption rather than on capability is described in checking the build assumption behind a cheap quote.

Where each variable is declared, and why the location matters

Locking a variable in an email does not lock it. The five variables have natural homes in the document set, and putting each one where it belongs is what makes it survive from quotation into production and into incoming inspection.

Two of those rows deserve a note. The copper map belongs on a drawing rather than in prose because prose cannot express a layer-by-layer assignment without ambiguity. And the finished-versus-base statement has to be adjacent to the figure, not in a general notes paragraph elsewhere, because a reader checking one number will not go looking for a qualifier three pages away.

The broader question of which document controls when two of them disagree is a frequent source of disputes on thick copper orders. How to make the specification sheet itself binding on the quote is covered in making the fabrication specification sheet binding on the quote, and the conflicts that commonly appear between notes and drawings are listed in the fabrication notes conflict checklist.

How this differs from the heavy copper product page

If you arrived here looking for what a heavy copper multilayer board is, which copper weights are workable, or what a supplier needs to be able to do to build one, the heavy copper PCB page is the right destination. That page answers capability questions: what the process supports and what the board is for.

This page answers a narrower and later question. It assumes you have already decided you need a thick copper multilayer board, already have quotes in hand, and need to know why they disagree. The two are not alternative readings of the same query — one is about what can be built, the other is about whether two prices refer to the same thing.

The same split applies to the multi-layer PCB page and to the stackup guide: those describe how stackups are constructed, while this page is only about which parts of a stackup must be fixed before a price means anything.

When to lock, and when locking is premature

A reasonable objection: locking five variables requires engineering decisions that may not be made yet. If the design is still moving, declaring a copper map prematurely can force a rework of the request later.

That objection is valid, and the answer is to be explicit about which stage you are in rather than to send an underspecified request and hope.

• Design still open. Ask for a budgetary indication and label it as such, stating the assumptions *you* want the supplier to use. A quote against stated assumptions is comparable even when the assumptions are provisional, because all suppliers used the same ones.
• Design frozen, sourcing round open. Lock all five and run a real RFQ. This is the only stage at which the lowest number is meaningful.
• Supplier selected, details still settling. Changes now are change control, not clarification, and they carry a different cost profile. The terms that govern this are discussed in repeat order change control terms on heavy copper boards.

The failure mode worth avoiding is the middle one done badly: a frozen design sent out as an underspecified request, quoted on five different assumption sets, and awarded on price. That sequence selects for the supplier whose assumptions were furthest from your intent, and the gap surfaces at first article rather than at quotation. What a supplier should be asked to evidence before you get there is set out in requesting process evidence from a heavy copper supplier, and the acceptance side is covered in heavy copper purchase specification and acceptance.

When the five variables are locked and you want a round of pricing against them, the quote request form is the entry point.

FAQ

Is the lowest quote on a 16-layer thick copper board usually the one with the lightest assumptions?

Often, but not always, and the point is that you cannot tell from the number. A low quote can come from a plant that genuinely runs this class of board efficiently, or from a reading of your request that produced a simpler board. Both look identical on paper. Asking for the proposed stackup separates them in one step, which is why that request should come before any discussion of price.

Can I just ask every supplier to quote their standard 16-layer thick copper stackup?

You can, and it will give you comparable *prices* for non-comparable *boards*. Each supplier's standard arrangement reflects its own materials and equipment. If your design has no constraints beyond layer count, that approach is workable. If the board has a finished thickness it must hit, a copper assignment it needs, or an enclosure to fit, their standard is unlikely to match your requirement, and you will discover the mismatch after award.

Which of the five variables matters most if I only have time to fix one?

The copper map. It is the variable with the widest spread of reasonable interpretations, and it propagates: once the copper assignment is fixed, the dielectric arrangement and the staging are heavily constrained, and the remaining ambiguity narrows considerably. Fixing only the finished thickness, by contrast, constrains much less than buyers expect.

The supplier says the price is subject to engineering review. Is that a problem?

It is a signal, not a problem. It means something in the request needed a decision the supplier made provisionally. Treat it as an invitation to ask which item triggered the note. A quote carrying that qualifier is not comparable with one that does not, since the two represent different levels of commitment regardless of the amounts.

Does requesting the proposed stackup reveal that I am comparing suppliers?

Suppliers assume they are being compared on any competitive RFQ. Asking for the build they intend to use is a standard engineering request and tells them nothing about the other bids. The information flows in your direction, which is the reason to ask.

If two quotes turn out to assume the same stackup, is the spread now real?

Largely, though check how each quote is structured before concluding it. Tooling folded into unit price versus charged separately, and different quantity assumptions, can still make matched-stackup quotes incomparable. Tooling reuse across reorders in particular changes the arithmetic, which is covered in heavy copper reorder tooling reuse cost decisions.

Related links

• A wide production hall with a green floor, white walls and ceiling fluorescent lights, two rows of large white machines facing each other along a central aisle, blue and yellow rectangular plastic trays on the floor, green and red indicator lamps on the machines, and one person in the distance.