PCB Volume Production Cost: What Actually Drives the Price

At prototype quantity a bare board price is mostly tooling, setup and paperwork. At volume the same board is priced by what it consumes in material and machine time, and the levers that move the number change completely. This article breaks a volume quote into its parts and sets out where a cost reduction is real and where it simply moves risk elsewhere.

The Four Cost Blocks

A volume price for a bare board is made of four blocks, and knowing roughly how they divide is enough to have a useful conversation with a supplier.

Material. The laminate, the copper foil, the prepreg, the solder mask and the surface finish. On a straightforward four layer board this is often the largest single block, and it scales directly with the area consumed: not the area of the finished board, but the area of panel used to make it.

Process. Drilling, imaging, plating, etching, lamination and the finishing steps. This block scales with the number of process steps, the tolerance required at each step, and the machine time. A board with two layers and coarse features passes through far fewer expensive steps than a fine line multilayer board.

Yield. The cost of everything that was scrapped, spread across the units that shipped. Yield is invisible in a quotation and decisive in the final price, and it is where aggressive designs hurt: a feature that is achievable but marginal will pass the first article and then quietly consume margin for the whole production run.

Overhead and amortisation. Tooling, films, test fixtures, engineering time, inspection, documentation and the packaging. At volume this block is small per unit, but it is what makes a 100 piece order expensive per board and a 10,000 piece order cheap.

Panel Utilisation

The single largest lever on a volume price is how many boards fit on a panel. Fabricators buy laminate in standard panel sizes and sell board area, so a design that leaves 30 percent of the panel as scrap pays for that scrap on every unit produced.

Utilisation is affected by the board outline far more than by anything electrical. A 100 mm by 80 mm board tiles badly, while the same area rearranged or split into two smaller boards can tile almost perfectly. Odd shapes waste area, and a curved or angled outline can waste more than the electronics inside it is worth.

The practical consequence is that a small change to the mechanical outline can change the price more than removing a layer. Before accepting a cost increase for a fine pitch requirement, it is worth asking the supplier what the utilisation is and whether a slightly different outline would improve it.

Layer Count, Thickness and Material

Each pair of layers adds a lamination cycle, more imaging and more drilling, so the step from four to six layers is a step change in cost rather than a linear one. Six to eight is a smaller increment, because the process is already multilayer. Below four layers, a two layer board is the cheapest option by a wide margin, and for many products the question of whether the design genuinely needs four layers is worth revisiting before a volume commitment.

Material choice is the next lever. Standard FR-4 with a normal glass transition temperature is the default and the cheapest. Higher Tg material costs more and is used where the assembly process or the operating temperature demands it. Low loss and high frequency laminates cost substantially more, and they also cost more to process because their drilling and lamination behaviour differs from FR-4. Choosing a specialty material for a design that does not need it is one of the most expensive mistakes in a volume build, and it is usually made because a prototype needed it rather than because production does.

Thickness follows the same logic. A thin board uses less material but can be harder to handle, so the cheapest option is usually the thickness that the process is set up for rather than the thinnest available.

Copper weight is a common point of confusion. One ounce finished copper is standard. Two ounces costs more in material and in etching, and beyond that the process changes again, because heavy copper requires different etching and plating control. A design that specifies 2 oz for thermal reasons but carries only signal current is paying for copper it does not use.

Feature Size and Tolerance

Line width, spacing, drill diameter and registration tolerance set the process class, and the process class sets the price. A board that can be built on standard equipment is cheap; a board that needs a tighter class moves to equipment with a longer queue and a lower throughput.

The thresholds that matter are the smallest trace and space on the design, the smallest finished hole, the aspect ratio of the hole to the board thickness, and the annular ring. Each one has a comfortable range and a stretch range. Living comfortably inside the standard range costs nothing extra. Pushing one feature into the stretch range usually triggers a class change for the entire board, so a single aggressive trace can reprice a design that is otherwise simple.

A supplier can often build a 75 micrometre line and will quote it, but the yield on that feature is lower and the price reflects the risk. If only one net needs that width, it is worth checking whether the routing can be rearranged so the requirement disappears.

Drill size behaves the same way. Small holes need better registration and more expensive drilling, and the aspect ratio limits how small a hole can be on a thick board. A design that keeps its vias in the standard range and uses one size of drill for as many holes as possible will drill faster than one that uses five different sizes, and the difference shows up in the price.

Surface Finish

The finish protects the copper and provides the solderable surface, and the options sit on a clear price ladder.

Hot air solder levelling is the cheapest and the most forgiving, but the surface is not flat, which makes it unsuitable for fine pitch components. Immersion tin and organic solderability preservatives are flat and inexpensive, with limited shelf life and handling constraints. Electroless nickel immersion gold is flat, durable, shelf stable and considerably more expensive, which is why it appears on fine pitch, high reliability and long life products. Hard gold on a connector edge is a separate operation and adds cost on its own.

Finish choice interacts with the rest of the process. A finish that needs an extra plating step adds a process step and a queue, and one with a short shelf life forces a production and shipping schedule that may not suit the customer. The cheapest finish that meets the assembly requirement and the storage life is almost always right, and the requirement should be written as a shelf life and a solderability standard rather than as a finish name.

Test Strategy

Testing is a visible line item in a volume quote, and it is often the one the buyer tries to remove first.

An electrical test on a bare board is standard, and it is done either on a flying probe or on a dedicated fixture. For small quantities the flying probe wins because there is no fixture cost, but it is slow. Above a certain quantity a bed of nails fixture amortises and the per unit test time drops sharply, which means the cheapest test strategy depends entirely on the order size. A quote that changes test method between 500 and 5,000 pieces is normal and worth understanding rather than arguing about.

Optical inspection, impedance coupons and microsectioning are quality controls rather than tests. What affects the price is their frequency: sampling versus every panel, and a coupon per panel versus a coupon per order.

Quantity, Lead Time and Setup

Setup and tooling are fixed costs, so their contribution per unit falls as quantity rises. This is why the unit price curve is steep at low quantities and flattens later, and why a buyer comparing two suppliers at 100 pieces is mostly comparing two sets of setup charges.

Quantity interacts with panel utilisation in a way that catches people out. Fabricators build in panel multiples, so an order that does not divide cleanly into panels may leave a partial panel that still has to be built. Ordering 4,800 pieces when the panel yields 100 boards produces 48 panels; ordering 4,850 produces 49 panels and pays for the 50 spare boards that go into stock or scrap.

Lead time is priced too. A standard lead time uses the existing queue, while an expedited order jumps it and pays for the disruption, and the premium grows with the number of steps the board requires. On a volume programme, ordering frequently against a forecast is cheaper than expediting, provided the forecast is accurate.

Yield, Rework and the Cost of Being Marginal

Yield is the block that is least visible and most damaging. A design that runs at 95 percent yield on a four layer board with standard features is unremarkable. The same design pushed into a fine line class may run at 80 percent, and the 20 percent that is scrapped carries the cost of all the material and machine time that went into it, not just the material.

Where a board is marginal, the failure mode is usually specific: a plated barrel that does not fully cover, an annular ring that breaks out after registration drift, or a fine trace that opens where the etch runs a little long. Each of those has a design-side mitigation, and it is cheaper to add copper or open a clearance than to accept a yield loss for the life of the product.

Levers That Genuinely Reduce Cost

• Improve panel utilisation by adjusting the outline or the panel layout before tuning the electrical design.
• Stay inside the standard process class for line width, spacing, drill size and aspect ratio.
• Choose the cheapest laminate that meets the temperature and loss requirement, and no better.
• Use one via size and one finished copper weight wherever possible.
• Match the surface finish to the assembly requirement rather than to habit.
• Do not remove electrical test to save a line item; agree the right test method for the quantity instead.
• Order in panel multiples, and check that the quantity divides cleanly.
• Separate the prototype build from the production build, so the prototype choices do not migrate into volume.

Most of these are decisions made during layout rather than during purchasing, which is why a cost review is worth doing before the design is frozen. A supplier who can see the design can usually point out the two or three features that are setting the price, and that conversation is cheaper than a redesign. It is also worth comparing quotes on the same basis: a quote built on a different finish, a different test method or a different laminate grade is not a lower price, it is a different board.

FAQ

• Does a higher quantity always lower the unit price? Yes, but with diminishing returns. The biggest fall happens between prototype and the first production quantity.
• Is it cheaper to design for HDI? No. HDI adds processes and cost. It is used to fit more function into a smaller area, not to save money.
• What is the cheapest surface finish? Hot air solder levelling, provided the component pitch allows it.

Summary

At volume a bare board is priced by the material it consumes, the process steps it needs, the yield it achieves and the fixed costs spread over the order. Panel utilisation is the largest lever and the easiest to overlook; layer count and material come next; feature size and finish decide which process class the board lands in.

Reducing cost at volume is therefore mostly a matter of staying inside the standard process, using the cheapest material that meets the requirement, and letting the fabricator see the design early enough to comment. Where a design has to be aggressive, a prototype build and a capability review before the volume release will show which of the aggressive features is actually necessary, and where the real cost sits.