In sourcing conversations, pcb fab is shorthand — and like most shorthand, it hides more than it reveals. A buyer who says they need fab is usually signalling that they want an unpopulated board: drilled, plated, masked, finished and electrically tested, with nothing soldered onto it. Yet the same two-word phrase gets used for a quick-turn prototype, a 58-layer rigid build, an any-layer HDI design, or a flexible circuit that folds into a wearable housing. Each of those pulls a very different set of process decisions into the quotation. Understanding the gap between “fab” as a purchasing step and “fab” as a stack of manufacturing constraints is often the difference between a clean first order and three rounds of clarification emails before the data package is even accepted.
Bare Board Versus Boxed Product: Splitting Fabrication From Assembly
When a buyer asks for a bare board, the scope is deliberately narrow. The deliverable is a finished laminate: copper patterned and etched, holes drilled and plated, solder mask applied, surface finish deposited, legend printed, and electrical testing completed. Nothing is populated. That distinction matters commercially because fabrication and assembly are priced on completely different cost drivers — one is driven by materials, layer count, drilling density and yield, the other by component count, placement speed, stencil setup, inspection and test.
The confusion typically shows up in two directions. Some buyers send a bill of materials with the Gerbers and assume the quote will return as a turnkey assembly price; others send a stackup drawing and expect assembly to be included because the same supplier does both. A clean fabrication request states the layer count, the laminate, the copper weight, the surface finish, the impedance requirements and the acceptance standard, with no components implied. A clean assembly request adds the BOM, the pick-and-place data, the approved vendor list and the test specification on top of all of that.
Suppliers that run both disciplines under one roof tend to compress this handover. Where fabrication and assembly sit in the same facility, the bare boards can move into SMT, through-hole or mixed-technology lines without a separate shipping and re-inspection cycle, and the assembly side can handle fine-pitch micro BGA down to 0.4 mm, components as small as 01005, large board formats up to 20” x 24”, plus X-ray inspection, IC programming and functional testing. For a buyer, the practical question is not whether one company can do both — it is whether the fabrication quote in front of them is for the board alone, or whether it quietly assumes assembly will follow.
It is also worth separating the word “fab” from the word “prototype”. A prototype order is a quantity decision, not a process decision. The same fabrication steps apply whether the buyer wants five boards or a short production run; what changes is how much setup cost gets spread across the order. Many suppliers now publish an online price calculator that returns an instant quotation for prototypes, quick-turn builds, HDI, RF, rigid-flex, flex and assembly work, which makes it easy to preview how layer count and finish choices move the number before any data is formally submitted.
What a Fabrication Quote Is Actually Pricing
Reading a pcb fab capability list before sending data is the fastest way to understand why two quotes for “the same board” can differ by a wide margin. Fabrication pricing is a function of how far the design pushes each individual process step, and capability statements describe exactly where those limits sit. Layer count is the obvious one, spanning 2 to 58 layers on rigid constructions, but copper weight, drill strategy, via fill and surface finish are often the real cost swingers.
Consider drilling and via structure. Mechanical and laser drilling for blind and buried vias, microvias filled with epoxy and copper-capped, and via-in-pad with conductive, non-conductive or copper plug fill are all standard offerings, but each adds process steps that a plain through-hole board never touches. Back drilling is another: a minimum hole size of 15.7 mils with a depth tolerance of ±6 mils is achievable, and that tolerance is what makes back-drilled stubs predictable at high signal rates. On the routing side, minimum trace width and spacing of 0.002” and BGA pitch down to 0.15 mm define how much of the design can be packed into a given area.
Materials drive the rest of the number. Standard FR-4 is not the same purchase as polyimide rigid laminate, low-loss high-speed laminate, or a microwave-grade substrate. Surface finish selection compounds it — ENIG, ENEPIG, hard nickel/gold, wire-bondable gold, selective gold plating, gold fingers, OSP, immersion silver and immersion tin all sit at different points on the cost and performance scale, and the right answer depends on whether the board will see wire bonding, press-fit connectors, repeated mating cycles or simple reflow. Impedance control to ±5% is a further line item, because it requires controlled dielectric thickness and measured trace geometry rather than a default build. Heavy copper up to 30 oz. and cavity processes sit at the far end of the spectrum, where the fabrication quote stops resembling a commodity price at all.
Beyond Standard Rigid: HDI, RF, Rigid-Flex and Flex Requests
Most buyers who say “fab” are still thinking about a rigid board, and that assumption breaks down quickly. HDI constructions using staggered, stacked or skipped microvias — up to a 9+N+9 any-layer build — are now routine requests for dense designs, and they change the fabrication plan from the first drill hit onward. Cavity boards, countersunk holes, press-fit holes, castellated edges and semi-flex constructions each require specific process knowledge, and a supplier that lists them is telling you which questions it already knows how to answer.
Radio-frequency work is a category of its own. Microwave and RF boards move between low MHz and millimetre-wave frequencies, and material selection is the single most critical design decision because dielectric constant, coefficient of thermal expansion and loss tangent determine whether signal integrity survives. Copper roughness matters too, since it governs the skin effect as signals travel along an interconnect. That is why high-frequency fabrication draws on distinct material families — Rogers grades such as RO4003C, RO4350B and RO4450F, Taconic TLY and RF-35 series, Arlon AD-series laminates, Isola high-performance and ASTRA MT77 materials, and Panasonic MEGTRON grades for low-loss high-speed networking — often combined in hybrid stackups where a single board mixes low-loss and standard laminates.
Flexible and rigid-flex requests add mechanical constraints on top of electrical ones. Feature sizes down to 25 µm and flexible dielectric cores down to 25 µm appear in high-reliability rigid-flex work, while flex fabrication draws on polyimide (often called Kapton), PET, PEN, LCP and fibre substrates, with thick copper flex circuits running up to 500 µm and sculptured flex circuits used where three-dimensional forming is required. Surface finishes shift accordingly — ENEPIG, immersion silver, immersion tin, immersion nickel/gold and electrolytic nickel/gold — because a flex tail that bends repeatedly or slides into a friction contact has different demands than a static rigid board. The practical lesson for buyers is straightforward: describe the fab request in the language of the hardest feature on the board, not the average one, and the quotation you receive will match the board you actually need built.
Ankara robotics engineer who migrated to Berlin for synth festivals. Yusuf blogs on autonomous drones, Anatolian rock history, and the future of urban gardening. He practices breakdance footwork as micro-exercise between coding sprints.