HDI PCB Design: Microvias, Via-in-Pad and Stackups

Modern packages — fine-pitch BGAs, chip-scale parts, dense connectors — stopped fitting on conventional PCBs years ago. HDI (high-density interconnect) is the answer: boards that route more connections in less space using laser-drilled microvias and thinner geometry. This guide covers what HDI actually is, the design choices behind it, and why an HDI board leans on front-end engineering far harder than a standard stackup.

What makes a board "HDI"

HDI is defined by feature size and via structure, not layer count. A board is HDI when it uses microvias — laser-drilled holes typically 0.15 mm or smaller — to reach density a mechanically drilled via can't. The payoff is escape routing: a 0.4 mm-pitch BGA simply cannot be broken out with standard through-holes, so its pads fan out through microvias into the inner layers.

Microvias: the core building block

A microvia is drilled by laser rather than a mechanical bit, which is what lets it shrink. Two rules dominate their design:

Aspect ratio — the depth-to-diameter ratio must stay low (roughly 0.75:1) so the plating fills reliably. That's why microvias only span one dielectric layer at a time.
Stacked vs staggered — to go deeper than one layer, microvias are either stacked directly on top of each other (needs copper-filled vias and adds cost) or staggered sideways (cheaper, but eats routing space).

Engineering note

Every microvia layer is built in its own lamination cycle. A "2+N+2" board laminates the core, then adds two microvia layers per side in sequence. More microvia layers means more lamination cycles — and each cycle is cost, lead time, and a fresh registration risk the front end has to control.

Via-in-pad

HDI almost always brings via-in-pad: the via sits directly inside a component pad instead of beside it. It's the only way to break out the tightest pitches, but an open via under a pad wicks solder away during assembly. The fix is a filled and capped via — plated shut, planarized, and plated over — so the pad stays flat and solderable. That's a fabrication capability your board's fab must actually have, and something DFM has to confirm before you commit.

HDI stackup types

HDI stackups are described by how many microvia layers sit outside the core:

Stackup	Structure	Typical use
1+N+1	One microvia layer each side of an N-layer core	Entry-level HDI, moderate BGA pitch
2+N+2	Two microvia layers each side	Dense BGAs, mobile-class routing
Any-layer	Microvias on every layer, no through-holes	Highest density (phones, wearables)

The right choice is driven by the tightest component on the board — usually one BGA sets the whole stackup.

Why HDI stresses the front end

A standard board tolerates loose front-end work; HDI does not. Three reasons:

Registration — sequential lamination stacks tolerance on tolerance. Microvia-to-pad alignment across builds has almost no margin, so CAM must compensate deliberately, not generically.
Fab-specific rules — microvia size, capture pad, and fill capability vary by fabricator. A design that builds at one fab fails at another. DFM against the target fab's real deck is mandatory, not optional.
Cost control — every extra microvia layer is real money. Good front-end engineering finds the stackup that meets the density need in the fewest lamination cycles.

Do you actually need HDI?

HDI is powerful but not free. You need it when a component's pitch forces it — sub-0.5 mm BGAs, high pin counts, or a form factor that won't allow more layers. If your densest part breaks out cleanly on a conventional stackup, HDI is cost you don't need. When you're unsure, that judgment call is exactly what a pre-production review answers.

See our HDI & microvia engineering service and the controlled impedance basics guide — impedance and HDI decisions are made together — or send us your stackup for a first-pass read on whether it's buildable as drawn.