Via-in-Pad Design Guide: Filled, Capped, and When You Actually Need It

What is via-in-pad?

Via-in-pad (VIP) is exactly what it sounds like: a plated through-hole or microvia placed inside the copper pad of a surface-mount component, instead of being "dog-boned" out to a separate landing. It became unavoidable around 2010 when 0.5 mm pitch BGAs went mainstream, and it is now baseline for anything with a ground-paddle QFN or fine-pitch BGA.

Done right, VIP shrinks your board, improves thermal performance, and cleans up BGA escape routing. Done wrong, it wicks solder into the via barrel during reflow, starves the joint, and gives you head-in-pillow defects that AOI may flag but the customer finds first. This guide covers the cost trade-offs, the four common VIP variants, and when you actually need each one — based on what we see daily on the SMT line at Energetika-VDS in Strumica.

The four VIP variants

Numbers above are for a 4-layer 100x100 mm board at 100 pcs from a typical EU fab. At 10 pcs the percentage adder roughly doubles because VIP carries a fixed NRE for the fill cycle.

When is VIP actually mandatory?

VIP is genuinely required, not just nice-to-have, in these cases:

• BGA pitch ≤ 0.5 mm. Dog-bone escape physically does not fit. You need VIP or you redesign the BGA out of the board.
• 0.65 mm pitch BGA with > 9x9 ball grid. You run out of channels in the inner layers without VIP.
• QFN thermal pads with > 4 thermal vias. Vias inside the pad must be filled and capped or solder volume drops below IPC-A-610 Class 2 limits.
• Boards thinner than 0.8 mm with HDI stack. Microvia-in-pad is the only viable interconnect.
• High-current grounds where via array carries > 5 A. Filled conductive VIP drops Rja by 15-30%.

For 0.8 mm pitch BGAs you can usually still dog-bone if you have 4+ inner layers. For 1.0 mm pitch BGAs, save the money — dog-bone is fine.

BGA escape routing: the real driver

The reason VIP exists is escape routing. Take a 0.5 mm pitch BGA, 15x15 balls. Without VIP you need a channel between balls wide enough for one trace plus two clearances. At standard 75 µm trace / 75 µm clearance you get exactly zero room. With VIP under every ball, you route on the inner layer directly down from the ball — no channel needed.

Rule of thumb for BGA channel count on a 100 µm trace/space process:

If your stack-up is going from 4 to 6 layers just to dog-bone a BGA, VIP usually pays for itself. Add up the bare-board delta versus the VIP adder before you commit.

How to spec VIP on your fab note

A clean fab note line looks like this:

Vias in pad: 0.20 mm finished hole, plated 25 µm Cu min, filled with non-conductive epoxy (Taiyo THP-100DX1 or equivalent), planarized, Cu-capped 20 µm min, ENIG over cap. Applies to all vias inside SMD pads of components U1, U4, U7.

Things that go wrong if you do not spec all four:

1. No fill material specified — fab picks cheapest, usually conductive Ag epoxy. Fine for grounds, bad for signal vias because of capacitance shift.
2. No planarization spec — you get a dimple, paste deposit varies by 30%+, reflow yield drops.
3. No cap thickness — thin caps crack at the edge during reflow, opening micro-paths for solder wicking.
4. No component callout — fab fills every via on the board and your unit cost doubles.

On every quote we run through the quote estimator, the second question after pitch is "are there vias in your BGA pads?" because it changes both bare-board cost and assembly yield.

Cost reality check

On a 4-layer, 80x60 mm board, 4 BGAs (0.5 mm pitch), 250 vias-in-pad total, 200 pcs production run from a Tier-2 EU fab in May 2026:

The "VIP is expensive" reflex is often wrong once you account for layer count savings. Run the math.

Where this fits in our workflow

Energetika-VDS handles VIP boards routinely on our DDM Novastar SPR-45 / LS60 / GF-120HT line. We will not quote a 0.5 mm pitch BGA without filled+capped VIP, because we have seen the failure modes too many times. If your fab does not offer Cu-capped epoxy fill, we will steer you to one of the EU partners we use through component sourcing and production support — typically Würth Elektronik or a similar tier.

For prototype-stage builds where VIP cost stings, the alternative is to relax to 0.65 mm pitch and skip VIP for the proto run, then move to VIP for production. We do this all the time on production transfer projects. For full traceability of which boards have which VIP spec, see quality and traceability.

Frequently asked questions

What is via in pad? A plated via placed inside an SMD copper pad, used instead of routing the via out to a separate landing — required for fine-pitch BGAs and most thermal-pad QFNs.

When should I use VIP? Mandatory for BGA pitch ≤ 0.5 mm, strongly recommended for 0.65 mm BGA escape, and for QFN thermal pads with more than 4 thermal vias. Optional everywhere else.

How much does via in pad cost? Typically +12-25% on bare-board cost depending on fill material and cap plating, plus a small NRE per design. Often offset by the layer-count reduction VIP enables.

Filled vs unfilled via in pad — which do I need? Filled-and-capped (non-conductive epoxy + Cu cap) is the safe default. Unfilled VIP wicks solder during reflow and is only acceptable on coarser pitches with low-reliability targets.

Does Energetika-VDS assemble VIP boards? Yes — VIP is standard on most of our SMT assembly work. We default to IPC Class 2 with Class 3 capability for medical and industrial customers.