PCB Hole Attributes and Special Drilling: Common CAM Review Conflicts

A drill chart with plated and non-plated flags often looks like the simplest part of a PCB release package. In CAM review it frequently is not. Size and XY location describe geometry. They do not, by themselves, establish plating intent, start and stop layers, tool family, residual stub, finished-hole class, countersink geometry, or bevel depth — or whether the feature is meant as a via, a component barrel, a press-fit interface, a mechanical hole, or tooling.

This article is for layout engineers, hardware owners, and manufacturing engineers who need PCB hole attributes and special drilling to get through CAM with fewer EQ loops. It maps the conflict families that commonly stop a job before plating and drilling are programmed: PTH vs NPTH when pad, hole, and net disagree; special mechanical features that cannot be run from a diameter alone; laser, blind, and buried via data that does not line up with structure; and hole, slot, and thickness tolerances that cannot be frozen from the files as released.

Exact stub windows, angle catalogs, laser-aperture charts, and finished-hole capability numbers belong on later problem pages. The aim here is how CAM reconstructs process from incomplete attributes, and what is worth deciding in layout before submit.

Hole Attributes, Plating Intent, and How CAM Reads a Drill File

CAM does not read a hole the way a CAD properties panel does. The layout tool stores a flag, a drill size, and a pad stack. The factory still has to turn that into a plated or non-plated flow, a tool, a depth, a mask opening, and an inspection method. Diameter and location are necessary. They are rarely sufficient.

PTH vs NPTH is a process decision, not only a CAD checkbox. A plated-through hole typically enters the plate sequence, grows copper on the wall, and is measured as a finished hole after plate. A non-plated hole generally follows a different path and is held to a drill or route size. Using the two flows interchangeably because the diameters match is a common way attribute errors become plating errors.

When the drill table is thin or inconsistent, CAM cross-checks pad presence, pad-to-hole size, and net connectivity. A hole with an annular ring and a net is often treated as electrical, but that reading still depends on mask, unused-pad rules, and the drill flag. A hole with no pad, no net, and a mechanical note is often NPTH — unless one of those signals is missing or contradicted. The difficult cases sit in between: same-size pad and hole with no net; an NPTH flag on a feature that still carries a pad on copper; a via-sized hole reused as tooling.

The same diameter can be a signal via, a component PTH, a press-fit barrel, a back-drilled structure whose stub is the real requirement, or a mechanical NPTH. The X/Y list rarely distinguishes those functions. The distinction usually lives in notes, layer pairs, drawings, and consistent pad-net-mask data — and that is the data that often fails to travel with the Gerbers.

Typical gaps include drill-table notes that never reach the PDF or Excellon header; NPTH flags that apply to only some hits of a tool; unused-pad rules that change inner-layer appearance without changing the plated flag; back-drill layer pairs written as a sentence instead of a start/stop pair; press-fit mentioned without a finished size; countersink or bevel left in a partial mechanical view; and laser versus mechanical via type inferred from size rather than named.

Design-Time Decisions That Lock Hole Type and Special Drill Process

Much of this EQ family is set before release.

Classify each hole by function first: electrical via, component PTH, press-fit, mechanical NPTH, tooling, back drill, laser or blind/buried via, slot, countersink, or gold-finger bevel. Reusing one drill code for two functions because the finished diameter looks similar makes plating intent and tool family harder to read.

Pad, annular ring, and net then need to line up with that function. An electrical hole generally needs a pad-stack and connectivity that survive unused-pad removal. A mechanical NPTH generally should not keep a circuit pad only for convenience. A press-fit hole needs the barrel and finished size the connector specification actually uses. When pad, ring, and net point at different functions, attribute review tends to stop.

Choose the process family while the stackup can still move. Mechanical through-hole, controlled-depth back drill, laser blind via, and buried via are difficult to add cleanly after the layer count is frozen. Blind and buried structures need to match the press-up. Back drill needs to match the layer pair that still carries the signal.

Write special-hole geometry as a complete requirement: depth or layer pair, side, angle, finished size versus drill size, tolerance class, and copper keep-out. "CSK" or "chamfer fingers" as a lone note leaves CAM reconstructing the feature. Countersink and gold-finger bevel belong in the mechanical drawing set because both intersect outer copper.

Treat hole, slot, and thickness tolerances as one system. Finished-hole meaning changes with plating and aspect ratio. A slot is not a round hit with a stretched pad. Copying one tight number onto all three callouts can close the process window even when each callout would have been discussable on its own.

Hole-Attribute Conflicts CAM Raises Before Plating

Before a drill program or plating rack-up is safe, CAM needs a coherent account of whether each hole is plated, electrical, or mechanical. Diameter by itself rarely answers that. A wrong call can change plating sequence, tenting or plugging, unused pads, and hole-wall quality — guessing NPTH on an electrical hole, or PTH on a hardware hole, is a scrap or reliability risk rather than a plot mark-up.

PTH vs NPTH when pad, hole, and net do not agree

The conflict is a set of signals that do not point the same way. The drill table may say plated. The netlist may show no connection. The pad may be the same diameter as the hole. Mask may be open as if the hole were a component barrel. Any two of those can be true in a clean design. All of them together, without a note, are not a reliable default.

A same-size pad with no net is not automatically NPTH. It may be an unused via the layout tool failed to remove, a mounting hole that still has a land, or a reused footprint. What CAM typically needs is a named function — via, component hole, mechanical hole, or unused pad with an explicit removal rule.

the two types of holes appear to be PTH holes but are currently defined as NPTH holes

NPTH declared on a hole that still has a circuit pad

The drill list or a fab note calls the hole NPTH, but copper still carries a pad on a trace. A pad on a net usually implies electrical use and, in many flows, plating. Leaving the pad while declaring the hole non-plated leaves open whether copper should pull back, whether mask should tent or open, and whether inner layers should keep the land.

Drilling an NPTH path through a live pad does not give a well-defined mechanical hole. The cleaner resolution is in the design data: remove the pad if the feature is mechanical, or restore PTH and connectivity if it is electrical.

 The highlighted holes are defined as Non-plated Through holes (NPTH) in the drill file, but they have pads on both sides.

Special Mechanical Drilling CAM Cannot Run From a Diameter Alone

Back drill, press-fit, countersink, and gold-finger bevel often appear in the standard drill list as another size, or they do not appear there at all. The list typically does not carry tool type, residual stub, press-fit finished class, countersink angle, or bevel depth. Incomplete callouts can force an EQ even when the board is otherwise fabricable. Common stop points are tool substitution and copper keep-out.

Back drill and residual stub

A back drill is mainly a depth-and-layer-pair problem. The usual requirement is the unused stub left after the unused barrel is removed, plus a limit so the drill does not enter the remaining signal layer. Hole size selects a tool. It does not define stub.

CAM tends to stop when the stub call is tighter than the package can confirm, or when start and stop layers are missing. Without a layer pair, leftover stub and a drill through the signal pad both remain possible. How the stub will be inspected is part of whether the requirement can be frozen.

the 16 holes indicated in the picture attached all need to do back drilling

Press-fit hole drill and finished tolerance

A press-fit hole is a finished-hole and barrel requirement, not a generic PTH. Connector specifications typically care about the plated hole after process and a tolerance class referenced differently from a standard component hole. The drill is chosen to land on that finished size after plate, not to match a CAD drill field copied from a via.

When the package only marks the holes plated and mentions "press-fit" in a note, CAM is left choosing between a standard PTH card and an unstated class. The press-fit requirement needs to sit in the release package, pointed at the hole codes that use it.

 Holes larger than 2.0mm cannot be made as press-fit holes

Countersink angle and tool substitution

A countersink is angle, side, depth, and finished diameter at the surface — not only the through-hole in the middle. Shop tools are discrete. An angle that does not match a standard cutter often becomes a substitution EQ.

Substitution can change screw seat and the copper cut at the cone. CAM generally asks before changing the tool. The useful answer is accepted or forbidden, then a revised drawing so the following lot does not repeat the question.

no countersink hole details

Gold-finger bevel depth and copper exposure

Gold-finger bevel is edge machining that intersects outer copper and contact plating. It sits in this topic because the review set treats it with special depth control, not as a connector-design essay. The object is depth versus copper pullback.

If the bevel cuts deeper than the setback, base laminate can be exposed or the plated contact edge cut. The feature is often absent from drill data — that is expected — but depth and keep-out still need to live on the mechanical view used at the machine.

the distance from the gold finger to the board edge is 0.65mm

Laser, Blind, and Buried Via Conflicts in CAM Review

For CAM, blind, buried, and laser vias are one object: aperture, process choice, registration, and pitch, read against the stackup. Diameter alone does not establish laser versus mechanical. A blind or buried label that does not match the press-up is an attribute problem before it is a drill problem.

Process choice follows size, depth, and stackup together. Files that omit via type leave CAM inferring from diameter. Pitch is a structure issue: capture pad, registration, and hole-to-hole distance affect whether two blinds stay separate. CAM typically flags that relationship before committing a laser program.

Blind/buried diameter vs laser process feasibility

Mechanical depth-controlled drilling and laser ablation are different tools, with different capture-pad expectations and different tolerance language. A tight finished-size tolerance on a laser via is also a measurement and capture-pad problem: the land has to receive the via after registration.

When via type is omitted, size becomes a hint. The package needs process, start-stop layers, and target size in one place.

For all the blind vias that size is more than 0.15mm, we will use laser to drill

Non-standard laser aperture

Laser tools and capture-pad libraries are built around a conventional size family. An off-grid aperture is neither automatically rejected nor automatically accepted. It becomes a feasibility and pad-stack question: does the capture pad still work, and is the aperture a mandatory functional size or a CAD default?

Non-standard laser aperture

Blind-via pitch and break-through risk

Close pitch plus registration error plus a modest capture pad can merge or rupture holes. The features do not have to touch in CAD to interact in lamination and laser alignment. The concern is structural: isolation between adjacent vias, or a pad that no longer captures the via. When a change is needed, it is usually pad, pitch, or stackup alignment — not only a smaller number on the same centers.

 the via holes are too close to the board edge

Hole, Slot, and Thickness Tolerances That Cannot Be Frozen From the Files

Finished-hole tolerance is the result of drill, plate (if PTH), route or nibble (if a slot), and measurement method — not the CAD tolerance field by itself. NPTH removes plating compensation but not drill wander or thickness interaction. Calling hole, slot, and board thickness to one tight number at once is often a control problem rather than three independent accuracy problems.

The useful review question is which dimension is actually functional.

Combined hole / slot / thickness tolerance stack

Three tight callouts remove window from different ends of the same process. Thickness changes aspect ratio and the meaning of a finished plated hole. A slot through that thickness is routed or nibbled, not drilled as a single hit.

The tolerance is recommended to be controlled at +/-0.13mm

Hardware fit, press-fit class, and card-edge thickness are different functional stories. Tightening all three because a title block says "unless otherwise specified" is how this EQ appears on boards that were otherwise ready to build.

Slot-hole tolerance treated like a round drilled hole

A slot is a path. It may be drilled at the ends and routed between them, or nibbled. Each path accumulates width and length error differently. A hole-style bilateral tolerance on both width and length often mismatches requirement form and process path.

When CAM reports that a slot tolerance cannot be held, the form of the requirement is frequently the issue. Restate which slot dimension is functional — width for a tab, length for travel, position to a datum — rather than copying the plated-hole block onto the slot.

For the slot hole,we will follow our standard  and control as +-0.1mm

NPTH hole tolerance

NPTH removes plating compensation. It does not remove drill wander, entry condition, or the effect of thickness on a deep unplated hole. "No plate" does not make an arbitrarily tight finished size free.

If the NPTH is a hardware hole, the functional fit belongs in the note. If it is tooling, the panel drawing can say so without borrowing component-hole tolerance language.

What These Conflicts Do on the Fabrication Floor

These findings do not stay in CAM as paperwork.

A wrong PTH vs NPTH call can change plating, hole-wall quality, and solderability. Found after plate, that is often scrap or field risk. Found before plate, it can still force a respin of drill, mask, and unused pads.

Special tools add setups. Back drill, countersink, bevel, and a press-fit drill card are extra programs and inspections. Unconfirmed geometry can idle the lot: panels may be in house, but the machine cannot be set until stub, angle, finished class, or bevel depth is frozen. The board may be buildable. The released data may not be.

Laser pitch and odd apertures show up as registration yield and capture integrity. Over-tight hole, slot, and thickness tolerances increase sort and extra measurement — or force a mid-job spec change that the next lot reopens.

If bevel or back-drill depth is guessed, the reliability paths are copper exposure on the bevel, and residual stub or a drilled signal layer on the back drill. Missing depth data is treated as a stop for that reason.

Clarification time is the common delay. Process capability is often not the blocker. Completeness of attributes is.

How CAM Findings Are Resolved Before Manufacturing

Resolution is a handoff: put the agreed state back into files and fab notes so plating and drilling can be programmed once.

Attribute conflicts close by correcting source data. Plated flag, pad, net, and mask need to agree in Gerbers, IPC-2581 or ODB++, and the drill table. A verbal "treat as NPTH" does not travel reliably to the next revision and does not protect unused pads.

Back drill, press-fit, countersink, and bevel close by completing the missing drawing: layer pair and measurable stub; finished-size class pointed at the hole codes; angle, side, and depth; bevel depth and copper setback.

Laser, blind, and buried features close by stating via type, start-stop layers, target size, and whether an unconventional aperture is mandatory.

Tolerance EQs close by naming the functional dimension and restating the others in fab notes. The title-block default does not need to apply to slots, NPTH, and thickness at the same time.

Tool substitution needs an explicit accept or reject, then a revision. If substitution lives only in email, the next lot can rebuild the same EQ from the old drawing.

Pre-Submit Checklist for Hole Attributes and Special Drilling

• Every hole in the drill list has an explicit PTH or NPTH attribute, or a named via type.
• No hole is same-size pad / no-net without a stated mechanical or unused-pad rule.
• No NPTH sits on a copper pad that still implies a net.
• Back-drill layer pairs and stub requirement are in the package.
• Press-fit holes point to a finished-size / drill callout, not a generic PTH size.
• Countersink angle, side, and depth are drawn; substitution is pre-accepted or forbidden.
• Gold-finger bevel depth and copper setback are on the fab drawing.
• Blind, buried, and laser vias state process, start-stop layers, and size; no unexplained odd aperture.
• Blind-via pitch has been reviewed against capture pads (pass or fail only).
• Hole, slot, and thickness tolerances are each present and are not copied as one blanket tight spec onto NPTH and slots.

FAQ

Q1: How does CAM decide PTH vs NPTH when the drill table and the pads disagree?

A1: It generally does not decide from one field. Plated flag, pad, net, and mask are read together. When they conflict, review stops. Diameter is a poor tie-breaker.

Q2: Is a same-size pad with no net automatically an NPTH?

A2: No. That pattern can be an unused via, a CAD leftover, or a mechanical hole that still has a land. Without a stated rule, it stays an attribute question.

Q3: Why is a back-drill stub not just another drill-size callout?

A3: Size selects a tool. Layer pair and measurable stub define the feature. Missing pair information leaves both leftover stub and a drilled signal layer possible.

Q4: What does a press-fit hole need beyond a PTH diameter?

A4: A finished-hole class pointed at the hole codes that use it, landing on the connector specification after plate.

Q5: Why does an unusual countersink angle generate an EQ?

A5: Cutters are discrete. An off-list angle is a substitution request. Substitution can change screw seat and exposed copper, so it needs a written accept or reject and a revised drawing.

Q6: Why is gold-finger bevel grouped with hole and special-drilling review?

A6: The review set treats it as edge machining whose depth intersects outer copper and plating. The risk is exposure or cut plating when setback is not on the fab drawing.

Q7: Does a small blind-via diameter mean the hole is a laser via?

A7: Not by itself. Process follows size, depth, and stackup together. Omitting via type is what turns diameter into a guess.

Q8: Why can an NPTH still have a tolerance CAM will not freeze?

A8: Removing plating removes plating compensation only. Drill wander, thickness, and measurement remain.

Q9: Should hole, slot, and board-thickness tolerances use the same tight default?

A9: Not as a stack. They interact through process path and aspect ratio. Assign the tight number only where the product uses it.

Q10: After an EQ, where should the agreed geometry live?

A10: In the release package: corrected attributes, completed layer pairs and depths, accepted substitutions in a revision. Email can close the current question; it does not reliably feed the next lot.