6 mil Traces on Single-Sided PCB: Copper Weight, Side Etch and Panel Uniformity

The short answer

Your supplier's capability table says min trace/space 2mil/2mil. You need 6 mil (about 0.15mm) on a single-sided board, so the margin looks enormous. What actually decides whether that board holds 6 mil is not the table value but three things: copper weight, etch side attack, and panel uniformity.

Every limit on a capability table is a single-parameter limit — it holds under one favourable process combination: thin copper, small panel, symmetric double-sided construction, etch parameters tuned for that case. Single-sided boards, chosen for cost, usually go the other way: bigger panels, heavier copper, copper on one side only. Each of those eats into fine-line margin.

So this article does not argue about whether 2mil is real. It answers the useful question: for your specific single-sided board, what inputs and what acceptance agreement make 6 mil actually manufacturable.

1. What failure looks like

Fine lines on single-sided boards rarely fail across the whole panel. They fail locally and by batch:

1. Trace narrower than drawn, but continuous. Electrical test passes, AOI passes, and the trace is measurably thinner than designed. Harmless on a logic signal, a latent problem on anything carrying current or holding impedance.
2. Localised opens. The opens cluster at specific panel positions — usually an edge or one side — rather than distributing randomly. That is a uniformity symptom, not a capability limit.
3. Batch-to-batch drift. First build measures fine, the repeat build comes back thin or heavy. That points at etch parameters or copper-foil batch variation, not at the design.

All three trace back to one chain: how much copper is removed, and whether every location on the panel loses the same amount.

2. Side etch: trace width is lost from the flanks

Etching is chemical, and the etchant does not act only downward. The resist protects the top surface; both flanks of every protected conductor stay exposed to the etchant for the whole etch cycle. The result is a trapezoidal cross-section: top width close to the drawn value, bottom width smaller.

Two engineering consequences:

• Heavier copper means a longer etch, which means more side attack. Clearing thicker copper requires more time, and more time removes more copper from the flanks. That is why heavy copper and fine lines are an intrinsically conflicting pair — it is the geometry of the process, not an equipment shortfall.
• The artwork needs pre-compensation. The fab compensates the image for copper weight and etch conditions so that the post-etch width lands inside the tolerance band. Compensation is a process parameter, not a design parameter — but the fab has to know which traces are width-critical, otherwise compensation can only be applied as one panel-wide value.

For a 6 mil trace the real question is not whether 6 mil can be etched. It is which copper weight you pick, how wide the tolerance band is, and whether the fab was told this panel carries fine lines that need separate attention.

3. Copper weight and trace width: which one to fix first

/capabilities/ lists copper weight from 0.5oz to >12oz, and min trace/space of 2mil/2mil. Those two numbers do not hold on the same board. The practical decision order is:

Fix copper weight first, then trace width. Not the reverse.

Copper weight is usually driven by current carrying, heat spreading, or connector requirements — functional constraints that are expensive to change. Trace width has routing flexibility. Lock the width first and you may be forced into heavy copper afterwards, landing in exactly the hardest combination.

For single-sided boards specifically:

• Separate fine-line areas from high-current areas. A single-sided board has no inner layers to share the load; all copper is on one side. Mix 6 mil signal traces into a heavy power path and the whole panel has to run at the heavy-copper recipe, compressing fine-line margin everywhere.
• If only a few traces need 6 mil, call them out and accept a tighter local requirement rather than tightening the whole panel tolerance. That is far more achievable.
• State whether copper weight is base or finished. Single-sided boards typically have no panel plating, so the two are close — but the basis still has to be explicit, or compensation is calculated against the wrong reference.

4. Panel uniformity: trace width is not one number per panel

This is the step most often missed on single-sided fine-line work.

A horizontal etching line applies etchant by spray. Spray flow field, etchant refresh rate at the surface, and the board's position along the conveyor all cause different regions of the same panel to etch by different amounts. In practice: centre and edge of a panel do not share one trace width, and leading edge may differ from trailing edge.

/capabilities/ lists a max board size of 1250×570mm. The larger the panel, the larger the absolute regional spread. Single-sided boards are often panelised large to spread cost — so fine lines and large panels are another pair pushing against each other.

Two factors specific to single-sided construction:

• Asymmetric copper distribution. Copper on one face only means the board behaves differently under wet-process handling than a balanced double-sided board. Slight bow affects how tightly dry film laminates; where lamination is less tight, etchant creeps laterally, and that is where traces come back thin.
• Lamination and exposure sit upstream of uniformity. The imaging step (Automatic Exposure Machine or LDI Laser Direct Imaging) sets resist edge definition. A soft resist edge means etching starts from an uncertain geometry, and no amount of etch stability recovers it. Fine lines are far more sensitive to this step than coarse ones.

So "can you do 6 mil" is better asked as: where on this panel can 6 mil be held, and what tolerance band applies.

5. Design inputs to put in the file

What lets a fab assess the combination and compensate correctly:

1. Copper weight, with unit and basis stated (oz or µm; base or finished).
2. Minimum trace width and minimum space, given separately, and noted as post-etch finished values or not.
3. Width-critical trace callouts. Which traces have functional width requirements (current, impedance, press-fit). Without the callout, the fab can only treat the panel uniformly.
4. Width tolerance and measurement agreement. Top width or bottom width, which panel locations, how many points. Skip this and acceptance becomes two parties measuring differently.
5. Panelisation intent. Whether the fab may adjust the array. Panel size drives uniformity, so granting freedom here usually helps fine lines.
6. Solder mask and surface finish requirements. Finish adds or removes material on the copper and changes final conductor geometry; the effect differs by finish type — see PCB surface finish comparison.

6. Acceptance: what counts as passing

Aligning acceptance before kickoff beats arguing afterwards:

• Width measurement method and locations per item 4 above, written into the first-article report requirement.
• Visual and pattern defect acceptance class per IPC class. The same physical nick on a fine trace is judged differently under different classes — see PCB defect classification and IPC class acceptance.
• The division of labour between AOI and electrical test. AOI finds pattern defects (nicks, shorts, copper residue); electrical test (Flying Probe Tester) finds continuity. Neither one guarantees trace width is inside the band — width needs separate measurement. This is the single most common misunderstanding on fine-line projects.
• Outer-layer pattern inspection items are covered in outer layer pattern inspection checklist.
• How to annotate tolerance on the drawing without ambiguity: see manufacturing tolerance annotation on PCB drawings.

Next step

Once copper weight, minimum trace width, width-critical callouts, and the measurement agreement are assembled, a fab can assess whether the combination is feasible. Full parameter ranges are on manufacturing capabilities; single-sided construction and use cases are on single-sided PCB; bring the files to quote and engineering review.

• Image alt: Conveyor section and spray chamber of a horizontal etching line, where copper artwork is etched to its final trace width
• Self-check: One core question only (table limit vs. per-board width realisation); all figures quoted from /capabilities/ and /tolerance/; no yield, lead-time, customer case, or equipment-count claims; 6 internal links, all en locale and deployed; 6 FAQs map to body sections with no new facts introduced.

FAQ

The table says 2mil. Why is 6 mil even a discussion?

2mil is a single-parameter limit achieved under one favourable process combination. Your board has its own copper weight, size, and construction; the difficulty of 6 mil on your board comes from that combination, not from the table value.

Does a 6 mil trace require LDI?

Not necessarily. Both LDI Laser Direct Imaging and an Automatic Exposure Machine can meet 6 mil imaging requirements; the choice depends on compensation needs, volume, and panel size. What matters is resist edge quality and compensability, not the equipment name.

Can heavy copper and 6 mil traces coexist?

Heavier copper means longer etch and more side attack, so the two requirements pull in opposite directions. The workable approach is zoning: heavy copper on the power path, fine lines in signal areas, with the fine lines called out separately.

Why are some locations on the same panel thinner?

Etchant action across a panel is not perfectly uniform, and the spread grows with panel size; add the slight distortion that one-sided copper distribution causes in dry-film lamination. This is a uniformity question, managed by agreeing measurement locations and a tolerance band — not by tightening the whole-panel tolerance.

If electrical test passes, is trace width acceptable?

No. Electrical test checks continuity and AOI checks pattern defects. Neither is a width measurement. Width is measured separately, by the agreed method and at the agreed locations.

Should I fix copper weight or trace width first?

Copper weight. It is driven by current and thermal needs and is expensive to change; trace width has routing flexibility. Locking width first and then being forced into heavy copper lands you in the hardest combination.