A batch of coated boards comes back from the customer rejected for appearance. Nobody disputes that the coating looks wrong. The argument is about what "wrong" means, and it splits into two questions that sound similar and are not: is this defect a reliability risk, and is it an acceptance failure under the class the order was placed against. A supplier can answer the first with a coating expert and the second only with the standard.
In practice the two get conflated, and both parties lose. A buyer rejects a board for orange peel that IPC-A-610 Class 2 would accept, and pays for a recoat that adds a second thermal excursion to a board that was already fine. Or a buyer accepts a band of dewetting along a connector that Class 3 treats as a reject, and discovers the problem eighteen months later as dendrite growth between two fine-pitch pads. The cost difference between those two mistakes is roughly three orders of magnitude.
This guide covers the six defects that account for most coating rejections, the process parameter behind each one, and how the verdict changes between Class 2 and Class 3. It assumes the board has already passed its pre-coat cleanliness check, because a coating defect on a contaminated surface is a different failure mode entirely.
Six Defects You Will Actually See
Coating literature lists dozens of defect names, most of which are regional vocabulary for the same physical event. The six below cover the overwhelming majority of what appears on a rejection report, and each one has a distinct cause family, which is what makes the name useful.
| Defect | What the inspector sees | Cause family | Reliability impact |
|---|---|---|---|
| Orange peel | Uniform pebbled texture, gloss reduced, film thickness normal | Atomisation and solvent balance | Cosmetic unless severe; can trap solvent |
| Fisheye | Circular crater with a raised rim, bare or thin centre, random positions | Surface contamination, usually silicone or oil | Latent: a thin spot is an uncoated spot |
| Dewetting | Coating pulls back from an edge, leaving a clean wedge of bare laminate or mask | Surface energy, contamination, cure profile | Latent: exposed conductor at the boundary |
| Pinhole | Pinpoint hole through the film down to the surface, often in clusters | Air or solvent release from below, or a particle | Latent: a direct path to the conductor |
| Bubbles and blisters | Domed film over a void, sometimes with a visible cavity underneath | Outgassing from vias, component gaps, or entrapped solvent | Latent: bubble collapse exposes the surface |
| Crazing and cracking | Fine network of cracks, usually at film edges or over tall components | Film stress, over-cure, thermal expansion mismatch | Latent, and progressive under thermal cycling |
Notice that only one of the six is genuinely cosmetic. That is the single most useful fact in this article: orange peel is a finish problem, and the other five are coverage problems wearing a cosmetic disguise. A rejection report that lumps them together as "coating appearance" is telling a buyer nothing about the risk in the shipment.
Orange Peel: Viscosity, Atomisation and the Dry-Film Trap
Orange peel is the film freezing in place before it can level. A sprayed droplet lands, begins to flow out under surface tension, and then the solvent leaves fast enough that viscosity rises past the point where flow is possible. Whatever texture the spray pattern left is now permanent. The defect is therefore a competition between two rates: the rate at which the film can level, and the rate at which it loses solvent.
Three parameters move that competition. Viscosity at the nozzle is the first: a coating at the top of its viscosity window levels more slowly than the same chemistry at the bottom, and many workshops run high because it reduces runs and sags on vertical surfaces. Atomisation pressure is the second: too low produces large droplets that leave a coarse texture, and too high produces a fine mist that flashes off solvent in the air before it lands, which is the classic over-atomisation orange peel. Flash-off time between passes is the third, and it is the one most often compressed when a line is pushed for throughput.
| Parameter | Typical window | Orange peel direction |
|---|---|---|
| Viscosity at application | Per supplier TDS, often 150-400 cps for spray acrylics | High end increases peel |
| Atomisation pressure | 1.5-3.0 bar for HVLP, per gun and coating | Both extremes increase peel |
| Flash-off between passes | 60-180 s depending on solvent | Too short traps solvent, too long stops flow |
| Board temperature at spray | 20-30 degrees Celsius | Warm boards accelerate solvent loss |
| Wet film per pass | Half to one third of final target | Thin passes level better than one heavy pass |
The practical test is a wet film comb or gauge immediately after spraying, read against the dry target. A film that measures on target dry but was applied warm and fast will show peel, because the same dry thickness arrived in fewer, heavier passes with less leveling time between them. Where orange peel appears only on boards sprayed late in a shift, the cause is usually board temperature from an upstream oven rather than anything the operator changed.
Orange peel becomes a real problem in one specific case: when the texture is deep enough to shadow a subsequent layer or to interfere with a connector or shield mating surface. Otherwise the film is continuous, and continuity is what the standard is about.
Fisheye and Dewetting: the Contamination Chain
Fisheye and dewetting are the same physics at different scales. A region of the surface has lower surface energy than its surroundings, so the coating pulls away from it. If the region is small and isolated, the film retracts into a crater with a raised rim and the result is called a fisheye. If the region is a band along an edge or around a keep-out, the film retreats as a front and the result is called dewetting. Both mean the same thing: something is on the board that the coating does not want to wet.
The contamination chain has a consistent order of likelihood, and working through it in that order saves a lot of guessing:
- Silicone, from mould release agents, thermal interface material, silicone masking boots, or a lubricant aerosol used anywhere in the same room. Silicone migrates through air and deposits on surfaces, which is why a single fisheye cluster can appear on boards that never touched a silicone part.
- Flux residue in the form of a thin, uneven film that passed ionic cleanliness because ionic testing measures conductivity, not surface energy. A board can be ionically clean and still dewet.
- Handling oils from bare-hand contact, which concentrate along the edges where hands hold the board and produce exactly the edge-band dewetting pattern that gets blamed on the coating.
- Mask residues from adhesive tapes and latex-free masking dots, where the adhesive leaves a low-energy footprint larger than the mask itself.
- Mould release on the laminate edge, a fab-side residue that only shows up where the coating crosses the routed edge.
The diagnostic that separates these is a surface energy check on the bare board before coating, using dyne pens or a contact angle measurement at the specific locations where the defect appears. A board that measures below roughly 38 dynes per centimetre in the defect zone has a contamination problem upstream of the coater, and no amount of parameter adjustment at the spray station will fix it. This is the point at which cleanliness verification has to go beyond ionic testing, a distinction covered in our guide to ionic contamination and cleanliness testing.
Dewetting along a connector edge deserves separate mention because it is partly a masking problem. A masking boot that is removed after cure leaves an adhesive footprint, and the coating applied on the next build pulls back from it. The result is a clean straight line of bare board exactly where the coating most needs to seal against the connector body. Our guide to coating masking techniques covers which mask types leave a residue footprint and which do not.
Pinholes and Bubbles: Where the Air Comes From
A pinhole is a hole in the film that goes all the way down. A bubble is a void under an intact film that will become a pinhole when it collapses or when thermal cycling opens it. The distinction matters for inspection because a bubble can pass a visual check under a gloss lamp and fail six months later, while a pinhole under UV inspection is obvious. The physical source of both is gas: either gas that came out of the board, or gas that came out of the coating.
| Source | Signature | Control |
|---|---|---|
| Entrapped air in via barrels and blind holes | Cluster along a via field or under a BGA | Vacuum defoaming, slower first pass, or a low-viscosity primer |
| Trapped solvent in a thick film | Bubbles distributed through the film, worse in high-build areas | Thinner passes, longer flash-off, staged cure ramp |
| Moisture in the laminate | Blistering after cure, localised and often on one board type | Bake before coating per J-STD-033 handling |
| Gaps under components | Ring of voids around large SMD bodies | Flow-out allowance, or underfill where the gap is structural |
| Particulate on the surface | Isolated single pinhole with a visible particle | Pre-coat inspection and blow-off |
The via-barrel source is the one that catches people out, because it is a function of board design rather than coating process. A via that is tented on one side and open on the other traps air during dip or spray, and the air escapes through the wet film as it cures. The same board coated by selective application, where the via field is outside the coated zone, has no pinholes at all. Where that is not possible, a first pass at reduced film build followed by a full pass lets the air escape while the film is still thin enough to re-close.
What IPC-A-610 Class 2 and Class 3 Actually Say
This is where most disputes should be settled, and where the standard is more permissive than inspectors often assume. The coating section of IPC-A-610 does not require a cosmetically perfect film; it requires coverage in defined areas, absence of specific defect types in those areas, and evidence that the specified thickness was achieved. The table below is a working summary for a buyer writing a purchase order, not a substitute for the clause text.
| Condition | Class 2 verdict | Class 3 verdict | Action |
|---|---|---|---|
| Orange peel, film continuous and on thickness | Accept | Accept | None. Do not reopen. |
| Fisheye with visible bare surface over a conductor | Reject in coverage area | Reject | Touch-up or recoat |
| Fisheye over solder mask only, no conductor exposed | Evaluate against coverage map | Reject if inside keep-in zone | Judgement call; document it |
| Dewetting at a connector boundary | Accept if outside coverage area | Reject if the boundary enters a keep-in zone | Zone definition decides |
| Pinhole exposing a conductor | Reject | Reject | Touch-up |
| Bubbles under 0.5 mm, not bridging conductors | Generally accept | Evaluate; reject if over a conductor gap | Document, monitor trend |
| Cracking through the film | Reject | Reject | Strip and recoat, investigate cure |
| Coating on a connector contact surface | Reject | Reject | Remove, re-qualify the assembly |
Two rows in that table cause most of the arguments. The fisheye over solder mask is a genuine grey area, because the standard is written around coverage of surfaces that need protection and a solder-mask-covered trace with coating over the rest of the board may still be protected. The connector boundary row is not a grey area at all, but it is treated as one whenever the coverage map was never defined in writing. Both are resolved by the same document: a keep-in and keep-out map agreed at NPI, which is the substance of our guide to IPC-A-610 conformal coating acceptance.
Class 3 is not simply "Class 2 with a better finish". The practical difference for coating is that Class 3 tends to remove the "evaluate" verdicts and replace them with a defined limit, and it expects a documented inspection method rather than a visual walk-by. A supplier quoting Class 3 pricing without a written inspection method and a UV or imaging record is quoting the price but not the class.
Recoat, Strip or Concession: the Three-Path Decision
Once a defect is confirmed, there are only three paths, and the choice is economic as much as technical. The mistake buyers make is defaulting to recoat, which sounds like the conservative option and is often the most damaging.
- Touch-up is appropriate for isolated pinholes and small fisheyes outside dense areas. It is fast, but it is also the least verifiable path, because a brushed touch-up has no controlled thickness. Require the touch-up material to be the same chemistry, require it to be recorded per serial number, and require a cure verification if the product is Class 3.
- Strip and recoat removes the film chemically and applies a fresh one. This adds a full thermal cycle and a solvent exposure to an assembly that has already been through reflow, and on boards with fine-pitch parts it risks the very migration the coating exists to prevent. It is the right path for cracking or for coverage failures that are systematic across the lot, and usually the wrong path for a single cosmetic defect.
- Concession with documentation is the path nobody proposes out loud and the one that is sometimes correct. If the defect is outside the coverage map, Class 2 accepts it, and the only thing wrong is that the drawing implied something stricter than the standard requires. A documented concession is cheaper than a recoat and does not touch the hardware. It requires a written engineering judgement on file, not a verbal agreement on a phone call.
The cost comparison is not close. Touch-up on a single board is minutes of work. A strip and recoat cycle is a chemical process, a second cure, a re-inspection and a scrap risk on every board it touches, and on a selective coating line it also consumes programming time. Where a defect is systematic, the recoat will reproduce it unless the parameter that caused it was changed first, which is why the recoat decision should never be taken before the cause is named.
Cracking and Crazing: the Defect That Gets Worse
Crazing deserves its own section because it behaves differently from the other five: it is not stable. A crazed film looks like a network of fine lines, usually radiating from an edge or concentrated over a tall component, and under thermal cycling those lines propagate until the film is discontinuous. The failure is progressive, so a board that passes inspection at time zero will fail in the field, and the inspection result at the factory is not evidence of anything.
Four causes dominate. Over-cure makes the film brittle and raises its glass transition behaviour past the point where it can accommodate movement. Under-cure leaves solvent in the film, which plasticises it initially and then, as the solvent slowly leaves in service, contracts the film and puts it in tension. Coating over a tall component or a connector body creates a film that has to bridge a step, and the film thins at the step and cracks there first. And a coating whose coefficient of thermal expansion is mismatched against a heavy copper plane will crack over the plane boundary when the assembly cycles, which is why cracking clusters over power planes and not over signal routing.
The acceptance consequence is blunt: cracking through the film is a reject at both Class 2 and Class 3, with no grey area. The process response is a cure verification rather than a visual check. A cure that is verified by FTIR against the supplier's reference spectrum, or by a solvent rub to a defined standard, is the only evidence that the film will not craze in service. Film that is merely dry to touch provides none. Where the coating chemistry is being selected in the first place, the comparison of chemistries and their stress behaviour is in our guide to coating types compared.
What to Put in the Purchase Order
Most coating disputes are not engineering disputes. They are specification disputes, and they are settled before production by seven clauses. A purchase order that contains all seven makes the table above enforceable; one that says "conformal coat per IPC-A-610 Class 2" makes nothing enforceable, because it delegates the coverage map, the defect list and the inspection method to whoever interprets it.
- Class and standard revision, named explicitly, with the revision letter and date. "Class 2" alone has meant different things across revisions.
- Coverage map: which zones must be coated, which must not, and which are conditional. As a drawing, not a sentence.
- Material and supplier, with the specific product code, because acceptance criteria for adhesion and thickness are chemistry-specific.
- Thickness target and tolerance per zone, with the measurement method named, since wet film, eddy current and cross-section give different numbers on the same board.
- Cure verification method, named, with the acceptance evidence specified as a record that ships with the lot.
- Defect verdicts for the grey areas: fisheye over mask only, bubbles under a stated size, dewetting at connector boundaries. Decide these once, in writing, instead of per shipment.
- Touch-up policy: permitted or not, by defect type, with the recording requirement and the re-inspection rule.
The seventh clause is the one that changes behaviour. A supplier that knows touch-up must be recorded and re-inspected will find the cause rather than brush over the symptom, because brushing costs more paperwork than adjusting a flash-off timer. Where the defect rate is high enough to be systematic, the route is through the qualification records rather than the shipping inspection, which is the subject of our guide to IPC-CC-830 coating qualification.
One further note for buyers of vacuum-environment or optical assemblies: a film with entrapped solvent can pass every visual criterion and still outgas in service. The TML and CVCM limits in that environment are a material property issue rather than an appearance issue, and they are covered in our guide to coating outgassing and TML/CVCM limits.
Frequently Asked Questions
Is orange peel ever a rejection? Under IPC-A-610 it is not, provided the film is continuous and meets the thickness target. It becomes a rejection only when the texture is deep enough to breach an adjacent coating layer or to prevent a connector or shield from seating, both of which are mechanical rather than cosmetic conditions.
How many fisheyes are too many? There is no count-based limit. The question is whether any fisheye exposes a conductor or falls inside a keep-in zone. Two fisheyes over a ground plane in a coated zone are a reject; twenty over solder mask at the far edge of the board may be acceptable in Class 2. Count without position tells you nothing.
Can a fisheye be recoated locally? Only if the coating is compatible and the surface under the fisheye is cleaned to the same standard as the original board. A touch-up applied over a contaminated spot fails the same way the original film did, and usually faster.
Why do pinholes cluster over a via field? Because the air has to come from somewhere, and the via barrels hold it. Clustering is diagnostic: it points to board design and board handling, not to the spray operator.
Does a thicker coat fix pinholes? No. A thicker film holds more solvent and takes longer to cure, which increases bubble formation. Thinner passes with proper flash-off, or vacuum defoaming of the coating, address the cause.
What inspection method should a Class 3 order require? Visual inspection under controlled lighting with a defined acceptance list, plus a coverage verification method that produces a record, plus a cure verification. A visual walk-by under a bench lamp satisfies none of the three.
