When a PCB needs conformal coating

Someone on the project has said the words “we should probably coat it,” and now the decision is yours. The quote from the assembler asks which chemistry and where the keep-outs are, the test engineer wants to know if the pogo pins will still make contact, and nobody can say what happens when a board comes back for repair. Conformal coating is a cheap-sounding line item that quietly changes the layout, the test plan and the service life of every board it touches.

What coating is for, and when it is not needed

A conformal coating is a thin polymer film applied over the assembled board. It keeps moisture, dust, salt and mild contamination off the copper and the component leads, and it adds insulation between closely spaced conductors. It is not a potting compound, not a mechanical shield and not a substitute for an enclosure. If water can pool on the board, coating slows corrosion; it does not stop a submerged board from failing.

The honest test is the environment. A board that lives inside a sealed, indoor product on a desk gains very little from coating and pays every one of its costs. Boards that condense, that ride in vehicles, that sit in an unheated outbuilding, that see salt air or agricultural dust, or that switch enough voltage that surface leakage matters, are the ones that earn it. The coating vendors say the same thing in their own guides: the decision turns on the environment, the sensitivity of the circuit and the durability the product needs, not on a general sense that coated is better (Chemtronics, The Ultimate Guide to Conformal Coating).

A common middle case is a product that will be sold indoors but occasionally used outdoors, or an industrial controller in a cabinet that is dry until someone hoses down the floor. Those are the ones to think hardest about. Coating covers the failure mode that a field return would otherwise teach you about a year in.

Picking a chemistry

Liquid coatings come in a handful of families, and the material vendors classify them the same way the qualification standard does. The standard most specifications cite, IPC-CC-830, defines acrylic, epoxy, silicone and urethane classes and puts each material through insulation resistance, thermal shock, flexibility, flammability and fungus tests before its maker can call it qualified (HumiSeal quick guide to IPC-CC-830B). Buying a coating that its manufacturer has qualified to that standard is the cheapest assurance available; it says nothing about your assembler’s process, only about the material.

Family	Strength	Cost you pay
Acrylic	Easy to apply, fast to dry, easy to remove with mild solvent, can be soldered through	Weak against solvents and fuel vapor
Urethane	Strong chemical and moisture resistance, abrasion resistant	Hard to remove; long full cure
Silicone	Wide temperature range, flexible, good for vibration and thermal cycling	Removal needs specialized solvent and agitation
Parylene	Very thin, uniform vapor-deposited film reaching under parts	Batch vacuum process, no rework by dissolving

For a small-batch product with any chance of field repair, acrylic is the default and the others need a reason. An acrylic like MG Chemicals 419D is tack free in about ten minutes, dries in half an hour with mild heat, and its maker states it can be soldered through for rework and removed with a listed thinner (MG Chemicals 419D product page). That combination is what keeps a coated board serviceable by a technician with an iron and a bottle of solvent.

Urethane and silicone are the right answer when the environment demands them: fuel or oil exposure for urethane, sustained high temperature or heavy vibration for silicone. Accept up front that a urethane-coated board that needs a component replaced is a project, not a bench job. Parylene belongs on dense, high-reliability assemblies where thin uniform coverage matters more than anything else and where nobody expects to rework.

Thickness is a real specification, not a detail for the operator. Liquid coatings are typically applied to a dry film in the tens of micrometers, with the vendors giving a working window of roughly 25 to 75 µm for acrylic and urethane and a somewhat wider one for silicone (HumiSeal thickness and standards guide). Too thin leaves pinholes and bare edges on tall parts. Too thick traps solvent, bubbles and cracks. Most coatings carry a UV tracer so an inspector can see coverage under a lamp; that inspection is only meaningful if the drawing says what thickness and coverage was wanted.

What coating does to test and rework

This is the part that catches teams the first time. A coating is an insulator, and it does exactly that job on a test point. A pogo pin pressed onto a coated pad reads open, or reads intermittently, which is worse. Bed-of-nails and functional test have to happen before coating, or the test points have to be masked so they stay bare. Either choice has consequences.

Testing before coating means any failure found after coating (a bad joint that opens under thermal stress, a part that dies in burn-in) requires removing coating to probe it. Masking the test points keeps the fixture useful for life, at the cost of a masking step on every board and a small bare spot the coating was supposed to cover. Connectors, card-edge fingers, switches, trimmers, mounting holes that carry ground through the screw, thermal pads that mate with a heatsink, and any pressure-fit or spring contact all need the same masking decision, and the assembler needs a drawing that shows which is which (I-Connect007 on masking during assembly and rework). Masking is done with precut tape dots, hand-cut tape, peelable latex, or reusable boots for connectors, and every one of those methods is easier when there is room around the feature to apply it and nothing tall crowding it.

Rework on a coated board always starts with taking the coating off the work area. With acrylic that is a solvent pen or a short soak, then solder, clean and recoat the spot. With urethane or silicone it is a longer soak and mechanical help. Thermal burn-through with the iron works on acrylic and leaves residue that has to be cleaned; on other chemistries it is a last resort. Plan the service procedure when you choose the chemistry, not after the first return.

Surface cleanliness is the other trap. Coating over flux residue, fingerprints or oil produces dewetting and delamination that shows up as bare patches under UV, and later as corrosion exactly where the residue was. If the paste is no-clean, either qualify the coating over that specific residue or add a wash before coating. The defect list that vendors publish (dewetting, bubbles, fisheyes, orange peel, cracking) reads as a list of process shortcuts, and most of them trace to contamination, thickness, or not letting solvent flash off between coats.

Designing the board for it

A board that will be coated should say so in the design files, not in an email. KiCad has no dedicated coating layer, and designers on the KiCad forum have settled on drawing keep-out regions on a spare user layer and naming it plainly. KiCad 9 lets you add user-defined layers and rename them in the board setup, so a layer called something like Coating.Keepout is visible in the fabrication drawing and in the exported Gerbers (KiCad 9 PCB Editor documentation). Put an outline around every masked feature, and put a note on the fab drawing naming the coating family, the target thickness, whether test points are to be masked, and which side or sides get coated.

Layout choices that make masking easier are worth making early. Group connectors and test points instead of scattering them. Keep small passives away from the sides of tall connectors so a boot can seat. Avoid deep pockets between tall parts where liquid coating pools and traps air. Keep vias away from the edge of a masked area so coating does not wick under the tape. Those are practical observations that coating houses publish because they see the same mistakes on every third board; treat their exact clearance numbers as the preference of that house rather than a universal rule, and ask your own assembler what their masking method needs.

Also read the datasheets of the parts being coated. Some sensors, MEMS microphones, pressure ports, crystals in open cans, and anything with a vent hole are damaged or detuned by coating and must be masked. Battery contacts and creepage paths for high-voltage isolation are cases where coating helps but only if it is complete; a partial film across a creepage gap can be worse than none.

We treat coating as a design decision with a test plan attached. If you send us the environment the product lives in, the service expectation, and the board files, we will tell you whether it needs coating at all, and if so which family, where the keep-outs go, and how the board will be tested and repaired afterward.