Most conformal coating problems are not coating problems. They are design and process decisions that were made months earlier, before the first board ever reached a coating machine.
In coating projects we review with EMS and OEM teams, the same failure patterns appear again and again: coating on connectors, uncured material under tall components, thickness that measures well on the coupon but fails on the board, and inspection that cannot see the defects it is supposed to catch. None of these are caused by a bad material. They are caused by decisions that were never made explicitly.
This article takes the opposite approach to a standard "what is conformal coating" guide. Instead of describing the chemistry, it lists the seven decisions that determine whether the process will hold up in volume production, and the questions a design or process engineer should answer before releasing a board to the coating line.
Coating selection usually starts with a material name: acrylic, silicone, polyurethane. That is the wrong starting point. The starting point is the environment the product will actually see in the field.
A board that spends ten years in a humidity-controlled cabinet and a board that is mounted on an outdoor cabinet in a coastal climate do not need the same coating, even if the circuit is identical. Salt mist, condensing humidity, sulfur-bearing atmosphere, cleaning chemicals, high voltage, and thermal cycling each attack the assembly in a different way, and each implies a different film property.
The practical rule is to write the environment first, then the protection target, then the material. If the material is chosen first, the process is usually tuned to fit the material, and the original protection requirement quietly disappears from the specification.
Every conformal coating process has areas that must remain uncoated: connectors, test points, press-fit pins, optical sensors, relays, and any surface that must make electrical contact or move.
When these areas are defined during design, they can be handled in the coating program, in the stencil-like masking fixture, or in the panel layout. When they are discovered on the production floor, they are handled with tape, silicone plugs, and manual labour, and the cost is paid on every single board.
The tell-tale sign of a board that was not designed for coating is a masking operation with more than a handful of manual steps. At that point the coating process is no longer automated, no matter how good the machine is.

Coating does not travel around corners. Tall components create shadowed areas where spray cannot reach, and dense clusters create capillaries that draw material away from where it is needed.
Two parameters control this in a selective coating machine: the angle of the valve relative to the board, and the distance to the surface. A vertical spray covers flat areas well but leaves the sidewalls of tall parts dry. An angled valve reaches sidewalls and the shadowed strip between tall components, but produces a thinner film on flat areas if the same program is reused.
This is why multi-axis machines with programmable valve tilt exist. It is also why a coating program written for one board layout rarely transfers to another without adjustment. If a process relies on "spray everything from above and hope", the shadowing problem will appear as a corrosion failure one or two years later, not as a defect on the line.

UV-curable coatings are attractive because they cure in seconds. But UV light only reaches what it can see. Under components, inside connectors, and in the shadow of tall parts, the coating stays wet unless the chemistry also has a secondary moisture or thermal cure.
The decision therefore is not "UV or thermal". It is "what is the shadowed-area cure path, and how long does it take". A line designed for a thirty-second UV cure and a five-minute dwell will fail if the shadowed cure needs four hours, because boards will be handled and packed long before the film is fully cured.
The same logic applies to infrared and convection curing. IR heats the surface quickly and suits thin films on exposed areas; convection delivers heat into cavities and is more forgiving for dense assemblies. Choosing the oven before the chemistry is a reliable way to create a line that runs fast and fails slowly.
"Coating thickness: 50 µm" is not a specification. It is a target without boundaries. What production needs is a window: a minimum thickness that guarantees protection, and a maximum that avoids cracking, bubbling, and interference with connectors and housings.
Thickness depends on valve type, spray pattern, stroke speed, material viscosity, and the number of passes. Change any one of them and the film changes. A stable process is one where the window is known, the measurement method is defined, and the machine parameters that keep the process inside the window are documented.
Equally important is where thickness is measured. Coupons, witness plates, and actual board locations can give different results. If the customer's acceptance criterion is based on a specific location on the assembly, measuring somewhere else produces paperwork that looks correct and protection that is not.
Inspection is often specified after the process is designed, which is backwards. The inspection method decides what defects will be caught, and what defects will reach the customer.
A realistic inspection plan combines these methods and states which defect each one is responsible for. When that mapping is missing, inspection becomes a ritual that generates data nobody uses.

Every coating line eventually produces a board that needs repair. How that board is repaired depends on the material: acrylic and UV coatings can often be peeled or dissolved locally, silicone is difficult but possible, polyurethane and epoxy are hard, and parylene requires mechanical removal.
If the repair method is not agreed during the design phase, the first defective board becomes a production stop. Rework is also where coating quality standards are most often broken: a board that is repaired, re-coated by hand, and returned to the line without thickness or cure verification is a warranty claim waiting for a date.
The simplest control is to define, in writing, which components may be reworked, which coating removal method is approved, and what inspection is required after re-coating.
The seven decisions above can all be tested on a small number of boards. A practical validation sequence looks like this:
This sequence takes days, not months, and it prevents the most expensive outcome in a coating project: a line that is installed, commissioned, and then re-engineered because the process window was never defined.

Conformal coating works when it is designed as part of the product, not added to it at the end of the process. Environment, keep-out areas, component shadowing, cure method, thickness window, inspection mapping, and rework path are decisions, and each one has a cost that is paid either during design or during production.
The same principle applies to equipment selection. A coating machine is only as good as the process window it is asked to hold, which is why the HP-730 selective conformal coating machine and the automated conformal coating line are specified together with the material, the program, and the inspection method rather than as standalone hardware.
Boards that follow this path can be validated in advance; the automotive electronics conformal coating case study shows how a double-sided production line was defined from the process window outward. If you are at the design stage, a coating sample trial on your own assemblies is the fastest way to find out which of the seven decisions your current design has already made for you.
Questions about a specific board or material? Reach our engineers through the Haipai contact page and send the layout plus the end-use environment; we will come back with a process proposal, not a catalogue page.
Most conformal coating problems are not coating problems. They are design and process decisions that were made months earlier, before the first board ever reached a coating machine.
In coating projects we review with EMS and OEM teams, the same failure patterns appear again and again: coating on connectors, uncured material under tall components, thickness that measures well on the coupon but fails on the board, and inspection that cannot see the defects it is supposed to catch. None of these are caused by a bad material. They are caused by decisions that were never made explicitly.
This article takes the opposite approach to a standard "what is conformal coating" guide. Instead of describing the chemistry, it lists the seven decisions that determine whether the process will hold up in volume production, and the questions a design or process engineer should answer before releasing a board to the coating line.
Coating selection usually starts with a material name: acrylic, silicone, polyurethane. That is the wrong starting point. The starting point is the environment the product will actually see in the field.
A board that spends ten years in a humidity-controlled cabinet and a board that is mounted on an outdoor cabinet in a coastal climate do not need the same coating, even if the circuit is identical. Salt mist, condensing humidity, sulfur-bearing atmosphere, cleaning chemicals, high voltage, and thermal cycling each attack the assembly in a different way, and each implies a different film property.
The practical rule is to write the environment first, then the protection target, then the material. If the material is chosen first, the process is usually tuned to fit the material, and the original protection requirement quietly disappears from the specification.
Every conformal coating process has areas that must remain uncoated: connectors, test points, press-fit pins, optical sensors, relays, and any surface that must make electrical contact or move.
When these areas are defined during design, they can be handled in the coating program, in the stencil-like masking fixture, or in the panel layout. When they are discovered on the production floor, they are handled with tape, silicone plugs, and manual labour, and the cost is paid on every single board.
The tell-tale sign of a board that was not designed for coating is a masking operation with more than a handful of manual steps. At that point the coating process is no longer automated, no matter how good the machine is.

Coating does not travel around corners. Tall components create shadowed areas where spray cannot reach, and dense clusters create capillaries that draw material away from where it is needed.
Two parameters control this in a selective coating machine: the angle of the valve relative to the board, and the distance to the surface. A vertical spray covers flat areas well but leaves the sidewalls of tall parts dry. An angled valve reaches sidewalls and the shadowed strip between tall components, but produces a thinner film on flat areas if the same program is reused.
This is why multi-axis machines with programmable valve tilt exist. It is also why a coating program written for one board layout rarely transfers to another without adjustment. If a process relies on "spray everything from above and hope", the shadowing problem will appear as a corrosion failure one or two years later, not as a defect on the line.

UV-curable coatings are attractive because they cure in seconds. But UV light only reaches what it can see. Under components, inside connectors, and in the shadow of tall parts, the coating stays wet unless the chemistry also has a secondary moisture or thermal cure.
The decision therefore is not "UV or thermal". It is "what is the shadowed-area cure path, and how long does it take". A line designed for a thirty-second UV cure and a five-minute dwell will fail if the shadowed cure needs four hours, because boards will be handled and packed long before the film is fully cured.
The same logic applies to infrared and convection curing. IR heats the surface quickly and suits thin films on exposed areas; convection delivers heat into cavities and is more forgiving for dense assemblies. Choosing the oven before the chemistry is a reliable way to create a line that runs fast and fails slowly.
"Coating thickness: 50 µm" is not a specification. It is a target without boundaries. What production needs is a window: a minimum thickness that guarantees protection, and a maximum that avoids cracking, bubbling, and interference with connectors and housings.
Thickness depends on valve type, spray pattern, stroke speed, material viscosity, and the number of passes. Change any one of them and the film changes. A stable process is one where the window is known, the measurement method is defined, and the machine parameters that keep the process inside the window are documented.
Equally important is where thickness is measured. Coupons, witness plates, and actual board locations can give different results. If the customer's acceptance criterion is based on a specific location on the assembly, measuring somewhere else produces paperwork that looks correct and protection that is not.
Inspection is often specified after the process is designed, which is backwards. The inspection method decides what defects will be caught, and what defects will reach the customer.
A realistic inspection plan combines these methods and states which defect each one is responsible for. When that mapping is missing, inspection becomes a ritual that generates data nobody uses.

Every coating line eventually produces a board that needs repair. How that board is repaired depends on the material: acrylic and UV coatings can often be peeled or dissolved locally, silicone is difficult but possible, polyurethane and epoxy are hard, and parylene requires mechanical removal.
If the repair method is not agreed during the design phase, the first defective board becomes a production stop. Rework is also where coating quality standards are most often broken: a board that is repaired, re-coated by hand, and returned to the line without thickness or cure verification is a warranty claim waiting for a date.
The simplest control is to define, in writing, which components may be reworked, which coating removal method is approved, and what inspection is required after re-coating.
The seven decisions above can all be tested on a small number of boards. A practical validation sequence looks like this:
This sequence takes days, not months, and it prevents the most expensive outcome in a coating project: a line that is installed, commissioned, and then re-engineered because the process window was never defined.

Conformal coating works when it is designed as part of the product, not added to it at the end of the process. Environment, keep-out areas, component shadowing, cure method, thickness window, inspection mapping, and rework path are decisions, and each one has a cost that is paid either during design or during production.
The same principle applies to equipment selection. A coating machine is only as good as the process window it is asked to hold, which is why the HP-730 selective conformal coating machine and the automated conformal coating line are specified together with the material, the program, and the inspection method rather than as standalone hardware.
Boards that follow this path can be validated in advance; the automotive electronics conformal coating case study shows how a double-sided production line was defined from the process window outward. If you are at the design stage, a coating sample trial on your own assemblies is the fastest way to find out which of the seven decisions your current design has already made for you.
Questions about a specific board or material? Reach our engineers through the Haipai contact page and send the layout plus the end-use environment; we will come back with a process proposal, not a catalogue page.