Process & DFM
How Surface Finish Changes Your Tolerances
The short answer
Most finishes add material to every surface they touch, and the drawing dimension is usually machined before the finish is applied. Type II anodizing adds roughly 0.005 to 0.013 mm per surface, hard anodize adds considerably more, plating adds whatever thickness you specified, and powder coat is thicker than all of them. On a bore or a slot the growth lands on both walls, so the feature closes by twice the coating thickness. Either machine the feature oversize by the known build-up or mask it, and say which on the drawing.
The assumption that breaks the part
A drawing says a bore is 12.00 mm with a tolerance of plus or minus 0.02. The shop machines it, measures it, and it passes. The part goes out for hard anodize and comes back at 11.93 mm. Nothing was done wrong, and the pin no longer fits.
Almost every finish is additive. Anodizing converts the surface of the aluminum into oxide, and that oxide occupies more space than the metal it replaced, so the part grows outward by roughly half the total oxide thickness. Plating deposits metal on top. Powder coat sprays a polymer film. Only a few processes, such as passivation or electropolishing, leave the part the same size or take a little off.
The second half of the problem is order of operations. Finishing happens after machining, usually at a different facility, and the person hitting your dimension is not the person applying the coating. Unless the drawing says otherwise, the machinist will hit the number you wrote, which means the number you wrote has to already account for the growth.
What each finish adds
Per coated surface, in the ranges you should design around. Confirm exact build-up with the finisher when it matters, since thickness varies with specification, rack position and geometry.
| Finish | Typical build-up per surface | Effect on a bore or slot | Behavior worth knowing |
|---|---|---|---|
| Type II anodize | About 0.005 to 0.013 mm | Closes by roughly double | Even and predictable on simple geometry. Color adds nothing dimensionally. |
| Type III hard anodize | Roughly 0.025 to 0.05 mm, sometimes more | Closes by roughly double, enough to kill a press fit | The one that most often breaks an assembly. Half the coating grows outward, half grows into the metal. |
| Electroplating (nickel, zinc, chrome) | Whatever you specify, commonly 0.005 to 0.025 mm | Closes by roughly double | Builds unevenly. Deposits heavier on edges and corners, lighter deep inside holes. |
| Powder coat | Roughly 0.06 to 0.15 mm, sometimes more | Effectively fills small features | Thick, and it will not reach deep into a hole. Threads must be masked. |
| Wet paint | Thinner than powder, varies by system | Small but real | Multiple coats stack. Ask for the total film thickness, not the per-coat number. |
| Passivation | None | None | Chemical, not additive. Safe on tight features. |
| Electropolish | Removes material | Opens slightly | The one that goes the other way. Removes a small amount and rounds sharp edges. |
| Bead blast | Removes a little, roughens the surface | Slightly opens, and changes the measured roughness | Usually a precursor to another finish rather than a final state. |
Two numbers matter and they are not the same: total oxide thickness and dimensional growth. On anodize, growth is roughly half the oxide thickness, because the conversion eats into the base metal as it builds outward.
A worked example: the bore that stopped fitting
Take a 12 mm bore in a 6061 housing that has to take a dowel pin as a light press fit, and say the design calls for Type III hard anodize on the whole part.
Assume a hard anodize specification giving about 0.038 mm of dimensional growth per surface. The bore has two opposing walls, so the diameter closes by about 0.076 mm. If the print calls the bore 12.00 mm and the machinist hits it, the finished bore is about 11.92 mm, which is not a light press fit. It is an interference fit that either will not go together or will gall on assembly. The engineering fits chart gives the clearance each class of fit needs.
There are three honest ways out.
Machine it oversize. Call the bore at 12.08 mm before finish and note the finished requirement. Now the machinist knows what to hit and the inspector knows what to check.
Mask the bore. The finisher plugs it, the bore stays bare aluminum, and the dimension holds. This costs money per part and leaves an uncoated surface, which may be unacceptable for corrosion or wear.
Machine after finishing. Anodize the part, then ream the bore to size. The bore ends up bare regardless, and it adds an operation, but it gives the tightest control.
Which you pick depends on whether the bore needs to be coated. What you cannot do is write 12.00 mm, specify hard anodize everywhere, and expect both.
Threads, blind holes and the places coatings behave badly
Threads. Coating lands on both flanks, so the effective pitch diameter closes by roughly twice the build-up, and threads are the single most common assembly failure after finishing. Either specify the thread to be masked, or machine it to a class that leaves room. Powder coat in a thread is not a tolerance problem so much as a re-tapping bill.
Blind and deep holes. Plating throws poorly. Coating is thick at the mouth and thin at the bottom, which means a deep hole is neither properly protected nor properly sized. If the hole matters, mask it and treat it as bare.
Sharp external edges. Plating builds up heavier on edges and corners, so a sharp corner gets a bead of extra material. A small break or chamfer on the edge gives a more even coat and a better-looking part.
Grounding and mating faces. Anodize is an insulator. If any surface needs electrical continuity, such as a chassis bond or a grounding lug, it has to be masked, and that has to be on the drawing rather than assumed.
Flatness on thin parts. Powder coat is cured at temperature. A thin sheet metal panel can come back with a bow it did not have, which is a flatness problem no dimension on your drawing predicted.
What to put on the drawing
Most of this disappears if the drawing answers four questions the finisher and the machinist both read.
State whether dimensions are before or after finish. A single note, such as "all dimensions after finish unless noted", removes the ambiguity that causes the failure.
Call the finish by its specification, not its name. "Anodize" is not a specification. Give the type, class and color, for example Type II Class 2 black per MIL-A-8625. For plating, give the alloy and the thickness. For powder, give the system and the color standard.
Mark what gets masked. Threads, bores, bearing seats, grounding pads. Call them out individually rather than writing "mask as required".
Say which surfaces are cosmetic. A finisher racks the part somewhere, and the rack leaves a mark. If you do not say where the mark may go, you have left that decision to somebody who has not seen the assembly.
How this changes what you ask for in a quote
Finishing is quoted separately from machining, usually by a third party, and it frequently has a lot minimum. That has two consequences for sourcing.
Small quantities pay a premium that is not visible in the per-part price. A finishing line has a minimum charge, so twenty parts and eighty parts often cost close to the same to anodize. If you are close to a quantity break, finishing is one of the places where ordering more is cheaper per part than ordering fewer.
Masking is labor, priced per part. Every plugged hole and every masked thread is a person doing something by hand. Four masked features on a small part can cost more than the coating. If the design can avoid masking by moving a tight feature off a coated surface, that is a real saving and it is a DFM decision, not a purchasing one.