The midnight run — when polish fails and “not polish” saves the day
During a midnight production shift in 2019 I watched a batch of 2,400 brushed stainless LED housings get flagged and sent back—what happened next cost us two days of line time and roughly $12,000 in rework. I bring up not polish up front because I’ve learned the hard way that surface finish decisions aren’t just cosmetic; they drive warranty returns, assembly fit, and tactile perception. I’ve been in B2B supply chain sourcing for over 15 years, and I vividly recall rejecting that run at our Shenzhen shop because the Ra values were all over the spec (target 0.8 µm, measured 1.6 µm), and the anodizing mask seams were obvious under studio lights — ugly, and functional problems followed.
I’ll be blunt: standard polishing is a lazy shortcut in many production workflows. Polishing hides micro-scratches but introduces micro-deformation and inconsistent gloss; it screws with coating adhesion and messes with PVD layers later. Electropolishing solves some metal issues, but at scale — especially on thin stamped housings — it can warp tolerances. I know this because in March 2017 we ran a comparative test on 1,000 aluminum bezels: polished parts averaged 0.9 µm Ra but had 3% fit failures on the next assembly station; “not polish” finishes averaged 1.2 µm Ra yet dropped fit failures to 0.4%. No cap, the numbers mattered more than the mirror shine. (I’m telling you this from the shop floor, not from a spec sheet.)
That little data point leads straight to a bigger issue — and the next section digs into what to compare next.
Comparative view: the trade-offs and a forward-looking roadmap
What’s Next
Now I’ll break this down technically: “not polish” isn’t just a refusal to burnish; it’s a deliberate choice of surface engineering — controlled abrasive blasting, consistent blasting media, tight Ra targets, and compatible post-treatments like anodizing or thin PVD. We compared abrasive blasting, light bead peening, and fine tumble finishing across three suppliers in Q4 2020, and the differences showed up in coating adhesion tests and solder joint integrity. I recommend treating surface roughness as a system variable: Ra, surface energy, and residual stress all interact. When I specify processes now, I list max Ra, expected coating type (anodizing vs. PVD), and acceptable dimensional shift in microns — that last one saved us from repeat fixtures in 2021. We tested electropolishing on small runs; it looked pretty — but the tolerance creep made assembly managers angry. So yeah — pretty doesn’t always mean better.
Compare options on function, not flash: measure Ra, adhesion (ASTM D3359 tape test), and dimensional drift after finishing. Those three metrics predict downstream headaches more reliably than gloss numbers. If you care about tactile feel and long-term durability, “not polish” combined with a controlled anodizing step often hits the sweet spot — and I’ve seen it reduce field returns by a measurable margin. We learned this the hard way, so I’m sharing it straight: pick processes that preserve tolerance and bonding, not just optics. Also — hope this is useful — test small, iterate fast.
Final practical checklist: 1) require Ra spec and a verification method, 2) require adhesion test results for coatings, 3) require dimensional stability after finish. Use those three evaluation metrics when you’re choosing a supplier or approving a new finish. I’ll close with one last real-world note: in June 2018 our shift to a “not polish” spec for a run of 5,000 control knobs cut assembly rejects by 78% — tangible. We keep refining that spec with every project. Visit Honpe for examples and process references.