Core answer: The three things that most often wreck a plastic part are: (1) a bad main-wall-to-rib ratio, which causes sink marks and warpage; (2) snap-fit cantilever strain beyond the material's limit, which breaks after a few hundred open-close cycles; (3) self-tapping boss diameter and driving torque out of balance, which cracks the boss on the assembly line. These three sets of rules are common to about 80% of consumer electronics and home appliances. Do the math first, then talk about feel.
Let me start with the conclusion: what actually sends a project back for rework is never a 0.1mm cosmetic surface deviation. It's wall thickness, snap fits, and screw bosses. I've done structural design in Dongguan for eight years, on projects ranging from smart speakers and beauty devices to small appliances and medical accessories — over 320 shipped products. Flip through the T0 to T3 trial-molding reports and you'll see it: sink, assembly interference, and cracked bosses together account for about 80% of rework causes. A cosmetic issue gets you scolded by the client. These three get you a new mold — over a hundred thousand RMB, plus three weeks of schedule gone.
1. Wall Thickness: Thicker Is Not Stronger, and Sink Marks Start Here
A lot of new structural engineers have the same reflex: the product needs to survive drops, so make the walls thicker. That's the classic pitfall. Every 0.5mm you add to a plastic wall increases cooling time by roughly 15-20%, the internal shrinkage becomes uneven, and the surface develops depressions — sink marks. The thick section also forms internal voids, so on a drop test it cracks right there.
Our working values for common materials: ABS at 1.8-2.5mm, PC at 2.0-2.8mm, PC+ABS at 1.8-2.5mm, PP at 1.5-2.2mm because of its high shrinkage. Keep the shrinkage rates in your head too: ABS around 0.5%, PC around 0.6%, PP 1.5-2.0%, POM up to 2.0%. On a precision mating part, getting shrinkage wrong by 0.3% means a 100mm part is off by 0.3mm, and assembly jams solid.
A real case: last year we did a desktop Bluetooth speaker with a transparent PC top cover and high cosmetic requirements. The first version used a 3.5mm main wall, justified as "drop protection." Trial molding produced a band of rippled sink marks across the top that no amount of polishing could save. We dropped the main wall to 2.2mm with a honeycomb rib pattern inside, rib thickness held at 1.2mm — 0.55x the main wall — and it still passed a 1.2m six-face drop test with the sink marks completely gone. Once rib thickness exceeds 0.6x the main wall, the back of the rib shows obvious whitening and a sink print. Never cross that line.
One more easily missed point: internal fillets. If the internal fillet is smaller than 0.5x the wall, melt flow is restricted at the corner and stress concentrates there — that's where a drop test will break. Our internal standard is an internal fillet of at least 0.5x wall, and 1x wall at load-bearing corners. External fillet equals internal fillet plus wall thickness, which is what keeps the wall uniform.
2. Snap Fits: Calculate the Strain, Don't Bend It by Feel
Broken snaps are one of the most common after-sales failures in consumer electronics. Many engineers design a snap by eye — if it clips together on the prototype, it's good. Then in mass production the user opens and closes the part a few hundred times and it snaps, and the complaints pile up.
Core answer: Snap-fit design must calculate cantilever strain: ε = 1.5 × t × δ / L², where t is root thickness, δ is the engagement interference, and L is cantilever length. ABS allows about 1.5% strain, PC about 2.0%, PP over 3%. If the number exceeds the material limit, either lengthen the cantilever, reduce the interference, or change material.
An example: a battery-cover snap on a beauty device, in ABS, first version at 1.2mm root thickness, 1.2mm interference, 8mm cantilever. Run the formula and the strain is 1.5 × 1.2 × 1.2 ÷ 64 — about 3.4%, far past the 1.5% ABS limit. Hand assembly in the prototype phase was fine because a person controls the force. But a user opening it once a day hits the problem within three months. On our insertion test it broke at 3,000 cycles.
There are three directions for a fix; we combined two. First, interference came down from 1.2mm to 0.6mm, with a lead-in chamfer on the engagement face — the feel did not get worse. Second, cantilever length went from 8mm to 12mm and root thickness down to 1.0mm. Recalculated, strain is 1.5 × 1.0 × 0.6 ÷ 144, about 0.63%, comfortably safe. After the change it survived 10,000 insertion cycles intact. One aside: give the snap root a generous fillet, R0.3 or more. A sharp root corner is the worst stress riser, and many fractures start from that right angle.
Assembly lead-in matters too. Put a 30°-45° lead-in chamfer on the male snap and a matching chamfer on the female side, so assembly is sliding friction rather than a hard push. Skip those chamfers and the line operators fight the part, and yield drops immediately.
3. Screw Bosses: Nine Out of Ten Cracks Are Not a Material Problem
A cracked screw boss is the most painful thing on a production line: every screw you drive blows out another boss, and a whole batch of housings is scrap. The first instinct for many people is to change material, swapping ABS for PC+ABS. The problem is usually dimensions and torque.
Start with dimensions. The inner diameter of a self-tapping boss is normally the screw's outer diameter minus 0.5-0.8mm. For an M3 screw, the boss ID is 2.4-2.5mm; for M2, 1.5-1.6mm. Boss outer diameter is typically 2-2.4x the ID, so an M3 boss is at least 6.0mm OD and up to 6.8mm at load-bearing locations. Always add a root fillet of R0.25-0.5, plus two or three ribs at 0.6x the boss wall thickness.
Here's a pitfall worth sharing. A humidifier bottom housing used M3 self-tapping screws with a 5.5mm boss OD and only 1.5mm wall. The line ran an electric driver at 0.6N·m, and a third of them blew out. Opening the parts, every crack started at the boss root. We changed three things: OD up to 6.8mm, an R0.4 root fillet, and three ribs — and worked with the line to bring torque down to 0.35N·m. On re-sampling, 500 parts with zero cracks.
Core answer: For an M3 self-tapping boss, use 2.4-2.5mm ID, at least 6.0mm OD, a root fillet of R0.25-0.5 with ribs, and keep assembly torque at 0.3-0.4N·m. Below 1.8mm boss wall thickness it is very hard to avoid cracking even with a different material — thicken the boss first.
Another detail: leave a counterbore (a "volcano") on the top face of the boss so displaced plastic has somewhere to go when the screw bottoms out. Without it you get white marks on the boss top and, in bad cases, the boss splits. A counterbore depth of 0.3-0.5mm is enough.
What to Take Away
All three point at the same thing: the value of structural design is finishing the math before the mold is cut. Change anything after that and it costs tens of thousands of RMB and two to three weeks. My habit now is to run a checklist before every 3D release: wall-to-rib ratio, are the fillets big enough, has snap strain been calculated, are boss dimensions and torque right. Ten minutes of work that saves a month of arguing later.
FAQ: Product Structural Design
Q: What main wall thickness should a plastic part use?
A: ABS at 1.8-2.5mm, PC at 2.0-2.8mm, PC+ABS at 1.8-2.5mm, PP at 1.5-2.2mm, with rib thickness held at 0.5-0.6x the main wall.
The exact value also depends on part size and load. Large parts can go thicker, but always pair that with ribs rather than adding wall alone, or you get sink marks and internal stress together. Transparent parts are especially sensitive to sink and are usually 0.3-0.5mm thinner than an equivalent opaque part.
Q: What do people most often overlook in snap-fit design?
A: Strain calculation and the root fillet. If strain exceeds the material limit, the snap assembles fine short-term and always breaks long-term.
Calculate with ε = 1.5×t×δ/L² and keep ABS under 1.5% strain, PC under 2%. Add a 30°-45° lead-in chamfer on the male snap and a root fillet of R0.3 or more, and insertion life improves noticeably.
Q: How many rounds does structural design usually take before mass production?
A: Conventional consumer electronics can reach mass production in one or two T-sample rounds. Complex products with sealing or transmission functions usually need two or three.
The key to cutting rounds is nailing wall thickness, snaps, bosses, and the tolerance chain at the design stage, plus a complete interference check and DFM review while still in 3D.
Q: What do timeline and cost look like for outsourced structural design?
A: A single product's structural design usually takes two to four weeks, with cost depending on complexity, part count, and whether trial-mold follow-up is included.
A one-stop package covering industrial design, structure, and CMF runs a little longer. Hezi Industrial Design is based in Dongguan and supports the full path from design to mass-production mold follow-up; pricing can be discussed directly.
Originally published on the Hezi Industrial Design website (hezidesign.com), Structure Design Field Notes column.
Top comments (0)