DEV Community: Asher Hu

Structural Design in Practice: The Balancing Act Between Feature Stacking and Cost Control

Asher Hu — Wed, 09 Sep 2026 04:48:15 +0000

The biggest fear with multi-function integrated products is that the internal structure stacks up like a game of Tetris — and you end up over budget on tooling with assembly yield dropping below 80%. I've spent 10 years in structural design and led over a dozen projects. My core answer is one sentence: when stacking features, you must use "modular decomposition + limit tolerance analysis" to hold cost down — otherwise it will fail.

Take a project we did last year for a Shenzhen brand: the client wanted wireless charging, a phone stand, a pen slot, and magnetic attachment integrated into one product, with tooling budget capped at RMB 250,000. Our team optimized the structure and pushed tooling cost down to RMB 220,000 with assembly yield at 97%. Here's that case, broken down into three real-world pitfalls.

Key Takeaways

The core conflict in product structure design: feature stacking vs. cost control
Real case: tooling cost cut from RMB 250k to 220k, yield at 97%
Three key actions: modular decomposition, limit tolerance analysis, mold structure simplification

1. Feature Stacking: Don't Let "Multi-Function" Become "Multi-Disaster"

Many designers try to cram every function into one shell from the start. The result: internal structures fight each other, and heat dissipation, assembly, and strength all suffer. The most common conflict in projects is between the wireless charging coil and the magnets. In our case, the client wanted 15W wireless charging while the magnets needed to hold an iPad Pro. The initial stack put the coil at the center of the base with magnets arranged around it. But the gap between magnets and coil was only 2.5mm — magnetic interference dropped charging efficiency below 12W.

How did we solve it? We changed the magnets from "surrounding" to "single-side." Specifics: N52 NdFeB magnets, 8mm × 2mm, placed at the right edge of the base with a 15mm center distance to the coil. Interference dropped below 5%, and charging efficiency stabilized at 14.5W. To boost holding force, we added a 1.5mm pure iron flux plate behind the magnets, raising pull force from 3N to 8N — the iPad Pro held rock solid.

⭐ Core principle for feature stacking: Follow "modular decomposition + limit distance analysis." Break each function (charging, magnetic attachment, stand) into independent modules; calculate the minimum spacing first (coil to magnet at least 15mm to avoid interference), then use tolerance stacking to set assembly clearance (0.1-0.15mm per side recommended), and finally simplify the mold structure to cut cost. Remember: the more functions, the more important modularity becomes.

Pitfall lesson: At first mold trial we overlooked interference between the pen slot and the phone stand. The pen slot was 12mm ID and the stand arm 3mm thick — when folded, the stand arm jammed the pen slot, and workers had to force pens in. We reduced the arm to 2mm, enlarged the pen slot to 13.5mm ID, and added a 0.5° draft angle. Always run an assembly interference check — ideally simulate in UG or Moldflow.

2. Cost Control: How to Cut Tooling Cost — Structure and Materials

These projects are typically low-volume, high-variety, and tooling is the biggest cost. The client's budget was RMB 250,000; how did we cut it to 220,000? The core: simplify mold structure + share the mold base. Specifically: make the base and top cover as two molds but share one standard mold base (800mm × 600mm), saving RMB 12,000. We also cut the base's slides from four to two. The original base had four side cores (forming the phone stand slot and pen slot); by adjusting the parting surface we converted two side cores to lifters at 8° with 10mm stroke. Slide mold cost dropped from RMB 15,000 to 8,000 each, saving another RMB 14,000.

Material selection matters too. The client specified PC+ABS for the base, but the base had ribs (2mm high, 1.2mm wide) at the bottom, and PC+ABS has poor flow — short shots caused sink marks. We recommended ABS + 10% glass fiber: better flow, shrinkage down from 0.6% to 0.3%, and higher strength. Material cost rose RMB 2/kg, but yield went from 85% to 97% — net cost actually dropped.

⭐ Tooling cost control formula: Tooling cost = shared mold base (saves 20%) + slides to lifters (saves 30%) + material flow optimization (yield +10%). Specifics: lifter angle 5°-10°, stroke under 15mm; material ABS+GF10% with about 0.3% shrinkage; mold life up to 300,000 shots. Remember: the simpler the mold structure, the lower the cost and the higher the yield.

One more easily overlooked point: wall thickness uniformity. In our case the base wall was 2mm but only 1.2mm at the stand pivot, causing severe sink marks during molding. We unified the wall to 1.8mm, thickened the pivot locally to 2mm, and added a 0.3mm vent groove — sink marks resolved. Keep wall thickness variation within 0.3mm, or the mold shop will charge extra for rework.

Finally: for products like this, structural design isn't about stacking features — it's about stacking balance. Function, cost, and yield take the optimum solution, not the maximum feature set. Our team always calculates the tooling budget first, then works backward to the feature stacking plan. That's how you avoid a crash.

FAQ: Common Questions on Product Structure Design

Q: With too many product functions, how do I avoid internal structural interference?

A: Use modular decomposition — each function becomes an independent sub-module, then use limit distance analysis to set spacing. For example, keep the wireless charging coil at least 15mm from magnets to avoid magnetic interference. Also run assembly interference simulations in UG or Moldflow to catch conflicts early.

Q: How do I control tooling cost?

A: Share the mold base, convert slides to lifters, and choose materials with good flow. A shared mold base saves about 20%; lifter angles of 5°-10° save about 30% of tooling cost; ABS+GF10% improves yield by 10%. Keep wall thickness variation within 0.3mm to avoid rework charges.

Q: How do I solve injection molding sink marks?

A: Unify wall thickness, add vent grooves, and choose low-shrinkage materials. Wall thickness of 1.8-2mm with variation under 0.3mm; vent grooves 0.3mm deep; prefer low-shrinkage ABS+GF10% (about 0.3% shrinkage) over PC+ABS at 0.6%, which is harder to control.

This article is adapted from the Hezi Industrial Design official website (hezidesign.com), "Structural Design Field Notes" column.

Huawei Mate XT 2 Tri-Fold Phone Launched: The Design Revolution in Hinge and Screen Ratio

Asher Hu — Tue, 08 Sep 2026 15:35:33 +0000

The most noteworthy point for designers in this news: the tri-fold form factor is moving from "technical spectacle" to "everyday tool." Huawei's Mate XT 2 engineering trade-offs in screen ratio, hinge feel, and overall thickness have set new design benchmarks for foldable products.

Last week's Huawei HarmonyOS 7 launch dominated feeds. What excited me as a designer wasn't the specs — it was the "unfolded feel" of the Mate XT 2 tri-fold in hand. Tri-fold was still a "concept device" last year; this year hands-on reviews already describe it as "moving from spectacle into daily life." As an industrial designer, I don't care how many times it folds. I care about: what difficult trade-offs did designers and engineers make behind the scenes to turn it from "spectacle" into "daily"?

What This News Says

Huawei Mate XT 2 is Huawei's second-generation tri-fold flagship phone, debuting with a new-generation Kirin chip, an all-new hinge structure, and a 16:9.5 special-ratio outer screen. When unfolded, the screen reaches 10.2 inches — close to tablet experience — and folded thickness stays within 13mm. It offers stronger hinge durability and a flatter folded screen, aiming to push tri-fold from "technology showcase" to "daily use."

Three Design Takeaways from an Industrial Design Perspective

1. Screen Ratio 16:9.5: A "Counter-Mainstream" Choice Backed by Deep Scenario Thinking

Every phone maker is pushing toward either "wider" or "narrower" extremes, but Huawei chose 16:9.5 — a nearly square ratio. At first glance it looks like a "step backward," but think about it: the core advantage of tri-fold is the "desktop feel" when unfolded. At 16:9.5, the outer screen when folded is close to a traditional phone ratio — comfortable for one-handed use; when unfolded it's close to a landscape tablet, where viewing-area efficiency for reading documents, charts, and note-taking far exceeds a conventional tall-narrow phone screen.

As a designer, I immediately think: this ratio is the compromise-optimal solution for both "folded" and "unfolded" usage states. It doesn't cater to any single scenario — it pursues continuity: as the user goes from folded to unfolded, the field of view transitions smoothly from "phone strip" to "desktop square" without a jarring break. That's far harder than stacking specs.

2. Hinge Design: Engineering Aesthetics from "Foldable" to "Unnoticeably Folding"

The biggest design challenge for a tri-fold phone is the hinge. Two things about the Mate XT 2 hinge impressed me: first, it reduces the folding gap — when closed, the gap between screens is nearly invisible and the device feels strongly unified; second, it optimizes the unfolding damping — changing from a "click" to an "even unfold," feeling like opening the cover of a hardcover book.

Behind this is a dual breakthrough in materials and mechanism design. What designers need to understand: the hinge is no longer a "mechanical part" — it's part of the product's overall experience. Users unfold the screen dozens of times a day; each time, the feel, sound, and smoothness build their perception of product quality. Huawei turned the hinge into an "interaction surface" rather than a "technical component."

3. 13mm Thickness: Finding the Sweet Spot Between "Folding Function" and "Handheld Feel"

A tri-fold structure is naturally thicker than a slab phone, but 13mm is an interesting number. Compared to the first-generation tri-fold, the XT 2 shaved off about 1.5mm. That 1.5mm may not stand out visually, but holding it in your hand versus putting it in your pocket is a qualitative change — 13mm is about the thickness of an ordinary wallet, so users psychologically feel "this is something I can carry anywhere," not "an electronic device that needs dedicated storage."

Designers should understand: thickness isn't an absolute number — it's a signal of fit with usage scenarios. 13mm means it slides easily into a pants pocket and doesn't feel like a "brick" in hand. This trade-off means Huawei didn't sacrifice battery or hinge strength for "thinner," but clarified the priority order of "thickness-feel-battery-durability" at the product definition stage. That's a very mature design decision.

What Designers Can Take Away

The most valuable thing I learned from the Huawei Mate XT 2: foldable design competition has moved from "can it fold" to "how does it seamlessly integrate into life after folding." Screen ratio, hinge feel, and overall thickness — these three seemingly basic dimensions are exactly what determines whether a product becomes a "daily tool" rather than a "tech toy."

As industrial designers, when facing new product forms we're most easily attracted by "innovation" itself, while neglecting "everydayness" — is it natural, handy, and unnoticeable for users to use every day? The Mate XT 2's answer: turn engineering problems into experience problems, and technical parameters into usage feelings. That's where the real value of industrial design lies.

FAQ

Q1: Why did Huawei choose the non-mainstream 16:9.5 screen ratio?

A: To achieve the best viewing-area utilization in both folded and unfolded states, 16:9.5 is the "compromise optimum." When folded it approaches a traditional phone's feel; when unfolded it's nearly square, suited to reading, documents, and multi-tasking split screens, avoiding the wasted ratio of a tall-narrow screen when unfolded. It's the result of deep user research by the design team into "dual usage scenarios."

Q2: What's the hardest part of tri-fold hinge design?

A: The hardest part is balancing "reducing the folding gap," "optimizing unfolding damping," and "long-term reliability." Huawei Mate XT 2 achieved a gapless close and even, smooth unfolding by improving hinge materials and structure. Designers need to treat the hinge as an "interaction component," not merely a "mechanical part," because the dozens of folding experiences per day directly shape the perception of product quality.

Q3: Is 13mm thickness "thick" or "thin" among foldable phones?

A: 13mm is a very reasonable thickness for current tri-fold phones, achieving the qualitative change of "carryable anywhere." It's thick compared to a slab phone, but considering the tri-fold structure, 13mm happens to match wallet thickness — users can slip it into a pocket easily without an awkward grip. This trade-off shows Huawei prioritized battery life and hinge strength — a mature product definition.

This article is adapted from the Hezi Industrial Design official website (hezidesign.com), "Industry News" column.

Plastic Welding Process Selection in Practice: Ultrasonic, Hot Plate, Vibration Friction, and Laser Welding

Asher Hu — Tue, 08 Sep 2026 05:21:28 +0000

Once a plastic housing is molded, how do you join two parts together? Screws take space and look bad. Snaps limit force and direction. Adhesive is permanent once applied. Welding is the most common permanent joining method in product structure design — but ultrasonic, hot plate, vibration friction, and laser welding differ enormously. Pick wrong and you waste mold money, stall yield, and watch scrap rates climb. This article lays out the four processes — application scenarios, cost thresholds, and design requirements — so structure designers can use it directly as a selection checklist.

When to Consider Welding Instead of Snaps or Screws

The logic of part joining isn't complicated. Snaps suit serviceable assembly at nearly zero cost, but force and direction are limited and engagement decays after repeated cycles. Screws are strong and serviceable, but take space and leave visible bosses. Welding is permanent: strength approaches the base material, sealing is best, and no extra parts are needed.

Three scenarios where welding wins clearly: hermetic sealing (water tanks, air chambers, humidifier tanks); high-appearance requirements with no visible screws (consumer electronics, handheld medical devices); and when two or more plastic materials must be integrated in one sealed package. In 2025, welding usage in consumer electronics grew about 22% year over year, driven mainly by wearables and small waterproof appliances. The cost of choosing wrong isn't just wasted equipment — it's yield stuck at 80% in mass production.

Core Differences Between the Four Welding Processes

Ultrasonic Welding — First Choice for Small to Medium Parts

The most widely used plastic welding method. A transducer converts an electrical signal above 20kHz into mechanical vibration; the horn delivers energy to the Energy Director (guide rib) on the part, generating frictional heat that melts and presses the parts together. The whole cycle usually takes under one second — extremely efficient.

Material suitability is critical. Ultrasonic works well on rigid plastics: ABS, PS, PC, PMMA, and PA transmit vibration efficiently and reach 80%+ of base-material strength. But soft plastics like PP, PE, and TPE basically can't be welded — the material absorbs and damps the ultrasonic energy, so the joint never fully melts. We've hit this limitation in projects many times; when a client asks whether PP parts can be ultrasonically welded, the answer is to switch to another process.

Two design points matter: the cross-section shape and height of the Energy Director. The standard is a triangular ridge with a 60-90° apex, 0.3-0.8mm tall. During welding this tip melts first and spreads to fill the joint. Every 0.1mm of height difference changes weld strength by 15-20%, so dimensional deviation beyond 0.1mm breaks weld consistency.

Equipment investment: a standard ultrasonic welder costs about RMB 8,000-15,000, with horn machining at RMB 1,000-3,000 — the lowest threshold of the four. Throughput is effectively unlimited; one machine can do thousands of parts per shift.

Hot Plate Welding — First Choice for Large-Part Sealing

A metal plate is electrically heated 30-50°C above the plastic melting point. The two parts are pressed against opposite faces of the plate until the contact surfaces melt; the plate withdraws and the parts are pressed together to cool. Weld strength can reach 90%+ of the base material — currently the strongest plastic welding method.

Hot plate's biggest advantage is material compatibility: nearly all thermoplastics can be welded, including the PP, PE, and TPE that ultrasonic can't handle. The weld face can follow the part shape freely without horn constraints, suiting large parts like water tanks, fuel tanks, and pipes.

The cost: tooling is expensive — a double-sided hot plate fixture runs about RMB 30,000-80,000, plus a machine at RMB 40,000-120,000. Cycle times are far longer than ultrasonic: a large part takes 20-60 seconds from heating to clamping/cooling, an order of magnitude lower throughput. The plate surface also needs periodic cleaning of carbonized residue or black spots appear on the weld.

Vibration Friction Welding — Good for Odd Shapes and Ring Structures

Two plastic parts are pressed together and vibrate linearly against each other at 100-240Hz; friction heats and melts the interface. Like ultrasonic it's friction-based, but at much lower frequency and much higher amplitude, so it can deliver far more energy.

The big selling point: it welds irregular contours and nearly all thermoplastics — PP, PE, PA, ABS, PC all work, and it can even join different materials (PP to PE, for example). Automotive has used it for years: intake pipes, fuel tanks, water tanks. A vibration-welded fuel tank weld can hold 0.1bar without leaking.

Equipment investment is similar to hot plate: a vibration welder runs about RMB 50,000-150,000, with vibration tooling at RMB 10,000-30,000. Cycle time is shorter than hot plate — about 5-15 seconds for large parts. The hard limitation: parts must be able to move relative to each other in the horizontal direction, so it doesn't suit assemblies with built-in electronics — the vibration will damage circuit boards.

Laser Welding — The Future of Precision Welding

Near-infrared laser passes through a transparent upper layer (which doesn't absorb the laser) and strikes a lower absorbing layer, which converts light to heat; the heat conducts up and melts the joint. Since there's no mechanical contact or vibration, precision reaches within 0.05mm, virtually no flash forms, and areas outside the weld don't heat up.

This process is used increasingly in medical devices and sensors. Pacemakers, glucose monitoring sensors, and miniature camera modules — products extremely sensitive to heat-affected zones — can only be met by laser welding. The upper material must transmit near-infrared (clear PMMA, PC, PA), and the lower material needs an absorber (carbon black is most common).

But laser welding has the highest barrier of the four: equipment runs about RMB 200,000-500,000, plus clamping fixtures and pneumatic systems. Weld faces need precision mating — gap no more than 0.1mm — demanding extremely tight part tolerances.

Selection Comparison: Four Dimensions

Material compatibility: Ultrasonic = rigid plastics mainly; Hot plate = nearly all thermoplastics; Vibration = nearly all thermoplastics; Laser = clear-on-absorbing pair.

Weld strength: Ultrasonic = 80%+ of base; Hot plate = 90%+; Vibration = 85%+; Laser = 70-85%.

Equipment investment: Ultrasonic = RMB 0.8-1.5万 (8-15k); Hot plate = 4-12万 (40-120k); Vibration = 5-15万 (50-150k); Laser = 20-50万 (200-500k).

Cycle time: Ultrasonic = under 1s; Hot plate = 20-60s; Vibration = 5-15s; Laser = 3-15s.

A Real Case: How We Selected the Welding Process for an Air Purifier Humidifier Tank

Last year we took on an air purifier humidifier tank. Requirements: no leaking, survive 30,000 thermal cycles without cracking, and no visible weld marks on the exterior. The tank material was set as PP — good chemical resistance and easy food-grade certification. The client initially wanted ultrasonic welding for low cost and speed.

But PP is a soft plastic — ultrasonic transmission is inefficient. We made ultrasonic samples: weld strength measured only about 50% of the base material, and leak testing showed 3 of 10 samples with micro-leaks. We told the client and re-evaluated two options: hot plate and vibration friction. Hot plate had the best strength, but the tank was thin-walled (2.5mm); hot plate heating risked local deformation, and the 50+ second cycle would hurt takt time. We chose vibration friction: weld strength measured 87% of base, 100% leak-test pass rate, about 12 seconds per part, equipment investment about RMB 90,000. Six months into mass production, yield stayed around 97% with zero leak-related returns.

Looking back, the logic wasn't complex: PP material rules out ultrasonic; thin walls rule out hot plate; hermetic requirements rule out screws; vibration friction sat right in the middle — acceptable equipment cost, high enough yield, sufficient strength. Selection isn't about finding the optimal process; it's about finding the option that satisfies every hard constraint across the four dimensions.

FAQ: Common Questions on Plastic Welding Processes

Q: Which welding method suits ABS plastics?

A: ABS is a rigid plastic — ultrasonic welding works best, reaching 85%+ of base-material strength with under-one-second cycles and the lowest equipment cost. For parts over 300mm or with complex shapes, vibration friction welding also works but costs more.

Q: What's the best Energy Director dimension for ultrasonic welding?

A: Standard design is a triangular cross-section with a 60-90° apex and 0.3-0.8mm height. The rib width is about 1.5x its height. Every 0.1mm of height deviation changes weld strength by 15-20%, so mold machining precision should be within ±0.05mm.

Q: Why can't PP be ultrasonically welded?

A: PP is a semi-crystalline soft plastic with low elastic modulus — the ultrasonic vibration energy is absorbed and damped by the material itself and can't concentrate at the weld face. For PP parts, choose hot plate or vibration friction welding; neither relies on the material's rigidity to transmit vibration.

Q: Does the welding process choice affect the product mold structure?

A: Greatly. Ultrasonic requires an Energy Director ridge at the weld face; hot plate requires a flat weld face with enough heating clearance; vibration friction requires relative motion space between the two halves. Decide the welding process no later than the mold design stage — changing after the mold is cut costs tens of thousands in rework.

Q: Can different plastics be welded together?

A: Same-plastic welds (PP to PP) have the highest strength. Similar materials (ABS to PC) can use vibration friction or hot plate, with strength dropping 10-25%. Completely incompatible pairs (PP to PA) have melting points more than 30°C apart — the weld faces can't melt simultaneously, so direct welding isn't recommended.

This article is adapted from the Hezi Industrial Design official website (hezidesign.com), "Structural Design Field Notes" column.

Product Hinge and Pivot Design: Hinge Types, Torque Calculation, and Life Testing — Field Notes on Flip Mechanisms

Asher Hu — Mon, 07 Sep 2026 05:54:50 +0000

A product has a flip lid that starts feeling loose after just a few dozen open-close cycles. Every structure designer has run into this. Notebook screens shouldn't wobble, but they also shouldn't require two hands to open. Smart home lids should hold exactly the angle you set. Handheld device flip heads should survive thousands of bends. Hinges and pivots look simple, but failures are expensive: at minimum a repair and housing replacement, at worst a full recall. This article covers selection, torque calculation, and life testing for flip mechanisms.

Hinge-Type or Torsion-Spring? Decide Before You Build

Flip mechanisms in products generally split into two camps.

Shell-type hinges (friction hinges): two housing halves joined by a shaft with a bushing. Friction between the bushing and shaft provides damping. Notebook hinges, tablet flip stands, phone folding mechanisms all use this type. High load capacity and complex positioning angles are possible, but more parts and higher cost.

Torsion-spring hinges: a metal tube with a torsion spring inside. Opening the lid loads the spring; closing releases it. Friction between the spring wire and the tube wall is the damping source. Portable beauty device lids and some smart speaker flip covers use this design. Simple structure, low cost, but the torque range is narrow — not suited for large-angle multi-position scenarios.

Choose based on three factors: opening angle (under 180° is fine for torsion-spring; beyond that needs shell-type), target life (under 10k cycles for torsion-spring; 20k+ needs shell-type), and budget (torsion-spring is 40-60% cheaper per unit than shell-type).

A Beauty Device Flip Lid: How Torque Is Calculated

Last year we designed a portable beauty device with a torsion-spring lid. Client requirements: 10k open-close cycles life, free-stop at any angle. The first prototype started wobbling at 3,000 cycles. The root cause wasn't the spring — it was bushing wear.

The torque formula our industry uses: T = W × L × sinθ × k. W is lid weight, L is the distance from pivot to lid center of gravity, θ is half the max opening angle, and k is a safety factor. That lid weighed about 45g, CG 35mm from the pivot, opening 150°. Plugging in: T = 45g × 35mm × sin75° × 1.6 ≈ 1.2 kgf·cm. Safety factor 1.6 covers spring preload margin.

The problem was the bushing. First version used standard POM against a 304 stainless shaft — dry friction coefficient about 0.2-0.3. After 3,000 cycles, the surface started peeling and torque dropped 42%. The fix: switch to MoS₂-filled self-lubricating POM (friction coefficient 0.12-0.15), coat the shaft bore with DLC (diamond-like carbon), and tighten the clearance from 0.08mm to 0.04mm. After the change, 12,000 cycles showed only 18% torque decay — passed customer validation.

Bushing Material Selection: POM, Bronze, Stainless Steel

Bushing material directly governs pivot life and feel. We've tested three mainstream options.

POM (polyacetal): good self-lubrication, friction coefficient about 0.18-0.22, lowest cost. Works for most consumer electronics pivots. But it absorbs moisture in high-humidity environments and swells — leave 0.05-0.10mm clearance margin. In our testing, POM bushings in dry friction showed stable friction coefficient fluctuation of 0.18-0.22 over 5,000 cycles, while dry bronze bushings started showing stick-slip after 2,000 cycles with increasing torque scatter.

Phosphor bronze bushings: high load capacity, good heat dissipation — first choice for 20k+ cycle scenarios. But bronze-on-steel dry friction is high (0.35+), requiring grease. Otherwise squeaking is guaranteed. Good for notebook hinges and applications with assembly space for lubrication.

Stainless steel bushings: corrosion resistant, mandatory for medical and food-grade products. But machining precision is demanding and cost is 3-4x POM. Stainless-on-stainless tends to gall — always pair with a different material.

Life Testing: Three Reference Lines

Clients care most about open-close cycles. Life targets vary by product category. Here are the baselines we use:

Product Type	Life Target	Test Conditions
Consumer electronics (beauty devices, speaker lids)	5,000-10,000	Full open-close, 10-15 cycles/min
Notebooks / foldable tablets	20,000-30,000	Full open-close + 90° hold test
Vehicle flip screens	10,000+	Add high-temp 85°C and low-temp -40°C

Monitor the torque decay curve during testing. More than 30% decay from initial torque = failure. In one humidifier lid pivot project, initial torque was 1.5 kgf·cm; after 8,000 cycles it dropped to 0.9 — 40% decay. The cause: bushing wall was only 0.8mm thin, and local stress exceeded the material yield point. Increasing wall thickness to 1.2mm brought the design past 11,000 cycles.

Three Critical Checkpoints for Structure Designers

1. Tolerances. Shaft-to-bushing clearance should be 0.02-0.05mm, ISO fit recommendation H7/g6. Too tight won't turn; too loose wobbles. Note: POM bushings have injection shrinkage variation of 0.3%-0.8% — batch consistency is a common failure point.

2. Shaft fixation. Don't let the pivot float on just one hole in the housing wall. Design support ribs that brace both sides. A cantilever pivot structure starts making noise or sticking around 5,000 cycles due to fatigue. Support ribs should be at least 1.5× the shaft diameter in height.

3. Cable routing clearance. If a flip housing carries an FPC flex cable, leave enough clearance. The cable must not rub against the pivot core during opening and closing. FPC bend radius should be no less than 1.5mm — below that, the traces crack eventually. We recommend molded cable guides inside the housing wall to fix the flex path.

Final piece of experience: pivot feel cannot be locked in by drawing parameters alone. The same torque design value produces very different feel with different friction materials, clearances, lubrication states, and machining burrs. Always build prototype samples, run them through a full aging cycle, and then confirm the mold. Changing feel after the mold is cut is expensive and time-consuming.

FAQ: Common Questions on Product Hinges and Pivots

Q: The actual pivot torque is much lower than the design value. What's wrong?

A: Most common cause is oversized shaft-to-bushing clearance. Mold tolerances and material shrinkage variation both affect real clearance. Best practice: verify torque with prototypes before mold release, and leave adjustment room in the mold design — for example, the bushing bore can be shrunk by 0.05mm during mold tryout.

Q: How do I eliminate squeaking in a flip lid?

A: Squeaking usually comes from three sources: (1) bushing and shaft running dry — switch to self-lubricating material or add grease; (2) loose shaft fixation — reinforce the support ribs; (3) internal flex cable rubbing against the housing wall — add cable guides. Check in this order and 90% of squeak issues are found.

Q: Identical pivot specification but some units are tight and some loose. What to do?

A: Manufacturing tolerance variation. POM bushing injection shrinkage typically varies 0.3%-0.8% between batches, causing inconsistent clearance. Solutions: machine the bushing bore as a post-processing operation (precision ID), or switch to a spring-loaded claw design that compensates automatically.

Q: How do I achieve free-stop at a specific angle?

A: Free-stop comes from balancing spring force against friction. The spring provides base holding force; static friction between bushing and shaft provides fine-tuning. The key is matching spring force to the friction coefficient — too little spring force and the lid won't stay, too much and opening feels crude. We typically tune by adjusting bushing clearance and spring preload turns.

This article is adapted from the Hezi Industrial Design official website (hezidesign.com), "Structural Design Field Notes" column.

Product Surface Logo Process Design: Screen Printing, Pad Printing, Laser Engraving, and IMD — How to Choose

Asher Hu — Fri, 04 Sep 2026 06:23:52 +0000

Choosing the right logo process is a key decision in product design. It directly affects the product's perceived quality and brand feel. Make the wrong choice and your mold is already cut — faded or blurry logos drag down the whole product's look.

Here's a practical comparison of four common logo processes, based on real projects:

1. Screen Printing: Lowest Cost, But Flat Surfaces Only

Screen printing pushes ink through a mesh stencil onto the surface. Simple design, few colors, high volume = extremely low per-unit cost.

Pros: Cheap at scale, good color saturation, thick ink layer gives a tactile feel.
Cons: Only flat or very gently curved surfaces. Curved and 3D surfaces are out. Medium wear resistance — a 3M tape peel test can lift the ink.

Best for: Large flat panel logos, instruction labels, packaging. We used it on a smart speaker's top surface logo — 100,000 units, per-unit cost $0.02.

2. Pad Printing: The Solution for Curved Surfaces

A silicone pad picks up ink from a cliché (etched plate) and transfers it to the product. Curved, concave, and convex surfaces all work.

Pros: Handles 3D surfaces easily. Better detail than screen printing — can do 0.2mm line widths.
Cons: Ink layer is thin — medium wear. Setup cost per color is moderate.

Best for: Earbud shells, mouse side buttons, game controller buttons — any complex curved surface that needs a logo. We pad-printed a medical device grip with a tiny logo on a concave surface — 0.3mm lines, 5,000 units, consistent quality.

3. Laser Engraving: Highest Precision, Never Wears Off

A laser burns the logo directly into the surface. No ink, no peeling.

Pros: Extreme precision (0.1mm), fine detail, graduated effects possible. Zero wear — can't rub off, solvent-resistant. No consumables per part.
Cons: Single color only (the material's reaction to the laser). Results vary significantly by material — ABS looks different from PC or metal. Slower per-part than printing.

Best for: Metal parts, high-gloss surfaces, medical devices, tools. We laser-engraved a stainless steel medical instrument — logo still crisp after 500 autoclave cycles.

4. IMD (In-Mold Decoration): Never Wears Out, But Expensive Mold

A printed PET film is placed in the mold before injection. The plastic bonds with the film during molding, embedding the logo inside the plastic.

Pros: Logo never wears — it's inside the material. Best color saturation of any process. Multi-color and gradient possible. Can do brushed-metal or textured looks.
Cons: Mold costs 30-50% more than standard. Film lead time adds 2-3 weeks to the schedule. Higher per-part cost due to film consumption. Minimum volume usually needed to justify the mold investment.

Best for: High-volume premium products — phone cases, home appliance panels, car interior trim. We did the front panel logo for a high-end air purifier — 200,000 units, zero logo wear complaints in 3 years.

5. The Decision Table: Which Process for Which Scenario

Factor	Screen Print	Pad Print	Laser	IMD
Surface	Flat	Any 3D	Flat/gentle	Flat/curved
Precision	0.3mm	0.2mm	0.1mm	0.1mm
Wear	Medium	Medium	Excellent	Excellent
Color	1-4 solid	1-4 solid	1 (material)	Full
Mold cost	None	None	None	+30-50%
Part cost	$0.01-0.05	$0.03-0.10	$0.02-0.08	$0.10-0.30
Min Qty	1000+	500+	100+	10000+

Quick guide:

Budget-limited + flat + high volume → Screen print
Complex 3D surface + low-medium volume → Pad print
Premium look + maximum durability → Laser engraving
Mass production + no-wear requirement → IMD
Already-available combination: flat panel logo with IMD + small 3D part with laser engraving

FAQ: Common Questions on Logo Processes

Q: Can I do a gradient effect with screen printing?

A: No. Screen printing is solid ink per layer. For gradients, use pad printing (fine halftone dots) or laser engraving (varying power density). IMD can do full gradients with the printed film.

Q: How do you attach a logo to a curved high-gloss surface?

A: Avoid screen printing — the squeegee can't conform strongly enough. Laser engraving is best if the material is compatible; pad printing is the runner-up. Test wear resistance on the curved area first — the stretched ink layer on curves can crack sooner.

Q: I need a metallic-looking logo on plastic. Which process?

A: Two options. IMD with a metallized PET film gives the best look but needs high volume. Hot stamping is a simpler alternative for lower volume — a metallized foil is pressed onto the plastic with heat and pressure; cheaper than IMD, good for medium quantities, but the foil edge may be visible after wear.

Q: Does laser engraving work on all plastics?

A: No. ABS and PC engrave cleanly (light gray to dark contrast). PP and PE absorb laser poorly — the mark is faint. Filled materials (glass-filled nylon) engrave unevenly. Always test on actual production material before finalizing. We've had projects where switching from ABS to PC changed the engraved contrast dramatically.

This article is adapted from the Hezi Industrial Design official website (hezidesign.com), "Structural Design Field Notes" column.

Product Sliding Guide Rail Design: Clearance Values, Rail Types, and Limit Locks — A Structure Designer's Field Notes

Asher Hu — Thu, 03 Sep 2026 05:41:29 +0000

How much clearance should a sliding channel have? Which material pairing should the rails use? How much snap force should the limit lock have so it doesn't loosen over time? When working on sliding fits, the worst problem isn't that the part jams — it's that it sometimes jams and sometimes doesn't. The notes below come from several products with sliding mechanisms we have developed, and have been verified on the mold shop floor and the assembly line.

1. Clearance Between Channel and Rail: Tighter Is Not Always Better

We once built a portable medical testing device. The housing was 120mm long, the top cover carried a sliding channel, and a PCB slid into place along the rail. The first version used a 0.05mm per-side clearance. After sampling, the board would not slide in. Injection-molded parts have shrinkage variation of around 0.15% per batch, plus 0.05mm of part warpage — together the actual clearance had gone negative. The second version went to 0.2mm. It assembled, but the wobble was obvious; shake the housing and you could hear internal parts knocking.

On the third pass we finally understood: clearance for a sliding fit is not a fixed value — it is a function of length and material. The rule of thumb we use: for every extra 50mm of length, add 0.05mm per-side clearance. ABS shrinkage varies 0.3%-0.6% per batch; leave margin for that range. The final product used 0.15mm per-side clearance with a 10mm lead-in chamfer on the guide section. Measured after 5,000 push-pull cycles: 0.02mm wear, stable feel.

A more demanding example was the sliding cover on an SD card holder. The holder is only 32mm long and needs a positive locating feel. We tried 0.05mm per-side clearance with a POM slider — firm but not jamming; after 800 insert/remove cycles the clearance grew to 0.12mm and push force dropped from 2.5N to 1.8N, still acceptable. With ABS-on-ABS at the same clearance, the joint was loose after only 300 cycles.

So the foundation of channel clearance is: first decide the friction material pair, then calculate clearance from length and shrinkage, and finally leave 0.1mm assembly margin for the line to adjust. Clearance design without prototype verification is basically designing on paper.

2. Choosing a Rail Form: Three Structures for Three Scenarios

Rail structures aren't better when more complex. In structure design there's an unwritten rule: more mechanisms means more failure points. For sliding mechanisms, there are three common rail designs:

C-channel rails are the simplest: the top cover and bottom housing each form a groove, and when closed they create a C-shaped guide channel. They suit short travel (under 30mm) and light loads (under 500g) — battery compartment covers, the sliding cover on a TWS earbud charging case. The weakness is sensitivity to molding deformation: poorly matched halves cause "edge scraping." The fix is to leave 0.3mm relief at the channel bottom so only the guide surfaces contact.

T-rails: one part has a raised rail tongue, the other a T-slot. Guidance precision is a step above C-channels — they constrain motion in both X and Y. We made a handheld thermal imager battery cover where the customer demanded a tight one-handed push-pull action. We used a T-rail plus two spring-loaded steel balls for detent — push to the end and it locks with a click. The tactile feedback is excellent, but mold cost runs about 15% higher because the T-slot needs slides or lifters for the undercut.

Dovetail rails can take a certain amount of pull-off force in one direction — used where parts must resist pulling out without adding screws. These need careful attention to draft; dovetail angles are usually 55°-60°, and outside that range the joint jams or slips. Frankly, dovetails are rare in consumer product design — they show up more in mold slide guides and precision instrument stages. For consumer goods, C-channels and T-rails are enough.

If your product's push-pull travel exceeds 80mm, or the frequency exceeds 50 cycles per day, add a stainless steel liner strip to the C-channel. Plastic-on-plastic sliding surfaces show obvious wear after about 20,000 cycles; a 0.3mm stainless strip can extend life 10x.

3. Limit Lock Design: The Bottom Should "Click"

Sliding structures fear two things: the user thinks it's fully seated when it isn't, and it seats fully but slides back at the lightest touch. The limit lock solves both.

The spring snap is the most common solution. The design key is the engagement depth — bigger is not better. At 0.3-0.5mm engagement, snap force is roughly 2-4N: a clear feel without strain. Above 0.8mm, the user feels a hard "stop" at the end of travel, and forcing it can snap the lock. Chamfers matter too: a 25°-35° lead-in angle slides in smoothly, and the retaining face should be vertical or above 85° to prevent slip-out.

A lesson from our own work: a desktop fan base slide cover used 0.6mm engagement and tested fine in ABS. At mass production, the mold shop changed the steel grade, shrinkage shifted, and actual engagement became 0.9mm. The assembly line reported 30% of covers wouldn't seat; the rest seated but the lock root whitened and cracked after three months. We set engagement back to 0.4mm and changed the lead angle from 30° to 40° — the problem disappeared.

Screw-based stops suit serviceable scenarios where users may adjust position or replace parts. Magnetic locating suits quick-release, low-force scenarios such as accessory bays and toolbox lids — but magnets pick up ferrous dust during sliding, so avoid them in dusty environments.

4. Tuning Damping Feel: Not a Game of Chance

On sliding products, feel is the user's first impression of quality. Too loose feels cheap; too tight feels crude. The key to good damping: don't try to nail it with one tolerance — stack multiple dimensions.

Friction ribs: add 0.2-0.3mm-high raised strips on the rail face, 0.5-1mm wide. This reduces contact area while raising pressure per point. Hard ribs use POM or PA; soft ones use TPU. On a Bluetooth remote battery cover, the sliding cover always felt loose; adding three 0.2mm POM ribs removed the wobble and raised push force from 0.8N to 1.5N. Users said it felt "premium."

Texture depth: mold texturing isn't just cosmetic — it affects feel. VDI 24 texture (about Ra 2.0μm) has roughly double the friction coefficient of a mirror finish (Ra 0.05μm). A car aromatherapy diffuser's sliding cover was too grippy in mirror finish, and grease didn't last. Switching to VDI 21 texture gave a steady 1.2N push-pull force with no lubricant, unchanged after 12 months of testing.

Spring preload: when the product needs precise, durable damping, a spring scheme beats friction tuning. A medical device drawer module used two 0.2mm wave washers pressing on the rail sides with 3N preload. No matter how the molded parts shrink, the spring supplies stable side pressure; push-pull force stayed within ±0.3N.

One easily missed point in damping: the break-in period. Fresh sliding fits lose 15%-30% of friction within the first 50-100 cycles. If the prototype feels just right, users may find production units loose — that's usually why. Our current practice: run 200 push-pull cycles on samples before measuring force, and use that data to set the acceptance standard.

5. Material Friction Pairing: Pick Wrong and It Won't Slide

The design core of a sliding structure isn't geometry — it's friction. Material pairing directly governs life, feel, and noise.

Measured on our own fixtures at 23°C, 50% RH, 2N load, 10mm/s sliding speed:

POM on steel: dynamic friction coefficient 0.15, lowest wear, longest life, quiet. Best for precision sliding and long-life moving parts. Drawback: POM shrinks a lot (2-2.5%), so dimensional stability is below PA.
PA on PA: coefficient 0.35-0.45, rising with moisture absorption; without glass fill it cold-flows easily. Fine for low-frequency parts that slide occasionally.
ABS on ABS: coefficient 0.5-0.6, noticeably noisy; smooth out of the box but shows white wear marks after about 1,000 cycles. Common in cost-driven products, but life is limited.
Self-lubricating plastics (POM+PTFE, oil-filled nylon): coefficient 0.08-0.12, almost no grease needed. Good for medical devices and mechanisms that run unattended. Cost roughly 2-3x.

A practical pairing we use often: POM slider on a stainless steel rail strip. On a smart lock sliding cover, we verified about 20 slides a day for 12 months — push-pull force changed less than 10%. If budget is tight, POM-on-POM works, but the sliding surfaces must be polished or textured; raw molded faces against each other shed white powder quickly.

6. Anti-Pull-Out Design: Past the End Is Trouble

Without end stops, a user shoving past the limit can send the part flying or damage it. Three common anti-pull-out schemes differ a lot in cost.

End-step stop: cheapest. Mold a 0.3-0.5mm step at the end of the channel. Keep the R corner at least 0.3mm or the mold will short-fill. Downside: no repeated assembly — once in, it stays in (unless you disassemble the product).

E-clip or C-ring: slip it over the rail during assembly; it opens and seats in a groove at the end. Removable — good for products that need maintenance. Adds about RMB 0.5-1 per unit. On a smart speaker's TF card slot cover we used a 0.4mm spring wire ring: 3N to push in, 5N to pull out. Users hear the "click" and know it's seated.

Anti-pull ribs: a few trapezoidal ribs on both sides of the rail; the slider seats into matching recesses at the end. Suits removable scenarios like push-in battery compartments. Ribs about 3-5mm long, 0.5mm high, angled 25°-30° — bigger and it won't come out.

When designing sliding products, budget for 3-5 prototype iterations. Version one proves it assembles; version two tunes the feel; version three tunes the life. No design handbook sets everything for you in one pass. Batch-to-batch shrinkage drift, mold mismatch, and assembly-line technique all challenge your clearance value. The good news: these pitfalls follow patterns. Fall into one once, and next time you'll know exactly how to give that 0.05mm.

FAQ: Common Questions on Product Sliding Rail Structures

Q: What clearance should the channel and rail use?

A: There's no universal number — it depends on length and material. Add 0.05mm per-side for every 50mm of length. For POM-on-steel, use 0.1-0.3mm for moving fits and 0.05-0.1mm for locating fits. Verify with prototypes: molding shrinkage variation and mold deviation both change the real clearance.

Q: How many push-pull cycles can plastic-on-plastic rails survive?

A: POM-on-POM with good fits wears 0.05-0.1mm over about 20,000 cycles. ABS-on-ABS is much shorter — noticeably loose after 3,000-5,000 cycles. For high-frequency, long-life requirements, add a stainless liner or switch to POM-on-steel.

Q: The slide feels too rough. Does grease help?

A: Grease works short-term but attracts dust — not recommended for home consumer products. The right fixes: check whether the clearance is too small, switch to a lower-friction pair (POM on steel), or polish / lighten the texture on the rail face.

Q: What engagement depth should the limit snap use?

A: Spring snaps generally use 0.3-0.5mm engagement with a 25°-35° lead angle. Above 0.8mm, insertion feels sticky and the snap root fatigues over time. After mass production starts, watch for mold shrinkage drift changing the real engagement.

Q: The sliding part pulls out at the end of travel. How do I prevent it?

A: Three options: an end-step stop (lowest cost, not removable), an E-clip (removable, adds RMB 0.5-1 per unit), or anti-pull ribs (good for removable parts like battery compartments). Pick by maintenance needs and budget.

This article is adapted from the Hezi Industrial Design official website (hezidesign.com), "Structural Design Field Notes" column.

Huawei Mate XT 2 Triple-Fold Leaks: New Hinge and Screen Tech — Structure Design Analysis

Asher Hu — Wed, 02 Sep 2026 00:45:19 +0000

Key Takeaways

Huawei's next-generation triple-fold phone, the Mate XT 2, has its core specs leaked from the supply chain: a 10.2-inch unfolded display, a new hinge structure, and a whole-device weight target under 280g. Triple-fold is moving from a "show-off" flagship to an everyday device — and that transition is worth discussing from an industrial design perspective.

News broke this week: as the follow-up to last year's Mate XT, the leaked changes concentrate on three areas — a larger unfolded screen, a revised hinge architecture, and a stricter weight budget. The triple-fold category proved the concept in generation one; generation two starts talking about daily usability. Both moves carry plenty of design lessons.

What the News Says

Per IT Home's report, the Mate XT 2 will adopt a larger inner display, roughly 10.2 inches unfolded, slightly up from the previous 10 inches. The hinge reportedly moves to a new composite hinge structure with targeted improvements in fold-cycle life and unfolded flatness. The whole-device weight target is under 280g. For a triple-fold form factor, hitting that number would be an impressive piece of industrial design. Battery-wise, the dual-cell approach continues — two cells wired in series, with capacity likely exceeding the previous generation's 5600mAh.

Three Points Worth Discussing from a Structure Design Perspective

The Triple-Fold Hinge: From "Making It Work" to "Making It Reliable"

The hardest part of a triple-fold phone is the hinge. It is not a single-axis fold — it is two hinge systems working together: one for outward folding, one for inward. The first-generation Mate XT already used a U-shaped bend to reduce screen creasing, but stacking two hinge modules makes thickness and weight hard to control. If the Mate XT 2 can shrink both hinge modules with a more compact composite structure, the gains in device thickness and weight would be qualitative.

We worked on a foldable hinge project last year, and the lesson that stuck was tolerance management. For single-axis folding, hinge clearance must be held within 0.05mm; a triple-fold has two hinge sets, doubling the cumulative tolerance. If Huawei pushes hinge consistency further — tightening the eccentricity of both hinges and closing the cover-plate clearance to single-axis levels — it would be a positive signal for the whole industry.

Internal Stacking: Accounting for the Extra Space

A triple-fold phone unfolds to a large area, but folded thickness is its weak point. To reach under 280g, structural parts must shed material. The battery occupies the most internal space, and the dual-cell design means two battery compartments plus much more complex FPC routing than a conventional foldable. The interconnect flex between the two batteries must cross the hinge area, and the routing must account for the flex cable's bend radius during folding.

The display support structure also needs redesign. A 10.2-inch unfolded area demands a denser array of mid-frame support ribs, or pressing on the screen causes watermark ripples. If the mid-frame uses magnesium-aluminum alloy, it is about 30% lighter than stainless steel, but rigidity needs reinforcement. How tightly the rib spacing, rib height, and wall thickness are tuned — without adding much weight — directly determines the feel in hand.

CMF Trade-offs: Balancing Lightness and Premium Feel

Lightweight triple-folds usually force CMF trade-offs. A titanium mid-frame is about 40% lighter than stainless steel with adequate strength, but machining cost is higher and surface finishing is harder — blasting and anodizing parameters for titanium differ from aluminum, and yield control is a real hurdle. A fiberglass-composite back panel cuts more weight but feels less warm than glass. How Huawei chooses between the two is a fascinating product-definition question.

The display cover material deserves attention too. The outermost panel (outward-fold side) needs a scratch-resistant glass cover, while the inner screen needs flexible CPI or UTG ultra-thin glass. These two materials trade off light transmittance, crease control, and impact resistance differently — generation two's material selection may shift from generation one.

What Designers Should Take Away

Triple-fold moving from concept to daily driver means two things for structure designers.

First, the hinge solution space is widening. Samsung's water-drop hinge is no longer the only play — U-shaped bends, composite hinges, and multi-link structures all work, and different fold geometries breed different hinge architectures. Designers have more options, but the workload of evaluating each solution's reliability and durability grows too.

Second, lightweight structural design skills are becoming more valuable. Whether triple-fold or other categories — consumer electronics, smart hardware, or automotive — the demand for thin-wall-plus-high-strength design is only increasing. How to shed weight without sacrificing rigidity, how to arrange ribs and fillets on thin walls to control deformation: these fundamentals show up most clearly in foldable products.

Whether the final production Mate XT 2 delivers on the leaked specs remains to be verified at launch. But the triple-fold line has already emitted enough technical signals, both in product definition and structural design, to be worth studying.

FAQ: Common Questions About Triple-Fold Phones

Q: Structurally, what does a triple-fold add over a regular foldable?

A: A triple-fold needs two hinge systems working together — one outward, one inward. A regular foldable has only one. Two hinges mean doubled cumulative tolerance, more complex FPC routing, and two separate battery cells wired in series. Overall structural design difficulty and part count both rise significantly.

Q: What kind of hinge life do triple-folds achieve?

A: Industry-standard foldable hinge life is generally 200,000+ cycles. With two hinges, each must meet that bar. Huawei's first-generation Mate XT hinge passed relevant reliability testing; if the Mate XT 2 switches to a new hinge structure, its life target should at least match the industry baseline.

Q: When will triple-folds drop below ¥10,000?

A: The hinge module alone costs 1.5-2x a regular foldable's, and with dual-cell batteries, a large flexible display, and a more complex mid-frame, the BOM cost won't fall quickly. Foldables took about three years to come down from ¥20-30k to around ¥15k. Getting a triple-fold under ¥10k will likely take another two to three generations — though as hinge designs mature and volumes grow, the downward cost trend is certain.

Injection Mold Ejection System Design: Pin Layout, Whitening Prevention, and Balanced Ejection

Asher Hu — Tue, 01 Sep 2026 04:55:05 +0000

Two months ago, a Bluetooth earbuds project hit trouble at first mold trial: 20 samples out, 8 with ejection whitening (stress marks), 3 stuck to the cavity. The mold maker took one look and said, "Not enough ejector pins, and ejection is unbalanced." We reworked the pin layout overnight, and the second trial passed completely. That experience stuck with me: the ejection system is the most underestimated part of an injection mold, yet it fails more often than people expect.

This article breaks down the key points of ejection system design: how to lay out ejector pins, how to prevent ejection whitening, and how to balance ejection force. If structure designers think these through at the drawing stage, mold trials go much smoother.

Three Principles of Ejection System Design

Distribute ejection force evenly: place pins close to deep cavities, ribs, and dense boss areas.
Use sufficiently large pin diameters: thin, long pins bend, whiten, and break easily.
Keep ejection distance short: enough to demold is enough; every extra millimeter adds scrap risk.

Ejector Pin Layout: It's Not About Quantity, It's About Position

The only principle of pin layout is: let the product be stressed evenly during ejection. Which positions need pins? Deep cavity sidewalls, thick bottom sections, rib intersections, boss roots, and the backs of snaps — these areas have the highest shrinkage grip, and pins must sit directly behind them.

Pin diameters generally range from 1mm to 6mm. Small pins (1-2mm) are used near narrow ribs and small bosses, but watch the length — slender pins over 30mm tend to bend or snap. Large pins (4-6mm) go on flat bottom sections, where the contact area is big and whitening is unlikely. You can mix diameters on one part, but each pin's ejection force should stay roughly balanced, or the thin pins take the load first and break.

Back to the earbuds project. The first version had only 6 pins of 2mm diameter spread evenly across the bottom. But the shell was a deep-cavity part — grip force concentrated on the cavity wall and the sound-port ribs, where there were no pins at all. The result: the cavity wall came out stressed white, and the sound-port area stuck to the mold. We added 4 pins behind the cavity wall and 2 blade ejectors at the sound port, and ejection balanced out. The numbers: pins went from 6 to 12, and the whitening rate dropped from 40% to below 2%.

Preventing Ejection Whitening: Start with Pin Heads and Draft Angle

Ejection whitening is the most common ejection defect — the pin pushes a white mark into the product surface. There are only two root causes: the pin head is too small, or the draft angle is insufficient, leaving the grip force too high.

Pin heads come in two common forms: flat-head and cup-head. Flat-head pins go on internal non-visible surfaces — simple to machine, low cost, but higher whitening risk; they suit positions with a large stressed area. Cup-head pins (also called locating pins) have a small boss on the head that seats into a tiny locating recess on the part back. Contact is more concentrated, whitening risk is lower, but the part needs a 0.3-0.5mm deep recess.

Draft angle directly decides how smoothly the part demolds. General guidance: for glossy plastic parts, at least 1°; the deeper the texture, the larger the angle — VDI 24 and above needs at least 2°. One medical device client used a VDI 30 texture but specified only 1.5° draft on the drawings; the trial showed high ejection resistance and a whitening rate over 15%. After changing the draft to 2.5°, the rate fell below 3%. This correlation held across several of our projects: every extra 0.5° of draft cuts the whitening rate by an average of 5-8 percentage points.

Balanced Ejection: How to Judge It, How to Adjust It

Unbalanced ejection is obvious: the part comes out higher on one side, or tilts at the end of the stroke. In severe cases the part jams halfway out and risks damaging the mold. The judgment method is simple: after the first trial, measure the ejection height at the four corners or symmetrical positions with a caliper. A difference over 0.5mm means visibly unbalanced ejection.

Adjustment takes three steps. First, add pins where grip is high — deep cavity sidewalls, under large ribs. Second, remove pins or shorten them where resistance is low, so they contact the part later. Third, tune the ejector plate stroke so ejection speed is even — too fast creates impact and causes whitening or cracking. A good rule: keep ejection speed at 30%-50% of the mold opening/closing speed.

One easily missed point: return pin (return pin / push-back pin) layout. Return pins sit on the ejector plate and push it back during mold closing. If they're laid out asymmetrically, the plate deflects during reset, shifting pin positions on the next cycle. We recommend return pins near the four corners of the ejector plate, symmetric, with equal lengths.

Special Ejection Methods: Sleeve Ejectors, Blade Ejectors, Stripper Plates, and Lifters

Standard round pins handle most cases, but some structures need dedicated methods.

Sleeve ejectors are for ring structures like bosses and round locating posts. A sleeve surrounds a core pin, so the ring face is stressed evenly and the boss root isn't damaged. When a boss exceeds 4mm outer diameter and 8mm height, consider a sleeve ejector first. Keep the sleeve wall at 0.8-1.5mm — too thin and it cracks.

Blade ejectors are for narrow ribs and thin-wall structures. The cross-section is rectangular, 1-4mm wide and 0.5-1.5mm thick, fitting tight spaces better than round pins. But machining precision is demanding: keep the fit clearance at 0.01-0.02mm — too tight jams, too loose lets flash form. For continuous narrow ribs longer than 15mm, a blade ejector is often the only workable option.

Stripper plates suit large thin-wall parts like transparent covers and faceplates. The whole plate ejects, stress is extremely even, and no pin marks remain — ideal for high-appearance parts that can't show pin marks. The trade-off: a more complex mold structure, costing 30%-50% more than a standard pin mold.

Lifters handle undercuts on the inner side of the part. A lifter moves sideways while ejecting, releasing the undercut. Lifter angle is usually 5-12°; larger angles give more travel but stress the lifter more and wear it faster. In a smart speaker grille project with 8 snap undercuts inside, we used lifters at a uniform 8° with guided lubrication — at 100,000 production parts, lifter wear was under 0.03mm.

The Drawing-Stage Checklist for Ejection Design

Before every drawing release, we run through these checks — here they are for your reference:

Is there a pin behind the deep cavity sidewall?
Is there a pin behind ribs, bosses, and snaps?
Do pins avoid visible surfaces, silk-screen areas, and logo areas?
Is the pin diameter-to-length ratio reasonable? (If the slenderness ratio exceeds 30, switch to a larger diameter or a blade ejector.)
Does the draft angle match the texture depth? (VDI 24 and above needs at least 2°.)
Are the return pins laid out symmetrically?
Is the ejection distance sufficient but not redundant? (Just enough to demold, plus 2-3mm safety margin at most.)

If all seven pass, the ejection system rarely causes big trouble. Internally we call it the "Seven Questions of Ejection" — every new project runs it before drawing release. Since we started in 2021, trial failures caused by ejection issues dropped from 6-7 per year to 0-1.

FAQ: Common Questions on Injection Mold Ejection Systems

Q: What's the difference between ejection whitening and ejection cracking?

A: Whitening is a white stress mark on the surface — stress whitening without material fracture. It's usually fixed by enlarging the pin head area or slowing ejection speed. Cracking is a local crack or fracture — material stress exceeded its limit. It needs structural thickening at that area, or more pins to spread the load.

Q: What pin diameter should I choose? Any standard reference?

A: General ranges: 1-2mm for narrow rib areas, 2-4mm for general areas, 4-6mm for large flat areas. The key is the length-to-diameter ratio — over 30, switch to a larger diameter or a blade ejector. Larger diameters spread force better and lower whitening risk.

Q: When should I use sleeve ejectors vs. round pins?

A: Round pins suit general flat and rib areas — low cost, easy to machine. Sleeve ejectors are specifically for bosses and round locating posts, where the ring face is stressed evenly and the boss root is protected. For bosses over 4mm OD and 8mm height, prefer sleeves over round pins.

Q: How do I set the lifter angle?

A: Generally 5-12°. Larger angles give longer travel but stress the lifter more and wear it faster. For deep undercuts, enlarge the lifter rather than the angle; if the lifter is too narrow, widen it or add ribs. For mass production, keep it within 8° to significantly reduce wear and maintenance.

Q: What if the visible surface can't show pin marks?

A: Three options. One: use a stripper plate — the whole plate ejects with zero pin marks, ideal for transparent covers and faceplates, at 30%-50% higher mold cost. Two: put pins on non-visible surfaces — lay them out inside the part or on the back, leaving only a few small-diameter pins on the visible side. Three: hide them with texture — if pins must cross a visible surface, a deeper texture (VDI 24+) covers the marks almost completely.

This article is adapted from the Hezi Industrial Design official website (hezidesign.com), "Structural Design Field Notes" column.

Hugging Face发布399美元开源机器人MicroDuck——工业设计师拆解小鸭子的3个看点

Asher Hu — Mon, 31 Aug 2026 07:59:26 +0000

Hugging Face 于2026年8月27日通过旗下机器人公司Pollen Robotics推出MicroDuck开源双足鸭形机器人，售价仅399美元。高25cm、重约700g，配备15个电机、摄像头、LiDAR和可抓取喙部，全开源硬件+软件栈。

一、这条新闻说了什么

2026年8月27日，全球最大的AI开源模型社区Hugging Face通过其子公司Pollen Robotics正式发布了MicroDuck，一款仅25cm高的双足鸭形机器人。这不是概念展示，而是可以直接下单预订的量产产品，基础售价399美元。

MicroDuck的完整配置相当扎实：15个伺服电机驱动全身关节，前置摄像头配合微型LiDAR实现环境感知，可活动的喙部能抓取小球等轻量物体。内置麦克风和扬声器，通过USB-C供电和通信，开机即可用手柄操控。最核心的是，整套技术栈完全开源（Apache-2.0许可）：从硬件设计文件到软件SDK，从MuJoCo物理仿真环境到强化学习训练管线，全部托管在GitHub上。

Hugging Face同步开放了7种预训练行为策略（走、跑、转身、踢球、站立平衡等），用户可以开箱即玩，也可以基于仿真环境自行训练新行为后部署到实体机器人上。可选4种配色：Cream、Graphite、Lavender和Sky。

二、从工业设计角度，这条新闻有三个看点

看点一：用"萌宠形态"降低机器人的心理门槛

MicroDuck最聪明的设计决策不是技术，而是选择了鸭子的仿生形态。对比市面上绝大多数机器人，鸭子形象天然具备亲和力。25cm的身高恰好是桌面级宠物尺寸。

这种设计思路和任天堂Labo或者索尼Aibo的逻辑一致：让机器人的第一印象是"宠物"而不是"工具"，用户对它的容忍度和互动意愿会高很多。

看点二：399美元背后的结构成本拆解

在15个电机、摄像头+LiDAR的硬件配置下把价格压到399美元，核心降本手段包括：

塑胶壳体一体化：鸭子外壳采用注塑成型，外观件和部分结构件合二为一
成熟的消费电子供应链：15个舵机采用标准微型伺服电机，BOM成本可控
功能取舍精明：没有做复杂手部、没有高端IMU、没有语音交互，预算集中在运动控制和视觉感知上

看点三：开源硬件+AI训练闭环，重新定义机器人的产品形态

传统消费机器人通常是"买来就是既定功能，不能改也不能变"。MicroDuck的产品定义完全不同：硬件开源意味着用户可以3D打印替换件、改装传感器；AI训练闭环意味着用户可以用MuJoCo仿真环境训练自己的行为策略。

这种"硬件作为AI训练平台"的产品定位，把完整的机器人研发流程打包成一个可消费的产品。

三、设计师可以带走什么

"萌宠形态"在产品设计中的通用价值。 仿生/宠物化的形态语言可以显著降低用户的心理防御。
双足机器人的小型化正在走向现实。 桌面级双足机器人的结构设计门槛已经大幅下降。
开源硬件时代，工业设计的角色在变化。 设计的核心竞争力从"做出这个形状"变为定义"为什么这个形状是最优解"。

原文链接：赫兹工业设计
参考来源：Pollen Robotics MicroDuck · GitHub

Product Thermal Design: Passive Heatsinks, TIMs, and Ventilation — Practical Notes

Asher Hu — Fri, 28 Aug 2026 05:59:25 +0000

Product Thermal Design Handbook: Passive Heatsinks, Thermal Interface Materials, and Ventilation Design — Practical Notes from a Structure Designer

A few years ago I took on a smart speaker project. The client wanted 30W of amplifier power in a sealed enclosure — no fan holes allowed. Our team figured passive cooling would be enough. The first prototype hit 62°C case temperature after half an hour of continuous playback. The client said "it can fry eggs." After that we did a full post-mortem and rebuilt our thermal design playbook — passive heatsinks, thermal interface materials, and ventilation layout — and haven't burned another project since.

Product structure design: thermal management seems straightforward, but it's where many projects fail. This article covers the key points from a practitioner's perspective.

Passive vs. Active Cooling: When to Choose Which

The first step in thermal design isn't crunching numbers — it's picking the approach.

Passive cooling relies on natural convection and radiation — no moving parts, low cost, high reliability, zero noise. It works for products under 10W: smart-home sensors, remote controls, low-power LEDs. We built a temperature/humidity sensor that dissipates just 0.5W — the plastic housing alone handles it, no heatsink needed.

Active cooling covers fans, heat pipes, vapor chambers, and liquid loops. Fan-based designs dissipate more heat but add noise and reliability risk (fan lifespan is typically 30,000–50,000 hours). Heat pipes and vapor chambers are passive conductors but active spreaders — heat pipes reach 5000 W/mK+, 25× that of pure aluminum, ideal for moving heat quickly from source to a radiating surface. On a projector thermal design we used a single 8mm heat pipe to channel LED heat to the housing's cooling fins — the temperature rise stayed under 15°C.

The rule of thumb is two thresholds: if heat flux exceeds 0.2 W/cm², or if the surface temperature limit is below 45°C, passive cooling alone probably won't cut it — go active or hybrid.

Heatsink Design: Aluminum or Copper? How Tight Should the Fins Be?

The heatsink is the most basic thermal component, but newcomers trip here often.

Material: Aluminum (ADC12 die-cast or 6063 extruded) conducts about 200 W/mK; copper does ~400 W/mK. Copper conducts twice as well but costs 4× more, weighs 3× more, and is harder to machine — fins require CNC or welding. In practice, 80% of projects are fine with aluminum. Copper is reserved for high heat-flux, space-constrained cases like high-power LED modules.

Fin pitch is where many designs go wrong. Too tight and air can't flow, dropping efficiency. Under natural convection, aim for 6–10mm between fins. For forced-air (with a fan), tight pitch down to 2.5–4mm is okay. Fin thickness should be 0.8–1.5mm — too thin and they deform; too thick and you waste material and weight.

On a router thermal design we used the die-cast ADC12 housing itself as the heatsink: 3mm base, 15mm-tall fins at 5mm pitch. Under full CPU load for 4 hours, the housing stabilized at 52°C — 8°C below the client's 60°C limit. The beauty is that the housing is the heatsink, saving a separate component and an assembly step.

Thermal Interface Materials: Grease, Pad, or Graphite?

The interface between die and heatsink is microscopically rough, and air fills the gaps — at just 0.026 W/mK. You need a thermal interface material (TIM) with conductivity of 1–15 W/mK to bridge it. Three common types each have their niche.

Thermal grease: highest conductivity (3–12 W/mK), lowest thermal resistance, but messy to apply — must be spread evenly to 0.1–0.2mm thickness. Not ideal for large areas or rework. On consumer electronics we use grease for CPU and GPU hotspots, typically 5 W/mK+.

Thermal pad: 3–8 W/mK, compressible, tolerates larger gaps, easy to apply, automation-friendly. Higher resistance than grease, suited to gaps of 0.3–3mm. On a power-supply design, the 1.5mm gap between MOS and housing was filled with a pad — first-pass OK.

Graphite sheet: exceptional in-plane conductivity (300–1500 W/mK), ultra-thin (0.025–0.1mm), used in phones and tablets where every micron counts. But it's anisotropic — through-plane conductivity is only 5–10 W/mK — so it's for heat spreading, not conduction. We tore down a branded tablet; the CPU heat was spread across the entire backshell via graphite sheet, paired with a metal mid-frame.

Ventilation Design: Position and Size

For naturally convected products, vent position and size decide efficiency. The basic rule: inlets at the bottom or low on the sides, outlets at the top or high on the sides — letting hot air rise creates natural flow. Vent area should be at least 2–3× the heat-source area. Hole diameters of 6–10mm work well — bigger risks EMI and dust ingress, smaller adds flow resistance. On a set-top box, we cut a row of slots (8×80mm) in the bottom shell and matched outlets on top — the chip ran 12°C cooler than the unvented version.

Real Case: Smart Speaker Passive Cooling — from Scorched to Ship

That smart speaker from the opener — how did it end? Three changes.

First, an aluminum heatsink on the amp chip: 45×45×15mm, 4mm base, 6mm fin pitch (natural convection). Second, the housing was switched to die-cast aluminum with a thermal pad on the interior contact. Third, an annular inlet ring at the bottom and an annular outlet ring at the top created a natural convection chimney. The revised prototype held 48°C after an hour of playback — 14°C cooler than the original, and the client signed off.

The lesson: passive cooling isn't "just don't use a fan." It's a four-dimensional exercise — heatsink, TIM, housing material, and airflow path — and every dimension has to be right.

FAQ: Common Thermal Questions

Q: How do you cool a product with a plastic housing?

A: Plastic conducts poorly (~0.2 W/mK), so it can't be the heatsink. Three options: (1) add a metal heatsink inside touching the heat source; (2) use thermally conductive plastic (2–5 W/mK) at higher cost; (3) add vents and a fan. Low-power products handle passive convection and radiation; above 10W, add a heatsink.

Q: Die-cast vs. extruded aluminum for heatsinks?

A: Depends on shape. Die-cast (ADC12) handles complex shapes and can integrate housing and heatsink, but conducts ~96 W/mK — lower than pure aluminum. Extruded (6063) hits ~200 W/mK but only does straight fins. High volume → die-cast; low volume or high performance → extruded.

Q: How often should thermal grease be replaced?

A: Consumer products typically never — grease lasts 5–8 years. In high-temp environments (70°C+) or under constant vibration, it may dry out or pump out; check at 3–5 years. Pads last 10+ years with no maintenance.

Q: Can you cool a waterproof (IPX5) product?

A: Yes. Options: metal housing as heatsink with sealing gaskets; potting the whole board with thermally conductive epoxy; running a heat pipe to a dedicated cooling zone on the waterproof housing. We built an IPX6 outdoor camera that used a thermal pad to conduct heat to an aluminum housing with large cooling fins — passed.

Originally published on the Hezi Industrial Design blog: Product Thermal Structure Design Guide

Insert Molding Design: How to Place Metal Inserts Without Disaster

Asher Hu — Thu, 27 Aug 2026 12:48:23 +0000

Insert Molding Design: How to Place Metal Inserts Without Disaster — 8 Years of Structure Design Notes

Every structure designer has been burned by inserts at some point — a nut seated crooked, an insert causing sink marks or cracks, pull-out force too low, an insert washed away by melt during injection. I've tripped on all of these myself. Insert molding sounds simple: drop a metal part into the mold and inject plastic around it. But metal and plastic have thermal expansion coefficients an order of magnitude apart, and every detail — shrinkage, grip force, locating method — can turn into a disaster. This article walks through the key design principles of insert molding, from insert types and locating structures to wall thickness and defect prevention. All of it is experience paid for with real money on real projects.

Three Common Types of Inserts

Insert molding falls into three categories by purpose, each with completely different design priorities.

1. Thread Inserts (Nut Inserts)

The most common type. Tapping threads directly into plastic fails fast — fine threads under M3 strip after a few cycles — so metal nuts are embedded in the plastic. Copper inserts dominate because copper conducts heat well (fast heat dissipation during molding), has moderate hardness, and gives clean threads after tapping.

We made a portable Bluetooth speaker with an ABS housing whose four corner posts needed M2.5 screws. Tapping the plastic posts directly stripped after three cycles. We switched to embedded M2.5×4mm copper nuts and measured over 45N pull-out, still stable after 500 screw cycles. The key: leave at least 1.5mm of plastic wall around the nut's outer diameter, or the area sinks and bubbles after cooling.

2. Locating / Support Inserts

These locate, support, or conduct magnetism — locating pins in motor brackets, magnetic cores in sensor housings. The biggest challenge is insert positioning accuracy and post-molding offset. In 2024 we made a smart lock panel embedding a stainless fingerprint-module bracket; stainless and PC have expansion coefficients nearly 5× apart. After cooling, uneven shrinkage around the insert caused stress whitening — the part failed cosmetics. We knurled the insert surface (0.3mm knurl depth, 0.5mm pitch) for grip and thickened the surrounding wall from 1.5mm to 2.0mm before the sink marks were controlled.

3. Conductive / Grounding Inserts

Ground springs, conductive terminals, shield interfaces in electronics. Beyond mechanical strength, these need electrical conductivity and reliable solderability. On a TWS earbud charging case, we needed to embed a pair of gold-plated copper pins inside the housing. At only 1.2mm diameter, direct in-mold placement let melt flow push the pins off position. We switched to a two-step process — mold the shell first, then press the pins in with a heat press — improving positional accuracy from ±0.3mm to ±0.05mm.

Five Core Design Principles

Principle 1: Don't Skimp on Wall Thickness Around the Insert

Plastic wall around a metal insert should be at least 0.3–0.5× the insert diameter. An M3 copper nut has an outer diameter around 5mm, so the wall should be at least 1.5mm, ideally 2.0mm+. Too thin and, after cooling, the metal shrinks less than the plastic, which gets pulled into sink marks — or outright cracks in severe cases. In 2023 a beauty device with an M2 nut in the handle kept only 1.0mm of wall; every insert site sank after trial. Three rounds of mold modification later, we finally thickened the mold insert.

Principle 2: Locate the Insert Firmly

How well an insert is held in the mold decides its final position accuracy. Four common locating methods:

Core-pin locating — insert slides over a mold core pin; best for annular inserts. Highest accuracy, but the core pin may pull the insert out when retracted.
Step locating — insert has a shoulder that seats against a mold step. Good for inserts under heavy axial load.
Magnetic locating — magnets in the mold hold the insert. Handy for small inserts, but magnetic force weakens at high temperature.
Spring-blade clamping — slide carries a spring blade pressing the insert. Handles odd shapes at the cost of complex, expensive tooling.

On a smart water cup we embedded an NFC antenna module (15×12×2mm) in the lid with only 1.2mm of wall. Magnetic locating held the module on the moving side; under high injection pressure the melt shoved it 0.8mm off, shifting the antenna and dropping read range from 30mm to 8mm. Switching to step locating plus spring-blade clamping held offset under 0.1mm.

Principle 3: Surface Treatment Decides Grip Force

Grip between insert and plastic relies mainly on surface roughness. Smooth inserts pull out easily. Common treatments:

Knurling — straight or diamond knurl, 0.2–0.5mm deep; the most reliable option. Straight knurl resists rotation; diamond knurl resists pull-out; combining both is best.
Grooves / rings — machine annular grooves into the insert; plastic fills them, forming a mechanical lock. Great when pull-out matters.
Blasting — adds roughness but gains little alone; rarely used by itself.
Chemical etching — creates micropores that plastic penetrates; highest grip, highest cost.

Reference data: in our 2024 pull-out comparison on M3 copper nuts (4mm effective length), a smooth surface gave just 12N, straight knurl (0.3mm) raised it to 38N, and diamond knurl + annular grooves hit 65N. If you need more than 40N, knurl-plus-groove is the standard kit.

Principle 4: Avoid Sharp Corners on the Insert

After molding, the shrinking plastic wraps the metal insert, keeping constant shrink stress around it. Sharp corners (e.g., a square insert's edges) concentrate stress and crack the plastic. We once embedded a square stainless plate in a PC+ABS housing with R0.2 corners. After trial, the plastic surface at all four corners cracked — stress cracks under magnification. We bumped the insert corners to R1.0 and thickened the surrounding wall from 1.2mm to 1.8mm; cracking stopped.

The rule is simple: every edge of an insert needs a radius of at least R0.5; square inserts want R1.0+ on corners. Round inserts are the ideal choice — the most uniform stress distribution, the lowest cracking risk.

Principle 5: Keep the Gate Away from the Insert

Melt shooting straight at an insert can shove it off position, deform it, or blow it right off the locating core. On a smart electricity meter housing, an M4 copper insert sat 15mm from the gate; at 80MPa injection pressure the melt displaced it 0.5mm and the follow-up screws wouldn't drive. We moved the gate 30mm away to an empty cavity and re-routed melt to flow around the insert — offset dropped to 0.05mm.

Basic gate rule: at least 15mm between gate and insert, and let melt fill the non-insert areas first so the insert gets wrapped naturally.

Common Defects and Fixes

Across five years of insert-molding tool trials, our data shows: sink marks around inserts account for 37% of defects, insert offset 24%, insert cracking 15%; the rest are flash, mispositioning, and so on. On-site fixes for the high-frequency ones:

Sink Marks

Symptom: depressions on the plastic surface around the insert. Root cause: metal shrinks less than plastic, so the plastic gets pulled by the insert. Fixes: thicken the surrounding wall (at least 1.5mm), lower packing pressure, extend packing time. If wall thickness is fixed, add an anti-sink groove opposite the insert — a small 0.3–0.5mm pocket that gives the shrinking material somewhere to go.

Insert Offset

Melt impact pushes the insert from its design position. Directions to fix: ① strengthen locating (step + spring-blade double clamping); ② move away from the gate; ③ reduce injection speed (multi-stage injection, slowing near the insert); ④ lengthen the fit between insert and locating core (at least 1/3 of insert length).

Insert Cracking

The insert itself cracks after molding or in service, common with high-hardness heat-treated parts (e.g., hardened steel). Root cause: the insert is too brittle and tears under plastic shrink stress. Fixes: switch to a tougher material (stainless instead of hardened steel), enlarge insert radii (R1.0+), preheat the insert (70–100°C) to cut thermal-difference stress.

Insert Molding vs. Post-Pressing: When to Choose Which

Insert molding isn't always the answer. The alternative is molding the plastic part first, then installing the insert via heat pressing, ultrasonic welding, or cold pressing. Our comparison data:

Insert molding: high positional accuracy (±0.05mm), high grip (60N+ with diamond knurl), high production efficiency (single step, no secondary operation). But higher tooling cost (insert-locating mechanisms), longer tooling trials, and inserts must be placed before every shot.
Heat-pressed inserts: simple tooling, low cost, adjustable insert position. But lower accuracy (±0.15mm), modest grip (typically 20–30N), and the added heat-press step lengthens cycle time.
Ultrasonic insertion: great for small inserts (under M3), fast, decent accuracy (±0.1mm). But plastic type matters (PC/ABS/PA work; POM/PP don't), and ultrasonic energy can damage plastic around the insert.

Our rule of thumb: if pull-out exceeds 30N or accuracy must be within ±0.1mm, go insert molding; if it's just fastening M3+ screws without high-frequency vibration, heat-pressed inserts suffice. For that smart lock, the fingerprint bracket needed 50N+ pull-out — insert molded; the battery-compartment nut (M3, 15N) used a heat-pressed insert to save tooling cost.

FAQ: Insert Molding Common Questions

Q: What's the minimum wall thickness around an insert?

A: At least 0.3–0.5× the insert diameter. For an M3 copper nut (outer diameter ~5mm), recommend 2.0mm+, never below 1.5mm. Below that, sink and cracking risk climbs sharply.

Q: Knurling or annular grooves — which works better?

A: Knurling mainly resists rotation; grooves mainly resist pull-out. Above 40N pull-out, combine both — machine grooves (0.3–0.5mm deep, 1.0–1.5mm pitch) and diamond-knurl between them. We measured 65N on an M3 copper nut with this combination.

Q: Which is cheaper — insert molding or heat pressing?

A: Per-part, insert molding wins (one step), but tooling costs more (locating mechanisms). Below 50,000 units/year, heat pressing is the cheaper total; above that, insert molding wins.

Q: Are some plastics unsuitable for insert molding?

A: Yes. Crystalline plastics like PP and PE shrink a lot (1.5–3.0%), concentrate stress around inserts, and risk sinks — they need thicker walls. PC, ABS, PC+ABS, and PA (amorphous or low-crystallinity, 0.5–0.7% shrink) suit insert molding best. POM shrinks moderately (1.8–2.5%) but has poor thermal stability — watch mold temperature control.

Q: Do inserts need preheating before molding?

A: Large inserts (length >20mm or diameter >10mm) should be preheated to 70–100°C to reduce metal-plastic temperature difference and shrink stress. Small inserts (e.g., M2–M3 copper nuts) usually don't need it. Steel conducts poorly — preheat to 80–100°C; copper conducts well, so 50–70°C suffices.

Originally published on the Hezi Industrial Design blog: Insert Molding Design Guide

Sheet Metal Design Pitfalls: Bend Radius, Flange Height, and Stiffeners — Practical Notes from a Product Structure Designer

Asher Hu — Wed, 26 Aug 2026 01:59:21 +0000

Sheet Metal Design Pitfalls: Bend Radius, Flange Height, and Stiffeners — Practical Notes from a Product Structure Designer

Last week a client brought in drawings for a medical device enclosure and asked if we could switch it from cast aluminum to sheet metal to cut costs. I had to smile when I saw the drawings — bend inner radius specified at 0.3mm, flange height of only 1.5mm. If that process could actually run in volume production, the stamping shop foreman would be swearing. Today let's talk about the traps hiding in sheet metal structure design — all hard-won lessons from years of product structure design work.

1. Bend Radius: Smaller Is Not Better

Designers who have never worked with sheet metal instinctively make the radius as small as possible, thinking smaller means more refined. But sheet metal bending is not injection molding — a small radius on plastic causes stress concentration; a small radius on sheet metal cracks outright.

The bend inner radius is directly tied to material thickness. The general rule: inner radius ≥ sheet thickness. For 1.5mm SECC galvanized steel, the minimum inner radius is 1.5mm — any smaller and it cracks. On one project, a client specified a 0.3mm inner radius on 0.8mm stainless steel; the toolmaker tried bending three times, cracked it three times, and it only held at 0.8mm.

If the design genuinely needs a small radius, there are two paths: add an extra leveling operation (bend first, then press the radius flat — cost goes up), or change material — SPCC bends better than SUS304 and tolerates tighter radii without cracking.

One more detail: the relationship between bend direction and rolling direction. Rolled sheet has a fiber direction; if the bend line runs parallel to the rolling direction, cracking risk is far higher than perpendicular. In 2023 we made a smart-cabinet door from 1.2mm cold-rolled steel with bend lines parallel to the grain — 17 out of 200 parts cracked in the batch. Rotating the blank layout 90 degrees solved it.

2. Flange Height: One Millimeter Decides Pass or Fail

Flanging is a common sheet metal feature for edge stiffness or assembly location. But flange height has a hard floor — flange height ≥ 3× sheet thickness is a basic requirement of stamping.

Example: 1.0mm aluminum needs at least 3mm of flange height. Specify 2mm and the stamping house will discover wrinkling and cracking mid-run — unusable. I learned this the hard way on an automotive product: 2.5mm flange height on 1.0mm stock. During tool trials every flange tore; the die was scrapped, the structure redesigned, and delivery slipped two weeks.

Flange height also depends on material and tooling: stainless flanges harder than cold-rolled steel; aluminum flanges easier than steel but wrinkles more. If height is truly constrained, consider two-step flanging (pre-flange then finish-flange) or flange coining — all of which mean higher tooling costs.

3. Stiffeners: More Is Not Better

Stiffeners (embossments) let thin sheet achieve high stiffness. But many designers can't stop once they start, turning whole panels into grids — and creating new problems.

Stiffener spacing and depth follow rules of thumb: spacing ≥ 5× rib height, depth ≤ 0.6× sheet thickness. For 1.2mm steel that means depth under 0.7mm and spacing above 25mm. Too close together, material can't flow during stamping and wrinkles form between ribs; too deep, the sheet tears outright.

Another trap: stiffeners near bend lines. In 2024, on a power-supply chassis in 1.5mm galvanized steel, a stiffener sat only 8mm from the bend line — during bending it was flattened completely, function gone. The fix: keep stiffeners at least 15mm away from any bend line so the bend zone has room for material flow.

4. Threads in Sheet Metal: Three Options Compared

You can't mold inserts into sheet metal like you can into plastic. Three common options:

1. Tapped extrusions (hole flanging + tapping) — punch, extrude, tap. Cheapest, moderate strength, good for M2–M6 thin-sheet joints. On a 2025 smart-lock project, 1.0mm stainless housing with an M3 tapped extrusion measured ~200N pull-out — sufficient.

2. Clinch nuts (e.g., PEM) — punched hole, press-fit standard nut. High thread strength, survives repeated assembly. Best for highly loaded fastening points. Requires a press riveting machine; unit cost runs about ¥0.3–0.5 higher than tapping.

3. Weld nuts — welded to the back face. Strongest option but more process steps and lower efficiency, for thick plate or extreme loads. In one medical device project, a load-bearing bracket in 3.0mm cold-rolled steel with welded M8 nuts exceeded 1000N pull-out.

Choose based on three factors: thickness, load, and disassembly frequency. Up to 1.5mm with light loads, tapping is fine; frequent removal or heavier loads call for clinch nuts; weld nuts are reserved for extreme duty.

5. Tolerances and Fits: Don't Copy Plastic Parts

Structure designers coming from plastics most often over-tolerance sheet metal. Plastics hold ±0.1mm or better, but sheet metal — with springback in both forming and bending — naturally runs looser.

Typical sheet metal tolerances: overall dimensions ±0.5mm, bend angle ±0.5°, hole positions ±0.2mm. Demanding ±0.1mm forces fine blanking or CNC finishing, doubling cost.

The case I remember best: on a consumer-electronics bracket, the design team located holes at ±0.1mm and quotes came back 60% above normal because every part needed CNC trimming. We switched the interface to a slotted-hole-plus-locating-pin floating design, relaxed tolerance to ±0.5mm, cut cost 40%, and assembly actually got faster.

One principle of sheet metal fit design: prefer slots over round holes, floating location over hard location. Sheet metal is "elastic" — let elastic deformation absorb the tolerance instead of clashing steel on steel.

6. Surface Treatment: Pick Wrong and Everything Else Falls Apart

Common finishes: zinc plating, powder/liquid coating, brushed, anodizing, e-coating. Different materials pair with different treatments — choose wrong and you take a double hit to schedule and quality.

SECC/SGCC galvanized steel — already plated; before coating you must abrade or phosphate it or adhesion suffers. Bare galvanized is fine, but watch cut edges (no zinc layer there) and add rust protection.

SPCC cold-rolled steel — must be finished or it rusts within 24 hours. Coating (powder or liquid) is the usual choice: cheap, full color range. On a 2024 smart-home control panel, SPCC housing with black powder coat passed 72-hour salt spray at about ¥8 per part.

Aluminum — brush then anodize: great feel and corrosion resistance at higher cost. Coating adheres poorly to aluminum; if you need color, brushed + anodized is the answer.

Stainless — usually brushed or mirror-polished, no extra corrosion protection needed. Confirm brush direction with the client — get it wrong and the part looks completely different visually.

FAQ: Sheet Metal Structure Design

Q: What's the minimum bend inner radius?

A: Generally inner radius ≥ sheet thickness (1.5mm steel → 1.5mm minimum). If smaller is unavoidable, change material (SPCC bends better than SUS304) or add a leveling pass. Leave margin and run trial bends before volume production.

Q: What's the minimum flange height?

A: Flange height ≥ 3× sheet thickness — 1.0mm aluminum needs at least 3mm. Two-step flanging helps when space is tight but raises tooling cost. Below this threshold expect wrinkling or cracking.

Q: How do I size stiffener spacing and depth?

A: Spacing ≥ 5× rib height, depth ≤ 0.6× sheet thickness. For 1.2mm steel: depth ≤0.7mm, spacing ≥25mm. Keep stiffeners 15mm+ away from bend lines or they'll flatten during bending.

Q: Tapped extrusion or clinch nut for threads in sheet metal?

A: Depends on the application. Tapping is cheapest for M2–M6, light loads, infrequent removal; clinch nuts give high strength and repeatable assembly at ¥0.3–0.5 extra per piece, for heavy or frequently serviced joints. At ≤1.5mm thickness, favor tapping or clinching.

Q: What tolerances are normal for sheet metal?

A: ±0.5mm overall, ±0.5° bend angle, ±0.2mm hole positions. Requiring ±0.1mm doubles cost via fine blanking or CNC. Use slotted holes and floating locators so the sheet's elasticity absorbs tolerance — far more reliable than rigid fits.

Originally published on the Hezi Industrial Design blog: Sheet Metal Structure Design Guide