Synthesized from 65 episodes of The Amp Hour · AI-generated, every claim cited to a verbatim transcript passage
mentions 2010–2026
Episodes65
Mentions100
Cited here36
First — last#75 — #715
Top guestsEric Van Wyk, Jason Huggins, John Saunders
Related3d printing · design for manufacturing · 3d printer · contract manufacturer · economies of scale

Injection molding is a manufacturing process in which plastic pellets are melted, driven forward by a reciprocating screw, and forced into a precision-machined mold under hundreds of tons of clamping pressure.[379] The process dominates the production of high-volume plastic parts because its economics invert those of most fabrication methods: the tooling carries nearly all of the cost, while the marginal cost of each additional part approaches zero.[130][405] That cost structure, together with geometric constraints imposed by the need to extract a solidified part from a steel tool, shapes a distinct body of design rules, failure modes, and production practices.[153][218]

Process

In a molding machine, plastic pellets are heated and pushed through a screw, which transfers torque to drive the material forward, and the melt is then injected into a mold held closed under multi-hundred-ton clamping force.[379] The mold itself is machined to very tight precision or cut by electrical discharge machining.[379] Die casting is the analogous process for metal: material is injected at high pressure, typically one to two hundred degrees above its melting point, so that it reaches every crevice of the cavity before solidifying.[325]

A spectrum of molding processes exists below full high-pressure injection molding, usable for test parts or short production runs without the multi-hundred-ton machine.[379] At quantities around a thousand parts, work is generally done on bench presses, by machining, by printing, or by low-pressure molding rather than on a true injection molding line.[379]

Design constraints

Draft and ejection

A molded wall cannot be perpendicular to the direction of draw: without a taper, or draft angle, the part tears against the tool as it is pulled free, so a nominally square feature emerges slightly trapezoidal.[153] The interlocking plastic brick is a rare molded product whose outer walls really are square and perpendicular, achieved by placing all of the draft on the inside surface of the part.[379] Because a part’s shape carries the evidence of how it was made — a face angled purely so the part will release reliably from the tool — teardown analysis of commercial parts is a recognized way of learning mold design.[485] When an experienced practitioner critiques a model, the first finding is typically missing draft on a face that would never release.[665]

Geometry and complexity

Pushing complexity into the tool is generally the favorable trade: features such as a battery retention detail require slides and draws that make the mold more expensive, while the extra plastic consumed per shot costs essentially nothing.[277] However, each variant of a design requires its own tool, so an engineer’s time spent modifying a model is trivial next to the cost of the mold the modification implies.[374] A geometry that a 3D printer will readily produce can be impossible to mold, and proving a feature on a printed prototype says nothing about whether it can be manufactured in volume.[127] For this reason, custom mechanical parts are designed so that a prototype can be machined or printed even when production will be die cast or molded, rather than discovering at the end of development that the production geometry cannot be prototyped.[436]

Gates and flow

Gate placement — the position of the injection points in the tool — is a classic first-product failure, surfacing only once parts emerge from the mold.[715] The effective mold designer works alongside the manufacturing engineer, understanding how the machine and the flow of the plastic constrain what the tool can be, and feeds those constraints back into the design.[712]

Tooling

Cost and geography

Tooling cost is the defining economic fact of the process. A production tool for a part as small as a plastic brick has been estimated at tens of thousands to a couple of hundred thousand dollars in the United States.[379] A single small proprietary connector plug carried a tooling cost of roughly eighteen thousand dollars.[167] Quotes vary by orders of magnitude with where the tool is cut; non-recurring engineering for a molded part in China has been quoted at around five hundred dollars.[379] Tooling can consume a crowdfunding raise outright: in one campaign, mold costs alone would have absorbed most of the roughly $170,000 raised.[219]

Hard and soft tooling

Soft tooling costs far less than hardened production tooling and wears out much sooner, but yields fully functional parts of reasonable quality — sufficient for regulated products that need a certifiable unit long before mass tooling exists, and good enough to pass regulatory testing while looking like the real product.[159] At a build of well under two thousand units, a custom hard tool for an enclosure is not the right call, with a soft mold the sensible alternative.[161] The threshold is not absolute: Michael Ossmann ran one product with an injection-molded case while manufacturing it in batches of a thousand at a time.[161]

Tool life and correction

A correction that adds material to the part is cheap, because it requires only cutting more steel out of the tool; designing so that fixes run in that direction keeps changes quick.[218] Worn tools reveal themselves as flash — a rim of excess material around the parting line where the mold halves no longer close cleanly.[586] Across tens of thousands of shots the tool wears and the plastic sags, so prototype shots that look acceptable are not evidence that a part is production-ready, and cosmetic standards shift once finished product is being judged.[287]

Tooling lead time

Tooling lead time has historically set the iteration rate of entire industries: appliance models appeared on cycles of five to seven years because design plus tooling took that long, and process rules were built around that pace even for products that would sell in the thousands.[159] On the Bus Pirate programme, Ian Lesnet fixed the board outline across successive product versions so that one enclosure tool could be amortized over several generations rather than remaining a single-product cost.[125]

The toolmaking skill

The scarce skill in a molded product is the toolmaker rather than the part designer: early units can be machined from solid aluminium at any cost, but production without a senior mold designer produces exactly the ejection failures that stop a line. Jeri Ellsworth, who brought a molded consumer product to production, described the necessary person as a “unicorn” who knows how to make the tools.[173]

Economics and production volumes

Past the tooling cost, the marginal cost of molded plastic is close to nothing: on an inexpensive vacuum cleaner the copper motor windings are the expensive content, while the entire plastic body costs around a dollar on a million-shot tool.[130] Ordinary molded consumer goods cost fractions of a penny per part.[405] Toy companies employ dedicated costing engineers who estimate a part by weighing the plastic, pricing the raw resin, and adding the molding cycle time for that specific geometry.[424]

The rough crossover where prototype-style processes stop being the cheaper route and tooling begins to pay sits around five thousand units, though the type of mold still matters below the high-volume threshold.[218] Design-for-manufacture guidance is correspondingly tiered by tooling budget: cheap tools come with one set of draft-angle and wall-thickness rules, expensive tools relax them, and the highest tier represents capability effectively unavailable to anyone else.[218]

No additive process competes with molding on throughput, because the cycle is simply heating plastic and shooting it into a tool.[172] One crowdfunded product that printed all of its enclosures ended up running a room of printers continuously to keep up — a choice its maker afterwards called a mistake.[172] Additive manufacturing is considered unlikely to displace molding at the commodity end of the market.[405] The process is, however, unforgiving of late commitment: crowdfunding publishes an idea before the team has any way to build it, and campaigns are routinely run by people with no route to a molded part or a manufacturable board who then cannot afford the engineers to get one.[336]

Defence and consumer molding operate at incompatible scales, so a supplier tooled for military volumes cannot simply take commercial work to stay busy between contracts.[705] Choosing among thousands of molders resists rating systems, because a buyer who had a good project leaves five stars while a buyer who had a bad one simply walks away rather than damage a relationship they may need again.[405]

Process variation and failure modes

A part almost never ejects cleanly from a new tool on the first attempt; on one design the part jammed so hard that the ejector pins sheared off before the plastic gave way.[173] Parts leave the tool carrying enough electrostatic charge to bounce back out of the collection bin, striking the side to discharge before they will settle.[218]

The same tool does not give the same part. Shot speed and melt temperature are all adjustable, so pre-production and production runs can differ, and a defect affecting a small percentage of parts is invisible in a fifty-unit build and obvious at five hundred or a thousand.[377] A field failure in a mechanical switch was traced to dimensional variation in the molded parts despite two years of prototypes on what was nominally the same tool never showing it.[377]

Plastic shrinks as it leaves the tool, so the drawing carries a shrinkage allowance rather than the measured dimension. Re-baselining the drawing to the shrunk parts, as a factory may request, would cause the next tool to be cut to the shrunk size and shrink again from there, ratcheting the part smaller with every generation.[564]

Colour contamination from an incompletely purged machine carries the previous run into the next; one thousand-unit order of light-coloured knobs arrived flecked with black, blue, and red, forcing hand inspection of every piece.[586]

Design for manufacture practice

Manufacturability is treated as part of the design rather than a downstream check, on the principle that a part that cannot be made is not a design at all.[712] A mold designer gives the same kind of free design feedback a contract manufacturer gives on a circuit board — thin walls, missing slope, and unbuildable geometry, all identified before any steel is cut.[350] The recurring pattern is that a factory reviewing a finished design finds savings that could have been designed in from the start, which is the argument for pulling manufacturing knowledge earlier into the process.[451]

Published design-for-manufacture curricula cover molding alongside die casting, rotational molding, factory selection, managing cost against quality and schedule, and waterproofing, with the molding module alone running to fifty or sixty slides.[451] Hands-on manufacturing courses have taken students from solid modelling through having tooling cut, treating surface finish and resin choice between ABS and polycarbonate as decisions the designer makes, alongside techniques such as in-mold lamination.[280] Toy-industry practice is to be physically present whenever a tool is opened, with trips batched around the run-up to the holiday season.[414] Taking a molded product to volume calls for a mechanical engineer with tooling experience in both molding and die casting, comfortable with tolerances and with working directly with contract manufacturers — a distinct hire from the board designer.[517]

Alternatives to custom tooling

An off-the-shelf enclosure is frequently good enough for the look and feel of a product, and going fully custom adds certification work on top of tooling.[287] Zach Dunham chose a stock container as the enclosure for a first product, removing the need to produce any enclosure at all and treating the constraint as deliberate rather than a compromise.[350] Where standard chip encapsulation could not be adapted to leave a window for a display, Marcus Schappi’s product instead used a molded enclosure over the assembled board to obtain the desired shape.[189] Jason Huggins held his product on printed parts while the design was still changing, treating assembly capacity — the maximum number of units that could be put together in a month regardless of demand — as the signal for when to move to tooling.[369]

Custom tooling buys freedom at volume, including parts such as bespoke battery form factors that cannot be bought from a catalogue, but it also removes every fallback: with a custom molded part there is no equivalent of buying a project box when the tool fails, and the designer becomes the supply chain with no backup.[365]

References

EpisodeTitleDate
125An Interview with Ian Lesnet - Bus Buccaneer BuilderDecember 10, 2012
127FPGA, Xess, 32 Bit - Quirky Qualitative QuestionsJanuary 7, 2013
130Boeing, PCBs & Startups - Awful Airplane AerationJanuary 28, 2013
153An Interview with Ryan O'Hara - Keyed, Kerfed KaptonJuly 8, 2013
159Interview with Eric Ries - Transorted Testing Tachydidaxy
161Interview with Michael Ossmann - Gifted Grimgribber GrokkerSeptember 2, 2013
167An Interview with Adam Wolf - Brick & Board BiunersOctober 14, 2013
172CAD courses and cross platform creation - Printing Propaedeutic PatternsNovember 19, 2013
173An Interview with Jeri Ellsworth - Intense Illusion IntroductionNovember 25, 2013
189An Interview with Marcus Schappi - Kit Ketch KenophobiaMarch 17, 2014
218An Interview with Eric VanWyk - Meiotic Mountenance MooshimeterSeptember 29, 2014
219Get Smart About Automation - Caducous Cyborg ConcernsOctober 6, 2014
277InterconnectoramaDecember 9, 2015
280New Year Education
287Pull The TriggerFebruary 17, 2016
325An Interview with David Kronstein (Tesla500)November 30, 2016
336An Interview with Bunnie Huang (2nd)
350An Interview with Zach DunhamJuly 3, 2017
365Wait, why is Jeff glowing?October 30, 2017
369An Interview with Jason HugginsNovember 26, 2017
374An Interview with Claire (née 'Clifford') WolfJanuary 7, 2018
377Debugger vs PrintefferJanuary 28, 2018
379An Interview with John SaundersFebruary 11, 2018
405An Interview with Spencer WrightSeptember 3, 2018
414An Interview with Scotty Allen (Strangeparts)November 5, 2018
424An Interview with Julia TruchsessJanuary 6, 2019
436Downward Sloping TraceMarch 31, 2019
451An Interview with Scott Miller (2nd)July 21, 2019
485An Interview with John DayMarch 22, 2020
517Depth and AI with Brandon Gilles and Brian WeinsteinNovember 15, 2020
564Pavlovian CheapskatesOctober 31, 2021
586Fran Blanche Version 3
665Really long needle nose pliersApril 24, 2024
705Psst...Hey buddy, wanna buy an Octopus?October 8, 2025
712Robots Everywhere with Aaed MusaJanuary 19, 2025
715Shiny New Pebble with Eric MigicovskyFebruary 9, 2026