IB Design Technology HL — All Flashcards

The relationship and interaction between people (aspects of the human body) and the products, systems and environments they use.

It makes a design more efficient, more usable, more functional, more effective and safer.

Comfort is only part of the people–product fit. A comfortable product can still fail the environment it is used in — for example a screen that is unreadable in sunlight.

The measurement of human physical dimensions, expressed as a percentile range, which presents the spread of physical characteristics across a population.

Static (structural) is measured with the body still — sitting eye height, popliteal height. Dynamic (functional) is measured while the body moves — forward grip reach, arm sweep.

Age, gender, ethnicity and disability. The age of the dataset itself matters too — populations have grown taller over recent decades.

The vertical distance from the floor to the underside of the knee when seated. It sets seat height, so that a user's feet rest flat and their thighs are not compressed.

The value below which a given percentage of a population falls. The 5th percentile of reach means 5% of people reach less far than that.

Reach is set from the SMALLEST user (5th percentile); clearance is set from the LARGEST user (95th percentile).

It fails half the population by construction, and nobody sits at the 50th percentile in every dimension at once. It suits only dimensions where being wrong either way is harmless.

90% of the user population. Covering the last 5% at each extreme usually costs far more in mechanism and tooling than it gains in users, so it is excluded as a stated decision.

Designing the product to be adjustable, and producing it in a range of fixed sizes. Many products use both.

Advantage: one product covers the whole 5th–95th range with one set of tooling. Disadvantage: the mechanism adds cost and weight, can fail, and only helps if the user actually adjusts it.

Advantage: each item is simpler, lighter and cheaper, with nothing to break. Disadvantage: several sets of tooling and stock, and the user must pick the right size.

A cycle helmet: two or three shell sizes keep each one light, and an adjustable cradle closes the remaining gap — a loose helmet does not protect in a crash.

The three-dimensional space a user can reach without moving from their position. Everything the product asks them to touch must lie inside it.

Normal reach is swept with the upper arm at the side and only the forearm extended — for constant and safety-critical controls. Maximum reach uses the whole extended arm — for occasional controls only.

Reach is sized from the smallest user, pulling surfaces lower and nearer; clearance is sized from the largest user, pushing them higher and wider. No single fixed geometry satisfies both, so the workstation adjusts.

Inside normal reach, however rarely it is used. Frequency sets the zone for ordinary controls, but a safety-critical control must be reachable instantly without stretching.

The study of the systems and biomechanics within the human body — their responses, limitations and capabilities. It tells a designer what a user can DO, where anthropometrics tells them how big the user is.

Visual accuracy, colour perception, strength, fatigue, muscle control and hearing thresholds.

Roughly 1 in 12 men has reduced red–green discrimination, so a colour-only code carries no information for them. Add a second channel — shape, position, a symbol, a pattern or text.

Fatigue lowers what the body can deliver as a task is repeated. A demand that sits just inside the limit for a rested user is outside it later in the shift.

The study of the human mind, including all the senses that send information to the brain — sight, sound, touch, smell and taste — and how they influence the design and development of products.

Smell: an odorant added to naturally odourless natural gas so leaks are detected. Taste: a bittering agent in detergent capsules so a child spits one out.

Carrying critical information through more than one sense — an alarm that both sounds and flashes. A single channel fails completely for part of the user group, while two degrade gracefully.

A flat screen gives no tactile detent, so the user has no confirmation that a press registered. Sound and haptics replace the feedback a physical button would have given by touch.

Waste, pollution and energy consumption — and all three are affected more by how long the product lasts than by what it is made of.

Doubling the time a product stays in use halves its material, energy and waste at once, because one product does the work of two. No material substitution comes close.

For anything powered the use phase dominates, so insulation and control matter. For anything passive the embodied impact dominates, so mass, processes and life matter.

A measure, a cost and an alternative — for example, "the recycled grade costs 12% more per part and removes the coating, recovered within a year".

Cyclic (creates no waste), solar (uses clean energy), safe (causes no harm), efficient (least material and energy), social (supports basic human rights).

That every part can return to a cycle — made from recycled or renewable material and recyclable or compostable at the end. A laminate or a glued joint usually breaks it.

Mass and energy only mean something against the service delivered. A chair 40% lighter that lasts half as long uses more material per year of use.

Below tier one of the supply chain: final assembly is visible and audited, while the fabric, fixings and raw materials are two or three tiers away and rarely traced.

Measuring a product against three accounts — people (social), profit (economic) and planet (environmental) — together rather than measuring profit alone.

Users, the community round the factory, everyone in the supply chain, and the people who repair, clean or dispose of the product. The supply chain is the forgotten part.

A design that loses money is never manufactured, so it helps nobody. Profit is a constraint to be met, not a villain, and the version that ships balances all three.

By changing what the product IS rather than optimising what it was: shipping concentrate instead of water, renting a tool instead of selling one.

Find the account that is quietly paying and name what it pays for. A product that seems unusually cheap or convenient is being subsidised by people, profit or planet.

A decision that moves a cost instead of removing it — a thinner bottle onto the user, a refill onto the household, automation from profit onto people. Check all three accounts after any change.

Safety and legality are not tradeable. Then favour the account that cannot recover — a lost livelihood or habitat — over one that can. Then take the whole-life view of profit.

Durability that costs more per unit is usually cheaper per year of service, so what looks like profit against planet at the till is often profit and planet over ten years.

Linear runs take, make, use, dispose — value created once and destroyed once. Circular is a closed loop where disposal becomes a return and resources are continuously repurposed.

Maintain keeps everything; reuse nearly everything; repair keeps the assembly; remanufacture keeps the parts; recycling keeps only the material.

It was designed to be thrown away and then partly rescued. Circular means the return was designed in — bolted not glued, separable materials, and a business that wants it back.

The embodied energy of forming and assembling a product is usually far larger than the energy in its raw material, so keeping the artefact is worth more than keeping the substance.

Longevity, upgradability, disassembly and dematerialization — with longevity first, because doubling the time in use halves everything else.

"Separated into material streams in under five minutes with standard tools", tested with a prototype, a screwdriver and a timer, by somebody who has not seen it before.

One unobtainable part. A product engineered for ten years with a five-year spares policy is a five-year product.

Extraction and processing waste ore, water and energy far exceeding the mass of the finished part, and manufacture adds offcuts, sprues and rejects.

A technical cycle, where metals and polymers are recovered and remade, and a biological cycle, where materials return safely to natural systems.

Where a product is used once, contaminated in use, lost during use, or too small to sort — situations where collection for the technical cycle cannot happen.

Degradable only breaks into smaller pieces. Biodegradable is converted by micro-organisms to water, carbon dioxide and biomass. Compostable is biodegradable within a defined time with no toxic residue.

A durable product should last; decomposition undermines the property that makes it sustainable. One recyclable polymer with a twenty-year life beats it on every account.

Reuse keeps everything; repair keeps the assembly; remanufacture keeps the parts; recycling keeps only the material. Take-back legislation keeps nothing itself but opens the route.

Standard fixings rather than glue or welds, a documented procedure, and spares available for the whole product life. A short spares policy cancels the design work.

It can require collection but cannot open a glued box or separate a bonded laminate. Every route depends on a design decision taken years earlier.

Rental and leasing, deposits and trade-ins, selling an outcome rather than an object, and producer responsibility — all of which make the return profitable rather than a cost.

Solar, wind, hydro and tidal, biomass and biofuel, and geothermal — each with a limit, which is why a real grid uses several.

Every loop involves collection, transport, sorting, cleaning, remanufacture and redistribution, none of which is free — and for a low-value material it can cost more than the material is worth.

The same material is moved and processed many times. If each movement burns fuel, the total emissions can exceed those of a single new item.

You cannot run a closed material loop on an open energy loop. If the energy is consumed and gone, the system is only half circular.

Examine how the product performs; determine its strengths and weaknesses; turn the weaknesses into opportunities for improvement.

An analysis ties every observation to a consequence, so it changes what you would design next. A description only says what the product is like.

So the redesign can be shown to have improved it. "Tips at 4 kg placed 80 mm in" can be re-measured; "a bit unstable" cannot.

State it with a measurement, say what it costs the user, name the feature responsible, and write it as a testable requirement ready for a specification.

Users: what happens at the task. Manufacturers: what is expensive and where rejects come from. Engineers: what the design is limited by. Service: what breaks and what they refuse to fix. Retail: why it is bought or returned.

People give reasons and priorities but misreport their own behaviour; testing gives numbers against a stated condition but only for what you thought to measure. The strongest findings are where the two agree.

It is existing user research in quantity, written by people at the moment they decided the product had failed them — and it is already in the client's records.

They have adapted to the faults and report the product as acceptable, and the people who gave up — who hold the strongest finding — are absent.

Strengths and weaknesses are internal to the product and true now. Opportunities and threats are external and about what is coming.

Function, performance, usability, features and materials — tagging each entry with its lens stops the SWOT collapsing into four opinions.

It is the weakness written again. An opportunity is external — a market gap, a regulation, a supplier or a rival's absence — that the product could exploit.

Strength with opportunity gives what to push. Weakness with threat gives what to fix first. Those two pairs are the output, not the four lists.

Use it first; test it whole; dismantle in order; analyse the parts; infer the decisions. Each ends in recorded data rather than a photograph.

The usability baseline — how long a first-time user takes, where they stop, what they try — can only be measured on a working product, and it is gone once it is in pieces.

Minutes to open, the number and type of tools needed, and how many parts were destroyed getting in — comparable numbers that can be set against a rival or a redesign.

It is usually there for a reason you have not thought of — a standard, a patent route, a manufacturing constraint, an old failure. Removing it blindly reintroduces a solved problem.

Everybody good = a saturated row, nothing to win. One rival ahead = a gap to close. You ahead = a strength to protect. Everybody weak = the unmet need, and the real opportunity.

A column tells you which existing product is best, and copying it produces a fifth version of what exists. A row nobody has answered gives you something rivals cannot match by tuning.

Otherwise you compare test conditions rather than products. Figures from four manufacturers' websites compare four marketing departments.

A market usually converges on one answer, so every competitor shares the same blind spot. A watering can and a fuel can both pour one-handed and neither competes with a kettle.

The deliberate habit of refusing to accept that something which works is as good as it can be — and converting the irritation into a stated need somebody could design against.

What everybody accepts; what you notice anyway, with a measurement; the question you refuse to stop asking; and the finding stated as an unmet need.

Wheels and suitcases both existed for decades before anyone combined them. Nobody lacked the technology — everybody had accepted that a suitcase is a thing you carry.

"Needs a lighter kettle" chooses the answer before ideation. "Needs to fill and pour without lifting it" leaves several very different products possible.

Empathizing (where does the current product fail?), defining (what baseline must be beaten?), ideation (how do others solve this?), and evaluating (did the new design beat the old?).

A before-and-after comparison on the same measurements — so the final evaluation can show improvement rather than claim it.

It tells you what a redesign would have to give up as well as gain. The part you cannot explain is usually there for a standard, a patent or an old failure.

Competitors share the market's assumptions, including the wrong ones. The principle that solves your problem has usually been found somewhere it was not called the same thing.

Global warming potential, air pollution, water pollution, soil pollution, ecotoxicity, and resource depletion.

Different gases trap different amounts of heat, so each is converted into the mass of CO₂ that would do the same damage, which lets a single total be quoted.

A cotton tote must be reused dozens of times before it beats a thin plastic bag, because growing and processing cotton is far more damaging per bag.

Where it starts and stops decides the answer. Leave out extraction and a recycled product looks free; leave out use and an inefficient appliance looks harmless.

Raw material extraction, manufacture, distribution and transport, use and maintenance, and disposal and recycling.

For a powered product, use. For a passive one, extraction. Ask whether the product consumes anything while it works.

Cradle to grave ends at disposal. Cradle to cradle recovers the material so it re-enters the first stage of the next product.

Doubling a working life halves the impact of everything that made it, per year of service — which almost no material substitution can match.

Design for process — does the shape suit the way it is made? Design for assembly — how many parts and how do they go together? Design for disassembly — can it come apart?

Every DfM change is a change to the drawing. Once the tool is cut the same improvement costs a new tool.

Assembly wants snap-fits, adhesive and the fewest parts; disassembly wants screws, one fastener type and separable materials.

A released snap-fit a tool can open at a named point, or screws where the product must be opened and clips where it need not be.

Draft on every face so it releases, uniform wall thickness so it cools evenly, radii at internal corners, and no undercuts the tool cannot open past.

It cools more slowly than the thin wall beside it and shrinks after the surface has set, pulling the skin in. The fix is a rib at the same wall thickness.

A feature the tool cannot open past, such as a hook on an internal face. It needs a side-action or lifter, or a window in the wall beneath it.

A part designed for machining is a different shape from the same part designed for moulding, so choosing afterwards means redrawing the product.

Does it move relative to its neighbour? Must it be a different material? Must it be separable for assembly or service? If all three are no, combine it.

Combine parts into one moulding; snap-fits instead of screws; symmetrical or un-mistakable parts; and assembly from one direction.

It takes a purchase order, a stock line, an inspection, an assembly operation and a chance of error with it.

When the parts are bonded into an inseparable lump, or when combining them needs a tool so complicated that the saving disappears.

What fails first, and how deep is it buried? In almost every powered product the answer is the battery.

Screws or released clips instead of adhesive; the shortest-lived part reachable first; one fastener type throughout; and materials marked and not bonded together.

It avoids a whole new product — all its extraction, manufacture and distribution. No material substitution comes close.

It costs assembly time on every unit, paid by the manufacturer, for a benefit somebody else receives years later — which is why it is regulated.

Process changes manufacture; assembly changes manufacture and distribution; disassembly changes use and disposal.

Design for disassembly, because repair and upgrade extend the use stage. A three-year life becoming ten avoids two whole products.

Bonding two different materials to reduce the parts count: assembly is faster and cheaper, and neither material can be recovered at the end.

Find the dominant life-cycle stage, remove parts, make the shapes easy to produce, then check it can still come apart.

Their needs (what the product must achieve), their wants (what they would prefer) and their limitations (what they cannot do).

By understanding users' needs and by carrying out their tasks in the specified environment — doing the task where it is really done, not imagining it from a desk.

Products that meet a need but ignore a want get abandoned. A medication reminder that works perfectly but looks medical ends up in a drawer.

A design process that pays particular attention to the needs of potential users of a product through the involvement of users at ALL stages of the design process.

Specify the context of use; specify the requirements; produce design solutions; evaluate against the requirements; iterate.

The product fits a diverse range of user needs and capabilities, and expensive mistakes are found while they are still cheap to fix rather than after tooling.

It takes longer and costs more up front, and a poor user sample lets a vocal minority pull the design away from the wider population.

The user, the task and the environment.

Because no single discipline covers the user, the task and the environment. Each specialist notices a kind of failure the others are not equipped to see.

An understanding of how users perceive information, make decisions and make errors — especially under stress or time pressure. It informs the USER side.

An understanding of how the setting, the culture and the daily routine shape the way a product is actually used. It informs the ENVIRONMENT side.

Field research, task analysis, user observation, interviews, surveys and focus groups.

Watching methods (field research, observation, task analysis) show what users DO. Asking methods (interviews, focus groups, surveys) report what users SAY. The gap between them is where design opportunities hide.

When you need to measure how COMMON something is across a population — after a deeper method has told you what to ask about.

One confident participant can pull the whole room, so the result reflects the loudest voice rather than the user population.

One named user built from research data who represents the attributes of a user population — with goals, context and constraints specific enough to force a design decision.

A written account of the persona performing a task in a specific situation. It surfaces constraints a persona alone does not — the time of day, the lighting, whether a hand is free.

An expectation a whole population shares about how something behaves: red means stop, clockwise means more. Designing against one causes errors under pressure, however logical the alternative.

Advantage: the team shares one concrete user instead of arguing about an abstraction. Disadvantage: a persona built on thin data gives the designer's assumptions false authority.

How closely the prototype resembles the finished product — in appearance, in materials, and in whether it actually works.

It is fast and cheap, so many ideas can be tested; and users criticise a rough model honestly, because it is obviously unfinished.

It is slow and expensive, so few can be made; and the team and users become reluctant to change something that already looks finished.

They explore ideas, refine them, and communicate them. Each job suits a different kind of drawing.

Informal (freehand sketch): fast and disposable, so many ideas can be explored. Formal (isometric, orthographic, exploded): unambiguous and to scale, so it can be manufactured from.

All three axes are drawn at 30° to the horizontal, and nothing gets smaller with distance — so measurements stay true along each axis.

Showing how a product assembles: the parts are separated along the axes they assemble on, in order, so sequence and every component are visible at once.

To answer a question about the design before the product is committed to — testing an idea, gathering user feedback, communicating the design, and finding problems while they are cheap to fix.

By the question. Anything the body judges (weight, grip, texture, fatigue) needs a physical prototype. Comparing many variants, or examining forces inside a part, suits a virtual one.

Comfort is a felt response to pressure, texture, temperature and time. A model can give mass and geometry, but none of those is an output of the simulation.

Scale, aesthetics, materials, function and performance.

As the mechanism alone in a rough open frame. It is quick to build, and testers comment on whether it works rather than on how it looks.

So the comparison is fair. If the shapes differ, any difference in feel, stiffness or finish could come from the shape rather than the material.

What happens over time — wear, heat, fatigue and the failures that only appear after thousands of cycles.

A surface model defines the outer skin only, used for complex curved styling. A solid model defines the whole volume, so it has mass, a centre of gravity and material properties, and can be analysed or manufactured.

The designer supplies constraints — loads, fixings, material, allowed space — and the software produces geometries that meet them, often shapes a designer would not have drawn.

Divides a part into many small elements and predicts stress, deflection and where it will fail under a stated load, before anything is manufactured.

All three simulate USE rather than form or strength: reach and clearance, being inside the design at full size, and force feedback on a virtual control.

Creating physical prototypes quickly, straight from a CAD model, so potential users and design teams can interact with them and give feedback that drives development forward.

A laser cures liquid photopolymer resin layer by layer. It gives the finest detail and smoothest surface of the three, so it suits appearance models. Parts are brittle and need supports.

A heated nozzle melts thermoplastic filament and draws each layer. It is by far the cheapest and most available, but has visible layer lines and is weaker across the layers than along them.

The part is built inside a bed of loose polymer powder that the laser fuses selectively. Overhangs rest on powder already present, so nothing has to be printed to support them.

By their physical, chemical and mechanical properties.

Measure it and put it back. If the material is unchanged the property is physical; if it reacted and was altered, it is chemical.

It lets a designer eliminate candidates systematically, and it lets unlike materials be compared on the same measured terms using published data rather than testing everything.

As natural (timbers, textiles, biomaterials) or human-made (metals, polymers, glass, composites, smart materials).

One that carries its loads through a skeleton of linear members, with the spaces between open or non-structural. A bicycle, a roof truss, a tent.

One where the outer skin itself carries the load, usually because it is curved. It encloses and protects while using very little material for the volume it contains.

Because real products use each structure where its advantage matters — a car has a frame chassis, shell body panels and solid brake discs.

Because the requirements conflict — light fights heat-resistant, warm-looking fights damp-proof — so no candidate satisfies every one and the designer must decide which matter most.

Physical, chemical and mechanical properties, plus aesthetic characteristics. The first three are measured; aesthetics are judged against the brief.

Safety requirements — food contact, flammability, electrical insulation — anything that would make the product fail in use, and anything a standard requires.

One that can be measured or observed without the material changing in any way.

Density, thermal expansion, thermal conductivity, melting point, electrical resistivity and electrical conductivity.

Mass per unit volume, in kg/m³ or g/cm³. It decides how heavy a product is for its size.

Bridges need expansion joints; a metal lid on a glass jar loosens under hot water; two bonded materials with different expansion rates warp or crack.

An aspect of a material that leads to it chemically reacting with another substance — so the material is altered by the encounter.

Corrosion resistance, reactivity (food safe), hygroscopy and flammability.

Its chromium reacts with oxygen to form a thin, self-repairing oxide film. Scratch it and the film reforms, so corrosion cannot get started.

The material absorbs moisture from the air. Timber swells and warps, nylon softens and cardboard loses strength — which is why some polymers must be dried before moulding.

An aspect of a material affected by the application of a force.

Strength is the force needed to break it; stiffness is how much it bends before then. A diving board is strong and deliberately not stiff.

Hardness resists scratching and wear at the surface; toughness is the energy absorbed before fracture. Glass is very hard and not at all tough.

Malleability is being hammered or rolled into SHEET; ductility is being drawn out into WIRE.

Elastic deformation returns to the original shape when the load is removed; plastic deformation stays. A bent paperclip has deformed plastically.

Two or more materials combined to enhance their properties — a matrix that gives shape and a reinforcement that carries the load.

The matrix gives shape, surface and protection, and transfers load into the fibres so they cannot buckle. The reinforcement carries the load and stops the matrix cracking.

GRP: glass fibre in polyester resin. Carbon fibre: carbon fibre in epoxy. Plywood: cross-bonded veneers in adhesive. Reinforced concrete: steel bars in concrete.

They cost more and are labour-intensive to lay up, they fail suddenly rather than bending first, and they are very hard to recycle because the constituents cannot be separated.

A material with one or more properties that can be significantly changed in response to changes in its environment.

Piezoelectricity, shape memory, photochromicity, magneto-rheostatic, electro-rheostatic and thermoelectricity.

A squeeze or vibration produces a voltage — and it runs backwards too: apply a voltage and the material moves. Used in gas-lighter sparks and ultrasound probes.

Because it replaces a mechanism. The material is both sensor and actuator, so there are fewer parts, no wiring and much less to wear out.

One that breaks down in the environment after disposal, or at the end of its useful life, into substances that rejoin natural cycles — water, carbon dioxide and biomass.

Heat, moisture and oxygen. Those exist in an industrial composter and not in a sealed landfill or the sea, so the label alone guarantees nothing.

They close the biological loop: grow the feedstock, make the product, use it, compost it — and its nutrients grow the next crop. Waste is designed out rather than managed.

For a durable product that will be collected and recycled properly. Biomaterials win where collection is unrealistic — packaging, single-use items, things used outdoors.

Following the load: what loads act on it, where each enters, which members carry it, and how it reaches the ground. Describing its appearance is not analysis.

In bending the stress is highest at the outer surfaces and nearly zero on the neutral axis, so material in the middle would add mass and carry almost nothing — the same reason an I-beam has thick flanges and a thin web.

Pure tension: no member is ever pushed, so nothing can buckle and every strand can be extremely thin. The same principle as a suspension bridge or a bicycle spoke.

A member carrying nothing is mass, cost and an extra joint that can fail, for no return — usually the fastest improvement available in an existing product.

The allowable load is exactly the load at which the structure fails, so any variation at all — material, manufacture, use or age — causes failure.

Nothing is made exactly to drawing, no material is exactly as published, no load is exactly as calculated, and nothing stays as strong as it started.

Up: a severe consequence, uncertain or dynamic loads, an inconsistent material, a part that cannot be inspected. Down: certainty, consistency, inspection, and a margin that costs a lot of mass.

Loads are known precisely, material is tested batch by batch, every part is inspected on a schedule, and every kilogram costs fuel for the life of the aircraft.

What carries the load. A frame carries it along slender members meeting at joints; a shell through a thin curved surface; a solid through its bulk, mostly in compression.

Ask what a hole would do. A frame loses one member, which bracing may cover. A shell loses most of its stiffness, because the surface was the structure.

A four-bar rectangle folds into a parallelogram with no member changing length. A diagonal makes two triangles, and a triangle cannot change shape unless a member stretches or shortens.

It carries load in bending, which is the weak way. Curvature, a fold or a rib converts bending into in-plane forces, which is what makes a shell stiff.

The supports: a pin lets the beam rotate there, a fixed end does not, a cantilever has one support only, a continuous beam has three or more.

A simply supported beam in the middle; a cantilever at its root; a fixed beam at the ends as well as the centre; a column sideways, by buckling.

A bracket is a cantilever and the bending moment is largest at its root and zero at its tip, so material is placed where the bending is.

It becomes unstable sideways at a load far below its crushing strength. Doubling its length makes it about four times easier to buckle, so shortening or bracing it helps far more than a stronger material.

Compression squashes, tension stretches, shear slides one part across another along a plane, torsion twists about the long axis, and bending stretches one face while squashing the other.

A loaded beam is in tension along one face and compression along the other, with a neutral axis between where the stress is zero — which is why an I-beam has thick flanges and a thin web.

Static forces are steady — self weight, a shelf of books, a parked car. Dynamic forces change — a person walking, wind gusting, a door slammed — and they cause fatigue at a fraction of the static load.

A closed section carries the twist as a continuous shear flow all the way round. Slotting it lengthways breaks that loop, so torsional stiffness collapses while bending stiffness barely changes.

Stress is the force divided by the cross-sectional area carrying it. Strain is the extension divided by the original length, so it has no units.

The elastic region, whose gradient is Young's modulus; the yield point where deformation becomes permanent; the ultimate strength at the peak; and fracture at the end.

Past yield the deformation is permanent. A part loaded to its ultimate strength has not broken, but it has deformed for ever — which for almost every product is already a failure.

Stiffness, which is the gradient of the elastic line. A stronger material with the same modulus deflects just as much; section depth is what changes deflection.

The gradient of the elastic line on a stress-strain graph — stress divided by strain. It measures stiffness: how little a material deflects under load.

A door seal that fills a varying gap and recovers; a snap-fit catch that bends to click and springs back; a suspension bush or grip that absorbs vibration.

Steel is about three times stiffer than aluminium, while doubling a beam's depth is worth roughly eight times — so changing the section is almost always cheaper and lighter.

Load follows stiffness, so the stiff part carries almost all of it — which is how a metal insert ends up taking the entire load in a plastic housing.

The forces sum to zero in every direction, and the moments sum to zero. Balanced forces alone still allow a structure to rotate.

A force multiplied by its perpendicular distance from a pivot. Doubling the distance doubles the turning effect for the same force.

Tipping, when the weight's line of action falls outside the support base; sliding, when a horizontal force beats friction or the fixings; and material failure, when a member yields, fractures, shears or buckles.

Two of the three failure modes — tipping and sliding — have nothing to do with strength, and no stronger material prevents either. Both are fixed by geometry, friction or a fixing.

Struts and ties, shape, lamination, and composite materials. Three of the four add almost no material.

It uses the same material formed differently, and bending stiffness rises with the cube of depth — so a fold, a rib or a tube buys enormous stiffness for no extra material.

Thin layers are bonded with alternating grain, so the weak direction of one layer is the strong direction of the next. It also resists splitting and warping and allows curved forms.

From the bottom hinge corner up to the top latch corner, so it works in compression. The other way it is in tension and a timber joint pulls apart.

The ratio of a structure's absolute strength — the load at which it actually fails — to the allowable load it is designed to carry. Allowable load = absolute strength ÷ SF.

Material variation, manufacturing defects, users overloading or misloading the product, degradation over time from corrosion and fatigue, and the assumptions made in the calculation.

The consequence of a failure, the certainty of the loads, and whether the part can be inspected. A rope carries a far larger factor than a shelf bracket.

It means more material, mass, cost and embodied impact, and sometimes a product that is worse to use. An over-specified part is a design failure, not a cautious success.

Linear, rotary, oscillating and reciprocating.

Both reverse. Reciprocating motion travels back and forth along a straight line; oscillating motion swings back and forth along an arc about a pivot.

Linear — a drawer or a lift. Rotary — a fan or a drill bit. Oscillating — a pendulum or a windscreen wiper. Reciprocating — a piston or a jigsaw blade.

Because a mechanism is defined by what it converts. "The handle" says nothing; "rotary motion applied by hand at the handle" tells you what has to come out the other end.

Changing the direction of a movement, altering the magnitude of a force by making the arms unequal, and making a part follow a particular path.

A fixed pivot is attached to the frame and cannot move; a moving pivot travels with the bars. Finding the fixed pivots first is how a linkage is read.

Reverse motion — one fixed pivot, the output moves the opposite way. Parallel motion — two fixed pivots with equal bars, the output stays parallel. Bell crank — a right-angled bar turning movement through 90°.

Because two equal bars stay parallel, so the tray translates without tilting. A single lever swings its far end through an arc and the contents slide off.

Input, process and output. The input and output are motions; the process is the mechanism that converts one into the other.

The type of motion, its speed, its direction, or the force it delivers.

Input: reciprocating motion from the hand. Process: a piston and washer in a cylinder with a one-way valve. Output: compressed air flowing one way into the tyre.

The system sensing its own output and correcting it — an engine governor closing a throttle as speed rises, or a thermostatic valve closing as a radiator warms.

MA = load ÷ effort — how many times a mechanism multiplies the force applied to it. On a lever it equals the effort arm divided by the load arm.

Distance. The effort end travels further than the load end, in the same ratio, because force × distance in equals force × distance out less friction.

When speed or reach matters more than force — tweezers, a fishing rod, a broom, the human forearm. All multiply movement and divide force.

Reduce friction with bearings and lubrication, stiffen members so movement is not lost in flexing, and shorten the chain so fewer joints add losses.

Gear-driven, belt-driven, cam, lever and linkage.

Gears — speed and force with no slip. Belt — speed across a gap, quietly. Cam — rotary into a timed rise and fall. Lever — force. Linkage — direction or path.

Driven divided by driver — teeth for gears, diameters for belts. Small driving large is slower and stronger; large driving small is faster and weaker.

Shafts far apart or not perfectly aligned, quiet running, lower cost, no lubrication needed, and a slip under overload that protects the motor.

Because each mechanism changes one thing. Chaining gives cheap standard parts, faults traceable to one stage, and ratios changeable by swapping a single component.

Efficiency, free play and wear. Efficiencies multiply: five stages at 95% each deliver about 77% of the input.

Write the motion at each junction first, then name the mechanism that converts each pair. One stage at a time: motion in, motion out, reason.

Belt — reduces motor speed. Gears — synchronise the two shafts. Crank — rotary into reciprocating for the needle. Cam — times the fabric feed. Linkage — clamps the presser foot.

Spur, bevel, rack and pinion, worm and wheel, ratchet and pawl, idler, and compound.

Teeth on the driven gear divided by teeth on the driver. A 10-tooth driver into a 40-tooth driven gear is 4:1 — a quarter the speed, about four times the torque.

The direction of the output, so it turns the same way as the input, and it bridges a gap between shafts. It does not change the overall ratio.

It gives a very large reduction in one step and cannot be back-driven, so the load holds wherever it stops.

A driver pulley on the input shaft, a driven pulley on the output shaft, a belt running in the grooves of both, and a tensioner or adjustable mounting.

Driven diameter divided by driver diameter. A 40 mm driver turning a 120 mm driven pulley is 3:1 — a third of the speed, about three times the torque.

It spans a gap with two parts, runs quietly with no lubrication, tolerates shafts slightly out of line, and slips under overload instead of shearing a tooth or stalling a motor.

Where two shafts must stay synchronised, because a belt slips and stretches. An engine camshaft, a sewing machine or a printer needs gears or a toothed timing belt.

A shaped disc turns on a shaft and a follower rides its edge. Where the profile is far from the shaft the follower rises; where it is close it drops. The profile is the motion.

Pear — a dwell then one rise. Circular and eccentric — one smooth rise, set by the offset. Triangular — three rises per turn. Oval — two. Snail — a gradual rise then a sudden drop, one direction only.

A stretch of constant radius on a cam profile, where the follower holds still while the shaft keeps turning. It is why a pear cam suits an engine valve.

The offset of the shaft from the centre of the disc, not the disc's diameter. Double the offset and the rise doubles.

By which of the fulcrum, load and effort sits between the other two. F, L, E in class order: fulcrum in the middle is first, load in the middle is second, effort in the middle is third.

Second class. The load sits between the fulcrum and the effort, so the effort arm is always longer — a wheelbarrow, nutcracker or bottle opener.

To multiply movement, speed and reach rather than force. A fishing rod, a broom and the human forearm all trade force away deliberately.

Effort arm divided by load arm, both measured from the fulcrum. Effort × effort arm = load × load arm is the same relationship written as moments.

Input — a device that senses a change and produces a signal. Process — the thing that decides. Output — a device that acts on the world. Feedback — a path carrying a measurement of the output back to the process.

Because a circuit cannot act on a quantity. Name the device that turns it into a signal — a thermistor, an LDR, a microphone, a push switch.

Ask whether anything in the circuit measures the OUTPUT. A timer counts whatever happened; a thermistor inside an oven measures the result of the heating.

Input: a thermistor in the cavity. Process: a controller comparing with the dial setting. Output: the heating element. Feedback: the thermistor measures the air the element is heating.

By the sense they reach: seen (LED, lamp, LCD, braille display), heard (buzzer, speaker, headphones), felt (haptic), done (motor, relay), kept (printer, plotter).

Right when the user is not looking — every alarm depends on it. Wrong in a shared or quiet place, because it reaches everyone, and useless to a deaf user.

A pin supplies a few tens of milliamps; a motor needs hundreds and a mains load needs amps. The pin switches a transistor or relay, which carries the load current from the supply.

A flyback diode. When the current stops, the collapsing magnetic field produces a reverse voltage spike that would destroy the switching transistor.

An open-loop system acts on its input and nothing measures the result. A closed-loop system measures its output, feeds it back, compares it with what was asked and corrects the difference.

To let the system correct for everything the designer could not predict — a cold room, a heavy load, a slipping wheel, a worn part. Accuracy is the consequence, not the purpose.

Open: a toaster on a timer, a doorbell, fixed-cycle traffic lights. Closed: an oven with a thermistor, a cruise control, a lift, a powered wheelchair with wheel encoders.

A loop that corrects too hard overshoots, corrects back and oscillates. A dead band and gentler correction are the usual fixes.

A high-gain voltage amplifier with a non-inverting input (+), an inverting input (−) and a single output. It amplifies the difference between its two inputs.

Its open-loop gain is about 100,000, so a millivolt of difference saturates the output. Two feedback resistors return part of the output to the inverting input, and their ratio sets a usable gain.

Interference picked up along a cable and drift with temperature appear on both inputs equally, so amplifying the difference rejects them. That is why sensor leads run as pairs.

Between the sensor and the microcontroller in a smart thermostat, connected doorbell, fitness band or kitchen scale — amplifying and filtering a millivolt signal so the converter can read it.

A microcontroller and its software built into a larger product to carry out one specific task — one program for the life of the product, invisible to the user.

Functionality — things the product could not do before. Efficiency — using only what the job needs. Automation — making decisions the user used to make.

The useful behaviour often exists only between them. A thermostat, a boiler controller and radiator valves each do one small job; the heating system lives in the messages they exchange.

Bugs, security updates for the product's whole life, repairs needing software nobody else has, and features that die when a server is switched off. Keep the core function working without the network.

So a circuit drawn in one country can be read in another by someone who does not speak the language. The diagram shows what is connected to what, not what the board looks like.

A variable resistor is a resistor box with an arrow across it. An LED is a diode triangle-and-bar with arrows leaving it. A relay is a coil drawn beside a separate pair of contacts.

The wires are connected. Wires crossing without a dot are not connected — missing this is the commonest way a circuit is misread.

As input, process, output. Find the supply rails, then the sensing chain, follow the signal to whatever decides, and end at the load the current finally does work in.

Safe in use and in failure; energy-efficient for the job; low standby power; and repairable or recoverable at end of life.

A switched-mode supply instead of a linear one, LEDs instead of filament lamps, and a processor that sleeps between events with a sensor waking it only when needed.

Because it is drawn every hour of every year in billions of devices. It is capped across the EU at 0.5 W per device precisely because the per-product figure is small and the total is not.

By shortening its life. A control board lasts a fraction of the mechanical parts, is rarely repairable and is not stocked for long, so a working machine is scrapped for a cheap part.

Analogue: any value between its limits, changing continuously. Digital: only a fixed set of discrete values, usually two — on and off, 1 and 0.

Everything between the samples, and the difference between each reading and the nearest available level. More bits and faster sampling reduce both without removing them.

A receiver only has to tell a 1 from a 0, so noise picked up on the way is discarded rather than accumulated. The thousandth copy is identical to the first.

Because the world is analogue at both ends. A microphone senses continuous air pressure, the signal is digitised for processing and transmission, then converted back for a loudspeaker.

Voltage — volt (V). Current — ampere (A). Resistance — ohm (Ω). Power — watt (W). Frequency — hertz (Hz). Time — second (s).

V = I × R, so I = V ÷ R and R = V ÷ I. Power P = V × I.

p (pico), n (nano), µ (micro), m (milli), the unit, k (kilo), M (mega), G (giga), T (tera). Each step is a factor of one thousand.

Ask whether the value is plausible for that component. Resistors are usually kΩ, capacitors µF, nF or pF, LED currents mA. A 4.7 MΩ resistor in an LED circuit would light nothing.

A circuit whose single output is decided entirely by the combination of its inputs, where every signal is 1 or 0. Its truth table lists every input combination and the output for each.

AND: on only when both inputs are on. OR: on when either or both are on. NOT: the opposite of the input. XOR: on when the inputs are different.

Four. Each input has two possible states, so two inputs give 2 × 2 = 4 combinations. Three inputs give eight.

A gate responds in nanoseconds, has no program to crash or hang, cannot be broken by an update, and its behaviour is fixed by the wiring — so it works when the software does not.

Passive components limit, store or divert what is already in the circuit — resistor, capacitor, switch, relay. Active ones control the flow — diode, LED, transistor.

Stores a small charge and releases it. It blocks steady voltages and passes changing ones, which is how it smooths a supply, sets a timing delay and couples an audio signal.

It carries far larger currents and keeps the two circuits electrically separate. That isolation lets a 5 V microcontroller switch a 230 V load safely.

When the current stops, the collapsing magnetic field produces a large reverse voltage spike. The diode gives that current a path to decay instead of destroying the switching transistor.

A resistance that falls as light rises — hundreds of ohms in bright light, megohms in darkness. It is not a voltage until it is placed in a divider.

A circuit cannot act on a resistance. In series with a fixed resistor across the supply, the sensor's changing share of the supply appears as a changing voltage at the junction.

Roughly equal to the sensor's resistance at the value to be detected, so the junction sits near mid-supply where the voltage change per unit of sensing is largest.

Switches are digital — on or off. LDRs, thermistors, humidity sensors and microphones are analogue, giving continuously changing values.

Making a real sensor signal usable with analogue components: amplifying it because it is tiny, filtering it because it is noisy, and comparing it with a reference so a decision comes out.

Converting the signal to a number and letting software decide. Averaging, delays, dead bands, displays and logging all cost nothing extra in components.

When the job is fixed, the response must be immediate, the cost must be pennies, and nothing must depend on software running correctly — a smoke alarm, an amplifier, a protection circuit.

As soon as the behaviour is complicated, needs changing later, needs memory or a display, or has to communicate with anything else.

A programmable integrated circuit — a processor, program and data memory, a clock and converters on one chip — with pins connecting directly to sensors and output devices.

Timers, comparators, counters and logic gates, often a whole board of them. In a washing machine it replaced a motor-driven cam stack.

Delays and sequences; averaging and dead bands; displays, menus and logging. Each would need extra components in a discrete design.

Software to write and debug, faults that may be bugs, a response in milliseconds rather than nanoseconds, and a product whose life is limited by how long its software is supported.

Additive, subtractive (wasting), forming, joining and finishing.

Ask what happens to the amount of material: more is additive, less is subtractive, the same in a new shape is forming, pieces becoming one is joining, and only a changed surface is finishing.

Moulding needs a material that can be melted or softened and set in a new shape. Timber burns rather than melting, and its fibres along a grain cannot be reformed.

A pairing of the technique with a reason — a material property, the volume required, a tolerance, a cost, or the environment the product lives in.

Protect the material from its environment, improve how the product looks and feels, and by doing both extend the product's working life.

It is not a coating. The oxide layer is grown out of the aluminium itself, so there is nothing on top of the metal to lift — though it is thin and brittle.

Zinc in a galvanised coating corrodes in preference to the steel, so it goes on protecting even where a scratch has exposed the steel beneath.

A finish that doubles a product's life halves the impact of everything that went into making it — usually a larger saving than any change of material.

In order: how the basic shape was made, what brought it to size, how the parts were joined, and what was done to the surface — with a reason at every stage.

A material property, the volume required, a tolerance, a cost at that volume, or the environment the product lives in.

At high volume, moulding is far faster and cheaper per part. The printed stage is where the shape was settled before the tooling was cut.

Deep-drawn body, trimmed and machined base, printed handle prototypes, riveted handle, anodised inside and polished outside.

Only in how each layer is made. LOM glues down sheets and cuts each outline, FDM extrudes melted thermoplastic, and SLA cures liquid photopolymer with light.

The bond between layers is weaker than the material within a layer, so the part breaks along a layer. Build orientation decides which direction it can be loaded in.

Nothing holds up material printed over air, so overhangs steeper than about 45° need support printed beneath and removed afterwards — costing material, time and clean-up.

Internal channels, lattices and shapes that would need a mould split into many pieces, because no tool ever has to be withdrawn from inside the part.

Not the machine but the specification. A base model answers one question and is thrown away; a production part must meet a specification, survive its service life and be repeatable.

There is no tooling to pay for, so the first part costs what the thousandth costs. Below a few thousand parts that beats moulding.

Making every unit different at no extra cost. A printer works from a file, so a hearing-aid shell or dental aligner matched to one person costs the same as a standard part.

At high volume, where moulding makes a part in seconds; on large simple shapes; and anywhere the layer-to-layer weakness or surface finish cannot be accepted.

A thin layer of powder is spread across a bed and a laser or electron beam fuses the cross-section. The bed drops, another layer is spread, and unfused powder supports the part.

The surrounding unfused powder holds the part up, so overhangs and internal cavities need nothing printed beneath them. The loose powder is recovered and reused.

No tooling, so the first part costs what the thousandth costs. Below a few thousand parts it beats moulding, and it removes the lead time and minimum order too.

Build volume caps part size, cycle times are hours, metal powders are costly and need careful handling, and safety-critical parts need qualification of powder, parameters and post-processing.

3D printing with a material whose physical or chemical state changes over time in response to an external stimulus — pH, temperature, water or light.

Printed in its permanent shape, warmed and deformed into a temporary one, then cooled so the chains lock. A stimulus lets them move again and it returns to the printed shape.

A vascular stent that opens at body temperature; a drug capsule that opens at intestinal pH; flat-pack that folds when wetted; shading that closes in direct light.

Few materials, high cost, poor recyclability, changes that are usually one-way or repeatable only a few times, and products that are hard to test, certify and trust.

Rotation of the print bed and rotation of the extruder head, which together let material be laid along a curved path rather than in flat horizontal slices.

A printed part is weakest between layers. When the layers follow the curve of the part, a load along that surface runs along the layers rather than peeling them apart.

The surface, because a curve is printed as a curve rather than a staircase, and the support material, because an overhang can be rotated until it is no longer an overhang.

Curved, loaded and made in small numbers — implants, prosthetic sockets, automotive brackets, aerospace ducts. Flat, lightly loaded or mass-produced parts gain nothing worth the cost.

In turning the work rotates and the tool is fed along it, so it makes round shapes only. In milling the tool rotates and the work is moved past it, so it makes almost any reachable shape.

It removes very little material at a time with many small hard particles, holding tolerances and producing surfaces a cutting edge cannot — grinding, honing and polishing.

Everything removed is waste that was bought, transported and often processed first. A machined aerospace bracket can start as a block ten times its finished mass.

Accuracy and surface finish no other category reaches, in any material, with no tooling to make first — which is why prototypes, one-offs and mould tools are machined.

Nothing is added or removed — the same material takes a new shape with almost no waste. All of them trade an expensive tool made once for a very low cost per part.

Blow moulding makes thin-walled parts with a narrow neck, such as bottles. Rotational moulding makes large seamless hollow parts, such as tanks. Vacuum forming makes open shapes with one good face.

The precision steel tool costs thousands and must be spread over the run. At 400 parts the tooling dominates; at 400,000 it is a fraction of a penny each.

A constant cross-section along the whole length — window frames, pipe, guttering, curtain track and aluminium sections.

Does it ever need to come apart? Temporary joints allow repair, upgrade and recycling; permanent ones are stronger, lighter, sealed and cheaper to assemble.

For dissimilar materials that cannot be welded, thin sheets a rivet would tear, and joints that must be sealed — because load is spread over the bonded area rather than at points.

Compatible materials only, heat that distorts and weakens the area around the joint, skill or a robot, and no way to reverse it.

It is one object that can only be shredded. A screwed product is a pile of materials that can be separated and sorted.

Research question → method → data you will collect → the design decision the data will settle.

A user group you could go and find this week, and a task those users carry out with a beginning and an end.

Every method has a blind spot. A watching method (observation, field research) shows what happens but not why; an asking method (interview, focus group) gives reasons but people misreport what they did.

Would a different answer change the design? If nothing changes whichever way the data falls, the question is decoration and the hours are better spent elsewhere.

Field research, user observation, interviews, questionnaires and focus groups.

The measurable facts that define a target population — age, gender, income, education, occupation, location, household type and relevant abilities. They define who is in the group; they never say what an individual wants.

It shows exactly what happens and never why. People also behave differently when they know they are watched.

Raw records → a repeated theme → a statement about the user → the want or need. Only the last belongs in a specification.

One specific, named, fictional person built entirely from real user research, standing in for a whole group of users so that design decisions are argued about a person rather than a category.

Who (name, age, situation), behaviour, environment, goal and frustration — each traced back to a research finding.

The design must satisfy it above all others: if the product works for nobody else, it must still work for the primary persona.

One person cannot represent the range of a real population, so needs at the edges get designed out — and an invented persona is worse than none, because the invention gets designed for.

Learnability, efficiency, memorability, errors and satisfaction.

Learnability is the first use — can a newcomer finish unaided? Memorability is a return after weeks — can they still do it without relearning?

How many errors are made, how serious each one is, and how easily it can be recovered from.

A measurement taken the same way before and after: for example 6 of 20 first-time users completing the task unaided, rather than "it is hard to learn".

A breakdown of a user's goal into the steps and sub-steps they actually perform, built from observation, used to find where the task goes wrong.

A goal written as the product's function assumes the solution. "Get the child safely strapped in" leaves every design open; "operate the harness" does not.

A step where errors happen, where the task takes far longer than it should, or where users abandon it. Each one becomes a specific design improvement.

You add sub-steps only where users hesitate, struggle or make mistakes. Breaking every step to the same depth produces a big diagram that points at nothing.

Empathize; define the project; ideation and modelling; designing a solution; presenting a solution.

Research findings from real users, a primary persona, and a user journey with pain points marked.

A problem statement, and a design specification whose criteria are testable and split into essential and desirable.

Because testing and users keep producing information that invalidates an earlier decision. A failed test sends you back to ideation; a missing requirement sends you back to the specification.

Diverge — generate many genuinely different approaches with judgement suspended — then converge, narrowing against the specification. They must never happen at the same time.

Brainstorming, mind mapping, thumbnail sketching, morphological analysis, and analogy or biomimicry.

They solve the problem by a different principle, so they fail and succeed for different reasons. If two ideas fail for the same reason, they are one idea.

Far enough to be judged and no further — card, foam or a quick CAD massing — all at the same fidelity, and modelling whatever part would kill the idea if it did not work.

The design specification and user needs — never personal preference. Criteria in the rows, ideas in the columns, every cell scored.

Score every cell, remove anything that fails an essential criterion, then use the desirables to choose between the survivors, and finally check the survivor against the persona.

Because several small desirables can then outvote a single essential — the exact decision the essential-desirable split exists to prevent.

An improved idea rather than a winner: the comparison shows which part of each idea was doing the work, and the developed design takes the best parts of several.

Model — build the cheapest thing that answers the current question. Test — with real users, measuring something. Refine — change the design because of what happened. Then repeat.

Because the questions get harder. Reach can be answered with card at the right height; stability needs real weight and real materials.

So an improvement can be attributed to a known change. Two changes at once hide which one worked, and can hide one making things worse.

When every essential criterion has been measured and met — not at the deadline and not when the model looks finished. Criteria not yet met are recorded honestly.

To generate performance data when tested with end-users — not to show what the product will look like.

From the question being asked. Card at the right height answers reach; a real pivot answers grip; real materials answer stability and mass.

Decide the measurement before the users arrive, give a task rather than an opinion prompt, do not help or explain, and record exactly the quantity the specification criterion names.

Because a printed part is weak across its layers and a card model has no weight. An evaluation that claims more than the model could show is worth less than one that names its limits.

Dimensioned orthographic views of every part, dimensions from a datum, a stated scale, detail views of joints at a larger scale, and an exploded assembly with numbered parts and a parts list.

Chained dimensions accumulate error along the chain; from a single datum each error stays independent.

Could somebody who has never seen the prototype build it from these sheets alone, without asking a question? Anything missing becomes a decision somebody else makes for you.

It carries appearance and proportion but no dimensions, scale, sections, materials or parts list — nothing can be cut or ordered from it. Its job is presenting the solution.

Why does this need to exist, what does it do, and how do you know it works — in that order.

It makes a point faster than a sentence could: the feature shown in use, a ghosted view of something hidden, or a section proving a claim such as cleanability.

Because measuring the miss and naming the next step reads as control of the project, while dropping the line reads as not having tested — and the gap is easy to see.

On the design intention, in the words of the problem statement — the stated need, now met — and, for a specific audience, what it changes for them.

Primary is first-hand data you collected for this project. Secondary is data collected and published by a third party for their own purposes.

Qualitative describes — reasons, feelings, behaviour, the users' own words. Quantitative counts — times, frequencies, dimensions, percentages.

Your own timings, tallies or measurements: for example how many seconds a nurse spends repositioning a table, recorded across twenty meals.

It is fast and free, it sets the standards and numbers you cannot measure, and it stops you spending limited primary hours on a question that is already answered.

User observations, interviews, surveys, questionnaires, focus groups, material testing and product analysis.

User requirements, design specification criteria, the persona, and suggestions for further development of a solution.

A stated condition rather than a vague quality — 200 wipes with a named disinfectant, not "is it durable" — plus identical samples, the same procedure each time, and a control.

Opinions tend to support whatever the designer already planned. A behaviour question produces a count that can contradict you, which is the only kind worth the hours.

Internet-based research, government data and statistics, university research, and a literature search.

To support or validate primary research: it supplies population data and standards you cannot measure, and it shows whether your own finding is normal or unusual.

Who wrote it and what do they gain? When was it produced? Where did their numbers come from, and from how many people? Who checked it before publication?

Nobody can check who produced it or how, so the figure cannot be defended. Citing also exposes the case where several of your sources are repeating one original.

Findings about users' experience, motivations and interactions — written up as a persona for each user group, plus a journey with pain points.

It is a problem the user found serious enough to solve themselves, so it shows both the problem and the kind of solution that is acceptable to them.

Name the groups, choose a primary and say why, then accommodate the others where it costs the primary nothing — resolving the conflict with a feature rather than averaging it.

They explain behaviour that otherwise looks irrational. A patient unwilling to be a nuisance will not press a call button, which rules out solutions that depend on asking for help.

The whole experience of carrying out a task, drawn as a storyboard of frames in order — starting before the user touches the product and ending after they leave it.

Marking how the task is going at each frame, so the troughs become visible. A trough you can point at is much harder to argue with than a sentence.

A frame where the journey goes wrong — the user hesitates, does extra work, feels unsafe or gives up. Each one must be stated as an observed action and effect.

A design opportunity: something specific enough to sketch and then test. A storyboard with troughs and no opportunities has diagnosed a problem and left it there.

Function — what it is for and what each part contributes. Performance — how well it does it, in numbers. Features — what it has, who each one serves, and which are never used.

It is the only one that produces baseline numbers. Without them, "better than the existing product" cannot be claimed or tested.

Either it is not needed — so it can go, freeing cost and weight — or it is not understood, which is a usability failure. Both are findings.

State the problem abstractly, find any product that solves it, and borrow the principle rather than the shape — then name the constraint that stops you copying it outright.

WHO has the problem (a recruitable group), what they NEED A WAY TO do (an outcome, never a product), and BECAUSE — the research insight with its evidence.

Because every idea generated afterwards becomes a variation of that one solution, so ideation has been decided before any analysis happened.

Could someone else recruit your users from it? Does it name an outcome rather than a product? Is the BECAUSE something you found out? Would three very different ideas all satisfy it?

It is what the final evaluation is judged against — whether the stated need is now met. A statement too vague to answer that was never specific enough.

A finding behind it, a statement of what the solution must DO, a number or condition, a test, and a band — essential or desirable.

Essential: the solution fails without it, so it filters ideas out. Desirable: a real improvement taken if it costs nothing else, used to break ties between ideas that all pass.

"Made from aluminium" writes a solution into the specification and rules out better ones. The criterion is the behaviour aluminium was wanted for.

To compare ideas against each other, to decide when development is finished, and to structure the final evaluation — which is why a vague one costs marks three stages later.

Isometric shows the whole object in three dimensions to communicate a concept. Orthographic gives front, plan and side in true size so a manufacturer can work to the dimensions.

Dashed is a hidden edge — real, but not visible from that direction. Chain-dotted is a centre line, marking the axis of a hole or cylinder.

Third angle places each view on the side you looked from; first angle places it on the opposite side. The symbol on the sheet says which, because reading it wrongly produces a mirror-image part.

When the question is about the ORDER of assembly — which part goes on first and which fixing goes where. An assembly drawing shows the fitted result but not the sequence.

Anything the body judges: weight, reach, grip, balance, fatigue, and fit in a real space with real obstructions.

Scale — full size wherever a body is involved. Shape — the form in the hand, weighted to the real mass. Space — whether it fits in the real place it must be used.

An aesthetic prototype tests how it looks and feels and need not work. A functional prototype tests whether it does the job and may look like nothing. One prototype doing both usually does neither.

Unweighted, it flatters every design — with no mass there is no wrist torque, so every grip shape feels comfortable and the test cannot tell them apart.

Surface knows only the skin. Solid knows the whole enclosed volume, so it can report mass and sections. Virtual knows geometry plus behaviour — motion, loads and materials.

It has no inside — only a skin with no thickness — so there is nothing for a density to act on.

The model is built from editable dimensions, so changing one updates every dependent feature. A variant costs minutes rather than a rebuild, which is what makes iterating affordable.

It cannot tell you what a product feels like to use — weight, balance, grip, fatigue or fit in a cluttered real space. Those need a weighted physical model.

A material, so stiffness and strength are known; the loads — how much, where and in what direction; and the constraints saying which faces are held still.

No. Red is the highest stress in that simulation. Failure depends on whether the peak exceeds the yield strength of the material, which the numbered scale tells you.

The load has to change direction through a small area. A generous fillet radius lets the stress flow round gradually, which is usually the cheapest fix available.

Anything outside the case described: being dropped, loaded off-centre, corroded, or fatigued over many cycles. It also assumes a perfect material, which a printed part is not.

Watertight solid, wall thickness against the nozzle or beam width, orientation, overhangs and reachable supports, and designed clearances at mating faces.

It fixes which faces print well, which surfaces overhang and need support, and the direction the part is weak in — so every other decision depends on it.

Typically 0.2 to 0.4 mm per mating face. Modelled at the exact nominal size, printed parts come off the bed fused together.

A printed part is not the production part: FDM is far weaker across its layers than a moulded part is in any direction, and SLA embrittles in sunlight.

They answer different questions — users about the body and the task, clients about budget, installation and what the organisation owns — and in one room the client talks while the users agree.

Decide the measurement before the session, give a task rather than an opinion prompt, say nothing and do not help, record behaviour as well as words, and ask the open question only at the end.

Pair it with what was observed, work out what it means about the design, then decide the change — and retest with the same task, users and measurement.

When the behaviour does not support the words: three users calling a tool heavy while nobody slows down is worth recording and watching, not redesigning for.

At the requirement, not the material. Each requirement names a property with a number, which a candidate either meets or does not.

As a filter: a candidate failing one is out however well it scores elsewhere. Mass, cost, availability and aesthetics then rank whatever survives.

Stiffness. Strength is the load before it breaks; stiffness is how little it bends before then. Section depth and shape affect stiffness at least as much as the material does.

The requirement, the property it demands, and the consequence for the user. A material name on its own earns nothing.

Texture, form and colour — each enhanced by a finishing technique after the material has been chosen on its properties.

Grip under a wet hand, hiding fingerprints, or preventing dirt being trapped — and it tells a user where to hold something with no instruction.

Anodising converts the surface of the metal itself so there is no coating to chip. Paint sits on top, and once chipped in a wet environment corrosion starts under its edge.

The colour goes right through, so a scratch shows the same colour beneath instead of a different substrate — the difference between looking worn and looking damaged.

Cost per part rather than per kilogram, including processing, waste and finishing — and over the whole product life, since a coating renewed twice is bought three times.

Available in the section and grade required, within the lead time, at a usable order quantity, and ideally from a second source so one supplier cannot stop production.

A product lasting fifteen years in a slightly worse material beats one lasting four in a better one, because the second is made and transported nearly four times over.

Essentials first — cost never buys back a failed property. Then whole-life cost, then what is genuinely available. Finally, state the trade-off you accepted and why.

The requirement from the specification, the property it demands, the evidence for that property, and the trade-off accepted. Missing the evidence link makes it a preference.

Secondary is published — property tables, standards, corrosion charts, supplier data sheets. Primary is your own testing on this material in this environment.

The supplier gains if you believe it and nobody independent checked it. It is evidence of what the material is offered as; a second source makes it a number you can defend.

Every choice gives something up. An answer listing only advantages reads as advocacy; naming the cost and arguing it was worth paying reads as judgement.

Simplify it to members and joints, add the real loads, follow each load to the ground, name what each member does, and say where it would fail first.

Triangulation for a mechanism; depth for bending, since stiffness rises with the cube of depth; a shorter span for deflection, which falls with the fourth power of span.

Strength and stiffness are different properties, and a stronger grade usually has almost the same stiffness. Geometry changes deflection enormously; material hardly at all.

At a joint, a fixing, or the most slender compression member — rarely in the middle of a solid part.

σ = F ÷ A using the original area; ε = ΔL ÷ L using the original length; E = σ ÷ ε taken on the straight part of the graph.

It is a length divided by a length, so the units cancel. A 200 mm bar stretched 0.4 mm has a strain of 0.002.

From the straight part only, before the yield point. Past yield the line curves and a gradient taken there is not the modulus.

Steel about 200 GPa, aluminium about 70 GPa, timber 10 to 15 GPa, polymers 1 to 3 GPa.

Overloading, material choice, size and shape — usually more than one at once, and size and shape are the cheapest to fix.

Stress concentrates there, and the sharper the corner the higher the local stress. A fillet radius spreads it, which makes a radius a structural feature.

The highest stress in that model, not necessarily a failure. Compare the value on the scale with the material's yield strength before concluding anything.

The loads, constraints and material data were all assumed. Analyse the wrong load case and the plot is precise, colourful and useless.

The forces balance — up equals down — and the moments about any point balance. Both must hold.

Take moments about one support so its reaction drops out, solve for the other, then use up-equals-down to find the first.

Arrows pointing away from each other along a member mean tension; arrows pointing towards each other mean compression.

Buckling, not just crushing. A slender member goes unstable sideways well below its crushing strength, and effective length decides it.

SF = ultimate ÷ allowable. Rearranged: allowable = ultimate ÷ SF, and required ultimate = allowable × SF.

The heaviest plausible user, plus anything carried, plus the dynamic peak from jumping or swinging, plus realistic misuse — never the average user.

Required ultimate load = working load × SF. Required area = that load ÷ the material's ultimate stress. Then round up to a standard stocked size.

It costs material, mass, money and embodied impact, can make the product worse to use, and can hide sloppy analysis. Reducing uncertainty by testing is often better.

Lever: effort arm ÷ load arm. Gears: driven teeth ÷ driver teeth. Belt: driven diameter ÷ driver diameter. Pulley block: the number of rope falls supporting the moving block.

The ratios multiply. Stages of 15 and 6 give an overall MA of 90, not 21.

When speed or reach matters more than force — tweezers, a fishing rod, a broom, the human forearm.

The calculation is pure geometry; friction at pivots, bearings and gear teeth takes a share of the effort before it reaches the load.

The distance the effort moves ÷ the distance the load moves, or input speed ÷ output speed for a rotating drive. It comes from geometry alone.

One mesh at a time. Gears sharing a shaft turn at the same speed, so each stage output is the next stage input. The overall ratio is the stage ratios multiplied.

Nothing — its tooth count cancels. It reverses the output direction and bridges a gap; the ratio is set by the first and last gears only.

VR is a ratio of distances or speeds, fixed by geometry. MA is a ratio of forces, reduced by friction. In a perfect machine they would be equal.

Efficiency = MA ÷ VR × 100%, or useful energy out ÷ energy in. VR is what the geometry promised; MA is what was delivered.

Gear teeth roll against each other; a worm thread slides across the wheel teeth, and sliding friction is far larger. A worm can be at 50% where a spur pair is at 97%.

It multiplies. Three stages at 95% deliver about 86%, so removing a stage is often worth more than improving two.

In a screw jack and a worm hoist, where friction stops the load running back when the operator lets go. Making them efficient would need a separate brake.

Input speed ÷ required output gives the overall ratio; split it into stages of about 5:1 or less; choose tooth counts that multiply to it; then check direction and physical size.

A single 30:1 pair needs a gear thirty times the driver's diameter. Two stages of 5.5:1 multiply to the same ratio inside a hand-sized housing.

Use an even number of external meshes, or add an idler. The idler restores the direction and leaves the ratio unchanged.

The same pair of teeth then meet on every revolution, so any imperfection wears a pattern into that pair. A ratio like 5.5:1 spreads the contact.

Largest radius minus smallest radius. That is how far the follower rises, and it sets the stroke of whatever the follower drives.

A stretch of constant radius where the follower holds still while the shaft turns. It is how a valve stays open for part of a cycle and how a feed waits for a needle to clear.

Backwards: draw the displacement graph of follower height against shaft angle, then wrap it round the shaft so each height becomes a radius at that angle.

A knife edge traces fine detail and wears fast; a roller wears well and is usual; a flat follower cannot enter a concave curve, so the profile must suit it.

Effort × effort arm = load × load arm, with both arms measured from the fulcrum.

A longer effort arm must give a smaller effort. On a first- or second-class lever the effort should be less than the load.

It shortens the load arm and lengthens the effort arm at the same time, so the ratio improves twice over.

The user. A large MA needs a long effort arm and therefore a long hand movement, and handles that must open beyond an arm's reach are useless.

The power supply, the input devices, any signal conditioning, the processing stage, and the output drivers — plus whether anything measures the output.

The largest integrated circuit, usually with a crystal beside it and a programming header nearby. Everything else feeds it or is driven by it.

Thin tracks carry signals and thick tracks carry power, so the widest tracks run from the supply to the output devices.

Saying what each part contributes and which stage it belongs to, and finishing with whether anything measures the output.

A motor, gearbox, position sensor and control circuit in one unit that moves to a commanded angle and holds it. Use one wherever a precise position matters.

When the user needs specific information — a temperature, a countdown, a fault code. A light can say something is wrong and never what.

It carries far larger currents and keeps the two circuits electrically separate, which is what lets a 5 V controller switch a 230 V load safely.

The current. A chip pin supplies a few tens of milliamps — enough for an LED with a resistor and nothing else.

Bluetooth about 10 m at very low power; Wi-Fi tens of metres at moderate to high power; 5G kilometres at high power. Reaching further always costs energy.

Bluetooth needs only the other device. Wi-Fi needs a router and a network. Cellular needs a network operator, a subscription and a SIM.

Is it battery-powered and must it last? Is it mains-powered inside a building with a network? Does it move, or is there no local network to join?

Keep the core function working without the network — a doorbell should still ring from its button if the server goes away.

Voltage across the component with the circuit running; current in line with the circuit broken; resistance with the power off and one end disconnected.

It has a very low resistance, so it becomes a short circuit. That is why the current socket is separate and fused on almost every meter.

Frequency, waveform shape, pulse width and duty cycle, and short glitches. A steady 2.5 V and a 0-5 V square wave read identically on a meter.

Count divisions and multiply by the setting: four divisions at 2 V/div is 8 V; a cycle across five divisions at 1 ms/div is 5 ms, which is 200 Hz.

V = I × R, so I = V ÷ R and R = V ÷ I. P = V × I, so V = P ÷ I and I = P ÷ V.

Converting to base units before substituting — milliamps into amps, kilohms into ohms.

Total charge used per day in mAh, then divide the cell rating by it. A 2,000 mAh cell supplies 20 mA for 100 hours.

It decides the physical size of the part. A quarter-watt resistor dissipating half a watt runs hot, drifts in value and eventually fails.

Resistors: R = R₁ + R₂ in series, 1/R = 1/R₁ + 1/R₂ in parallel. Capacitors: C = C₁ + C₂ in parallel, 1/C = 1/C₁ + 1/C₂ in series.

It must be smaller than the smallest resistor in the combination. An answer larger than both means the wrong rule was used.

V_out = V_supply × R₂ ÷ (R₁ + R₂), with R₂ the lower resistor. 9 V across 10 kΩ and 5 kΩ gives 3 V.

Much smaller wastes current and heats the board; much larger makes the junction a high impedance that the next stage loads and that picks up interference.

A rounded box for start and stop, a parallelogram for input and output, a rectangle for a process, and a diamond for a decision.

Exactly two, both labelled YES and NO. Three outcomes need two diamonds in sequence, not three lines out of one.

Otherwise it reads its sensor once and stops, leaving the output in whatever state it was in. The loop is what makes it monitoring.

A gap between the switch-on and switch-off thresholds, drawn as two decision diamonds with different values — 25 °C to open and 22 °C to close.

The supply rails — positive at the top, 0 V at the bottom. Every branch then hangs between them, input on the left and output on the right.

The wires are connected. Crossing without a dot means no connection.

Without it the base takes whatever current the previous stage can supply, risking the transistor and stopping a divider working as a divider.

Coil in the low-voltage circuit, contacts as a separate switch in the load circuit, joined only by a dashed line — which shows the two circuits are isolated.

Accelerometer — motion and orientation. Ultrasonic — distance by echo timing. Photoresistor — light. Hygrometer — humidity and air temperature.

It measures distance without contact, works in the dark, and is unaffected by the colour of the obstacle.

Whether its output is analogue or digital, whether it needs a divider, whether it needs amplifying, and exactly where it physically sits.

A sensor in the wrong place reads the wrong thing perfectly. A thermostat above a radiator switches the heating off while the room is cold.

A 100 nF capacitor beside the supply pins supplies the brief surge when an output switches, so the rail does not dip and reset the chip.

A switch alone connects the pin to 0 V when pressed and to nothing when released. The pull-up gives it a defined level the rest of the time.

Reading a mechanical switch twice a few milliseconds apart, because it bounces on contact and one press would otherwise look like several.

A motor draws several times its running current at start-up. Sharing a rail with the chip drags the supply down and resets it as the motor switches on.

Two to the power of the number of inputs: four for two inputs, eight for three, sixteen for four.

Combinational output depends only on the inputs now. Sequential output also depends on the previous state, so it needs a latch, flip-flop or counter.

Eight plus zero plus two plus one is eleven. Each place is twice the one to its right: 8, 4, 2, 1.

When it must remember something — a latching alarm that stays on after the intruder has gone, a counter, or a response that depends on what happened before.

Craft, mechanised, automated, assembly line, hybrid, and computer integrated manufacturing — in order from person to machine.

In a mechanised system a person operates the machine for every part. In an automated one the machine runs itself and a person supervises.

Ask how many are made, how identical they are, and how much judgement each needs. Many identical units with no judgement means an assembly line.

Real products have repetitive stages and judgement stages. Machines take the repetition; people take inspection, awkward assembly and final set-up.

It buys consistent quality, continuous running and a collapsing labour cost per part. It costs high capital, slow expensive changes, and bad parts made in bulk before a fault is noticed.

Repetitive demotivating work, and the whole line stopping if any one station does. It is also inflexible once laid out.

Machines take the repetitive work and people take the judgement, so it copes with variation full automation cannot. The cost is two systems to balance.

It links design, production, stock and quality into one system, so a design change flows straight to the machines and every part is traceable — and the plant depends entirely on that software.

One-off, batch, mass, mass customisation and continuous.

Tooling is a fixed cost divided by the number of parts. A £20,000 tool adds £2,000 to each of ten parts and 2p to each of a million.

The set-up between batches: tools changed, machines reset, first parts scrapped. Fifty made ten times costs far more per unit than 500 made once.

Restarting a float glass line or a chemical plant costs more than running it, because the process must be brought back to temperature and the first output is scrap.

The type of part, the material, the scale of production, the production system available, cost constraints, and environmental considerations.

Scale, because it decides whether the tooling can be paid for — though a shape or a material the process cannot handle rules it out whatever the volume.

Start with a cheaper process at lower volume, prove the market, then tool up — and write down the volume that triggers the change.

A factory uses what it has and what its people know, so it may machine a part a moulder would mould, or buy the moulded part in.

Its material, the process that made it, how it joins to the next part, and why each was chosen.

It is the best single predictor of assembly time and cost. Fewer parts means fewer stations, less stock, fewer mistakes and usually lower impact.

Grouping work so each station takes about the same time. One station at 90 seconds among others at 30 sets the whole line to 90 seconds per unit.

Where the fault is cheapest to fix. Testing a board before it is sealed inside an enclosure costs minutes; afterwards it costs the whole product.

Draft angles so the part releases, uniform wall thickness so it cools without sinking, radii instead of sharp internal corners, and no undercuts.

The line where the two halves of the tool meet, which cannot be avoided. A designer chooses where it runs so it looks deliberate rather than like a fault.

Parts are designed so they fit only one way round and the product assembles from one direction — which is why so many products have all their screws on one face.

Colour changes mean a machine changeover and wasted material, so colours are run in batches from a limited range.

To the client who pays, to the community that lives with the product, and to the environment that supplies and absorbs it. Almost every decision reaches all three.

Because a per-unit improvement can be an overall loss. A bottle using 18% less plastic but replaced twice as often uses more material across a year.

Welding an earcup shut is chosen for cost and thinness, and it decides that any fault scraps the whole product. Most environmental damage is a side effect of a decision about something else.

Four questions asked before a decision is fixed: who else lives with this, for how long, what would change it, and can I say why.

A hazard, a test and a limit — so the same product either passes or does not, and two people can agree about the result.

Design the hazard out; if it must exist, guard or interlock it; warn only when neither is possible. Relying on the user is never a safety measure.

Every clause exists because something went wrong somewhere, usually to somebody. No individual designer could discover all of those hazards alone.

That the product is safe for your users in your context. A standard written for adults says nothing about a nursery, and one tested dry says nothing about wet hands.

Planned, functional, technological, style and social. What separates them is what actually ended the product.

Functional means it genuinely broke and cannot economically be repaired. Planned means the useful life was deliberately limited by a design decision.

Measuring a product against people, profit and planet rather than profit alone. Obsolescence is where the three conflict most sharply.

A long-lived product builds the reputation the next one is sold on, avoids total warranty replacements, and avoids legal exposure as repairability rules spread.

Designing one product that works for as many people as possible regardless of age or ability, without a separate special version — covering physical, sensory and cognitive differences.

"Who does this exclude, and what would let them in?" It produces a specific barrier and a specific design change instead of an opinion.

One channel always excludes somebody: a beep fails deaf users and noisy places, a colour change fails colour-blind users and bright sun. Sound plus sight covers both.

One arm; a broken wrist for six weeks; holding a toddler. All three need one-handed use, and only the first appears in a disability statistic.

The middle value of one measurement across a population: half of people measure less and half more on that dimension. It says nothing about any other dimension.

Nobody is at the 50th percentile in several dimensions at once, and combining a few such requirements leaves a band far narrower than half the population.

Where comfort rather than access is at stake and adjustment is genuinely impossible — a fixed seat pitch, say. Even then, name who is disadvantaged and what alternative exists.

State who the product does not serve and why, bound the exclusion with real data, and provide an alternative where one exists — rather than claiming it suits everybody.

Setting each requirement from the user at the limit rather than the average: reach, force and readability from the least capable user, clearance and capacity from the largest.

Ask whether it lets somebody IN or lets something THROUGH. Access requirements come from the least capable user; space and capacity requirements from the largest.

Nobody is designed out, because the requirement is met across the whole range; and one product serves everybody, so there is no special version to design and stock, and no stigma.

Dropped kerbs, designed for wheelchairs and used by every pram and suitcase; lever door handles, an arthritis solution now standard; subtitles, for deaf viewers and watched by about a third of viewers.

Physio — how it feels to the body. Socio — what it does between people. Psycho — how it feels to think with. Ideo — whether it matches what the user believes.

A design decision behind it. "The dial is weighted and damped so it turns with resistance" is physio-pleasure; "it feels premium" is an adjective.

It is agreement with the user's values, so a recycled material is a real pleasure to one buyer and irrelevant to another.

Not every product should chase all four. A smoke alarm has almost no socio-pleasure and should not try, and saying so shows the framework is being used.

Attract — desirability. Converse — usability. Transact — usefulness. The user notices it, works out what it does, and gets what they came for.

Each stage is a filter. A product that fails to attract is never used however good it is, and one that attracts but cannot be understood is returned.

Nobody reports not having picked a product up. The company sees poor sales with good reviews and often blames the marketing instead.

ACT is the sequence; the pleasures are what is felt at each stage. Physio and ideo dominate attract, psycho dominates converse, and usefulness decides transact.