E-Ink Display Maths: Particles, Pixels & Waveforms

Why an E-Ink Screen Is a Mathematics Lesson

Why is mathematics important in an electronic-paper display? An E-Ink screen does not simply “hold ink” behind glass. It controls charged pigment particles with electric fields, addresses a rectangular grid, schedules voltage waveforms through time, and assigns finite display states to tones or colours. Algebra, geometry, ratios, probability, electricity and optimisation all meet in the quiet page of an e-reader.

E Ink describes its electronic ink film as millions of tiny microcapsules containing charged particles in a clear fluid. Applied electric fields move differently charged pigments towards or away from the viewing surface. The company also explains that its displays are reflective: ambient light reaches the pigments and is reflected to the reader, rather than a backlight shining through a transmissive panel. These are useful physical facts, but mathematics is what lets us predict pixels, image size, update choices and energy consequences.

The most important idea is state. A display pixel has a location, a previous optical state, a desired state and a permitted sequence of electrical commands. A black-to-white transition may need a different drive history from a grey-to-grey transition. That is why refresh behaviour is not captured by one number such as “frames per second.” The controller must manage space, time and memory together.

Quick Reading Routes

Students: begin with pixel grids, aspect ratio and grayscale levels.
Parents: read the sections on reading conditions, ghosting and specification claims.
Teachers: use the image-size, thresholding and energy-accounting investigations.
Career explorers: connect the topic to materials science, electronics, signal processing, typography, accessibility and embedded systems.

The activities below use screenshots, printed grids and approved display settings. They do not require opening a device or touching high-voltage display circuitry.

Reflective Displays Use Light Differently

Reflection is not emission

An emissive display produces light at or near its pixels. A reflective electronic-paper display relies mainly on external light reflecting from its surface. Some devices add a front light that shines across the display layer, but the optical path is still different from a conventional backlit liquid-crystal display.

This distinction matters because brightness is not just a slider value. Perceived readability depends on illumination, surface reflectance, contrast, viewing angle, font geometry and the reader. A reflective page may look clearer in strong ambient light, while a dark room may require a front light.

Reflectance and contrast are ratios

Suppose a white region reflects 40 units of measured light and a black region reflects 5 units under the same geometry. A simple contrast ratio is 40:5, or 8:1. If glare raises the black measurement to 10 units while the white remains 40, the measured ratio becomes 4:1. The pixels did not necessarily change; the viewing environment did.

A ratio needs a stated method. Instruments, angles, spectra and ambient light can change the result. Comparing numbers from unrelated tests as if they were identical is poor measurement practice.

Front-light uniformity is spatial data

Divide a page into a 3-by-3 grid and record brightness at each cell. If the mean is 100 arbitrary units but corners measure 84 and the centre 112, the variation can be described by deviations from the mean or by a range of 28 units. A single centre reading would hide this pattern.

This is a useful example of why maps can be more informative than averages. The location of a difference matters to reading comfort.

Charged Particles Turn Voltage into Appearance

Electric field links voltage and distance

In a simplified parallel region, electric-field magnitude can be represented as E = V/d, where V is potential difference and d is separation. A charged particle experiences force F = qE. These equations are idealised, yet they show why voltage polarity, magnitude, electrode geometry and particle charge matter.

If the field reverses direction, the force on a given charged pigment reverses. In a two-pigment system, opposite charges can move in opposite directions. Bringing light pigment towards the viewing surface makes a region appear lighter; bringing dark pigment forward makes it darker.

Real electrophoretic films contain fluids, particle interactions, boundaries and material-specific behaviour. Students should use the simple equations as models, not as complete manufacturing descriptions.

Motion takes time

Particles do not teleport. Their response depends on field, viscosity, temperature, travel distance and previous arrangement. A controller applies a time sequence of voltages—a waveform—to drive the transition reliably.

We can represent one pixel's drive as V(t), a voltage that changes with time. The visible state after the sequence depends on more than the final voltage. Duration, polarity order and starting state may all matter. This is a physical example of a system with memory.

Bistability changes the energy story

E Ink describes its ink as bistable: once particles have been moved to a stable visible state, the image can remain without continuous power to maintain that state. Energy is still required by the controller, memory, wireless system, lighting and page changes. “Uses power only when changing pages” is therefore an oversimplification when applied to a complete device.

A better energy model is:

total energy = update energy + lighting energy + processing energy + communication energy + standby losses.

The importance of each term depends on use. A static timetable with rare updates differs from an e-reader using a bright front light and frequent wireless synchronisation.

Pixel Grids Turn a Page into Coordinates

Every pixel has an address

A raster display can be modelled as a matrix. A 1,872-by-1,404 panel has 1,872 columns and 1,404 rows. Its total pixel count is 1,872 × 1,404 = 2,628,288 pixels.

Coordinates might begin at the top left. A point (x,y) then identifies a column and row. Software uses rectangles, masks and transformations to decide which locations belong to text, images, margins and controls.

Aspect ratio describes shape

The ratio 1,872:1,404 simplifies by dividing both numbers by 468, giving 4:3. This says the pixel rectangle is four units wide for every three units high. If pixels are approximately square, the physical screen has the same shape ratio.

An image with another aspect ratio must be fitted. It can be letterboxed, cropped, stretched or reflowed. Each choice preserves some properties and changes others.

Pixel density connects digital and physical size

Suppose the 1,872-by-1,404 panel has a 7.8-inch diagonal. Its pixel diagonal is √(1,872² + 1,404²) = 2,340 pixels. Pixel density is therefore 2,340 ÷ 7.8 = 300 pixels per inch.

The approximate pixel pitch is 25.4 millimetres ÷ 300 ≈ 0.0847 millimetres per pixel. A 24-pixel-high glyph occupies about 2.03 millimetres before font metrics and scaling are considered.

“Resolution” can mean two things

People use resolution for total pixel dimensions and for pixel density. “1,872 by 1,404” describes a grid. “300 pixels per inch” describes density. They answer different questions, so a careful comparison states which meaning is intended.

Grayscale Is Quantisation

Finite states approximate a continuum

If a pixel supports 16 grey levels, each pixel can represent one of 16 indexed optical states. Four bits can encode 2⁴ = 16 possibilities. The numeric code is not automatically proportional to perceived brightness; a tone curve may be needed.

Quantisation replaces a continuous target with one of a finite set of levels. The rounding difference is quantisation error. More levels can make gradients smoother, but only if the hardware, waveform, image processing and viewing conditions support them.

A simple mapping example

Suppose an 8-bit source image uses values from 0 to 255 and must be mapped to 16 evenly spaced codes from 0 to 15. A simple code is round(15x/255). A source value of 170 maps to round(10) = 10. Reconstructing with 255(10/15) gives 170, while a source value of 164 would also map near code 10 and lose some distinction.

This is not necessarily the device's actual mapping. It is a transparent classroom model.

Dithering trades local accuracy for spatial pattern

When the requested tone lies between available levels, nearby pixels can be assigned different states so that their average appearance approximates the target. A checker pattern of black and white has an average code halfway between them, though it is not physically grey at each location.

Ordered dithering uses a repeating threshold matrix. Error-diffusion methods pass part of a pixel's rounding error to neighbours. The methods create different textures and costs.

Text and photographs need different decisions

Text benefits from sharp edges and stable stroke widths. Photographs benefit from gradual tone changes. One thresholding strategy rarely serves both perfectly. A reading application may therefore use different processing for glyphs, diagrams and images.

Refresh Waveforms Are Sequences, Not Single Flashes

A transition depends on where it starts

Imagine four coarse states: white, light grey, dark grey and black. Moving from white to black may use a different pulse sequence from moving from black to white. A small grey adjustment may use a shorter path. The waveform controller can consult a transition table indexed by old and new state.

With N states, there are N² ordered old-to-new pairs. Sixteen levels give 256 possible pairs before temperature and mode are considered. This does not mean 256 wholly independent stored waveforms are always used, but it shows why the control problem grows.

Full and partial refreshes make different trade-offs

A full refresh can strongly reset a region and reduce accumulated image memory, but it may take longer and cause visible flashing. A partial refresh changes less of the page and can feel faster, but residual patterns may be more visible.

The best policy depends on content. A book page can tolerate a deliberate transition. Handwriting, menus or animation demand quicker feedback. Controllers can schedule occasional cleaning refreshes after several partial changes.

Refresh rate is not the whole experience

If one update takes 120 milliseconds, the reciprocal is about 8.3 updates per second. That arithmetic does not establish smooth 8.3-frame-per-second animation. Input processing, waveform choice, region size, display response and application scheduling all contribute.

Report both update duration and what was updated. A tiny cursor region and a full page are not equivalent experiments.

E Ink's waveform work shows continuing engineering

E Ink publicly announced its Ripple waveform architecture for Spectra colour displays, describing a goal of smoother transitions and reduced apparent flashing. This is useful evidence that waveform design remains an active engineering problem. It does not imply that every E-Ink product or monochrome panel uses that architecture.

Temperature Adds a New Variable

Materials respond differently when conditions change

Particle mobility and fluid behaviour can vary with temperature. Controllers may use a temperature reading to choose compensation parameters. A waveform suitable at one temperature may not give the same optical result at another.

A simple classroom model could let update time T depend on temperature θ:

T(θ) = a + b(θ₀ − θ) for a limited cool range.

This linear equation is illustrative, not a universal E-Ink law. Real response can be nonlinear and product-specific.

Interpolation needs boundaries

If a table gives a 200-millisecond update at 10°C and 140 milliseconds at 20°C, linear interpolation estimates 170 milliseconds at 15°C. Extrapolating the same line to −20°C would be risky because it extends beyond observed conditions.

This is a valuable scientific habit: interpolation between measurements is usually safer than extrapolation far beyond them.

Image Memory and Ghosting

Residual appearance is history-dependent error

Ghosting means part of an earlier image remains faintly visible after an update. We can model the observed new image O as the desired image D plus a residual fraction α of the previous image P:

O = (1 − α)D + αP.

If α = 0.05, five per cent of the old signal remains in this simple model. Actual ghosting is spatial, state-dependent and more complex, but α gives a measurable idea.

Measure rather than rely on memory

Display a black rectangle, update to a uniform grey page, and photograph under fixed lighting and camera exposure if the device permits such testing. Compare mean brightness inside the old rectangle with a control region. Repeat after different numbers of partial refreshes.

The camera introduces its own noise and tone curve, so results are comparative rather than laboratory certification. Still, controlled repetition is better than “it looked worse yesterday.”

Cleaning has a cost

Frequent full refreshes may reduce residuals but increase visible flashing and waiting. Rare full refreshes may feel fluid but allow more residue. This is a multi-objective optimisation problem: minimise time, disturbance and error simultaneously. There may be no single setting that is best for every task.

Colour E-Paper Adds Geometry and Combinatorics

Colour can be created in different ways

Some systems use coloured pigments; others can use a colour filter array over a monochrome layer. The exact construction depends on the product. Never assume all “colour E-Ink” panels share one optical design.

A filter pattern allocates spatial elements to colour components. If groups of subpixels collaborate to make one colour sample, effective colour detail can differ from the underlying addressable grid. Product specifications should be read carefully.

Colour spaces use coordinates

A source image may store red, green and blue coordinates, but reflective pigments have a different achievable gamut. A conversion algorithm maps requested colours into available states. Out-of-gamut colours must be clipped, compressed or approximated.

Two colours with different RGB triples can map to the same display state. This is a many-to-one transformation, so information is lost. Good design uses contrast, labels and patterns rather than depending on a subtle colour distinction.

Palette size grows exponentially

If three independent channels each have four levels, the ideal mathematical combination count is 4³ = 64. Physical interactions may prevent all combinations from being equally distinct. Counting theoretical codes is not the same as measuring perceptually usable colours.

Storage, Bandwidth and Update Regions

Frame size follows a simple product

For a 1,872-by-1,404 image stored at four bits per pixel, raw storage is:

1,872 × 1,404 × 4 = 10,513,152 bits.

Dividing by eight gives 1,314,144 bytes, about 1.31 megabytes using decimal units. Compression may reduce file size, but the result depends on image content and algorithm.

Small rectangles can save work

If only a 200-by-80 status box changes, it contains 16,000 pixels—about 0.61% of the full 2,628,288-pixel grid. Updating just that bounding rectangle can reduce transferred and processed data, although alignment rules and waveform constraints may expand the actual region.

Dirty-region algorithms use set geometry

Applications track rectangles that changed. Overlapping rectangles may be merged. A merge can reduce command overhead but update some unchanged pixels. Keeping many tiny rectangles reduces extra area but increases bookkeeping.

This is a practical optimisation question: compare the area of the union, the bounding box and the number of commands.

Energy Accounting with Honest Units

Power and energy are different

Power is the rate of energy use, measured in watts. Energy is power multiplied by time, often expressed in joules or watt-hours. A 2-watt front light used for three hours consumes 6 watt-hours. A 0.5-joule page update repeated 600 times consumes 300 joules, which is 300 ÷ 3,600 ≈ 0.083 watt-hours.

These illustrative values show why usage pattern matters. They are not specifications for a particular device.

Average energy per page can mislead

If a reading session consumes 1.2 watt-hours over 240 page turns, division gives 0.005 watt-hours per page. But the energy includes screen lighting, processor activity and time between turns. Calling it “display update energy” would be wrong unless those components were separated.

Battery percentage is not a precision meter

A displayed percentage is an estimate derived from voltage, current history, temperature and battery models. A drop from 80% to 70% does not necessarily represent exactly one tenth of the original energy capacity. For comparisons, use longer repeated trials and similar conditions.

A Worked Page-Layout Example

Suppose a 1,872-by-1,404 panel uses 120-pixel left and right margins and 90-pixel top and bottom margins. The content rectangle is 1,632 pixels wide and 1,224 pixels high.

At 300 pixels per inch, its physical dimensions are 1,632 ÷ 300 = 5.44 inches by 1,224 ÷ 300 = 4.08 inches. In millimetres, that is about 138.2 by 103.6.

If a line of body text is set in a box 1,500 pixels wide and an average rendered character including spacing occupies 15 pixels, a rough capacity is 100 characters per line. Real typesetting is not monospaced; words wrap, glyph widths vary, and shaping matters. The calculation is a first estimate, not a layout guarantee.

If line height is 42 pixels, the 1,224-pixel content height fits floor(1,224 ÷ 42) = 29 full lines, leaving 6 pixels. Increasing line height to 48 reduces the count to 25 lines with 24 pixels left. Readability and page-turn frequency trade off.

Mathematics in Typography and Reading

Font size is not glyph size

Point size, pixel size, cap height, x-height and line height are different measurements. Two fonts set to the same nominal size may look different. A reading setting should be judged by actual glyph geometry and comfort, not only by a menu number.

Edge placement uses sampling

A diagonal stroke crosses the pixel grid. Rasterisation decides which pixels represent the ideal outline. Antialiasing uses intermediate states to smooth the apparent edge. Limited grayscale or a different tone response changes the result.

Reflow is a constraint problem

When the user increases text size, the layout must preserve words, margins, images, headings and navigation inside a smaller effective capacity. The software solves a sequence of line-breaking and page-breaking decisions. A reflowable book is therefore not a fixed stack of screenshots.

Accessibility needs more than enlargement

Larger text can help some readers, but contrast, spacing, line length, navigation and content structure also matter. A display cannot repair a poorly structured document. Mathematics helps quantify dimensions; the reader's experience decides whether the result is useful.

Common Misconceptions

“An E-Ink screen never uses power while showing an image”

The electrophoretic image can be bistable, but the complete device can consume power for lighting, processors, radios, memory and monitoring.

“More pixels always make reading better”

Density helps render fine detail, but typography, contrast, lighting, software scaling and eyesight also matter.

“A partial refresh is simply a faster full refresh”

It may update different regions or use different waveforms, with a different balance of speed and residual error.

“Sixteen grey levels mean every photograph has only sixteen colours total”

Each pixel may choose among sixteen states, and spatial dithering can create many patterns. The underlying per-pixel state count remains finite.

“Ghosting means the panel is permanently damaged”

Residual appearance can be temporary and reduced by a cleaning refresh. Persistent faults are a different issue and should follow manufacturer support guidance.

“Reflective means readable in every light”

Reflection requires usable illumination and manageable glare. No display removes the geometry of light.

Practical Student Investigations

Measure usable page area

Take a screenshot with known pixel dimensions. Mark margins and calculate the fraction of the grid used for content. Compare portrait and landscape layouts.

Build a 16-level converter

Use a spreadsheet to map 0–255 source values to 0–15 codes with round(15x/255). Calculate the reconstructed value and error. Plot the staircase function.

Compare thresholding and dithering

Create a small greyscale gradient. Convert it once with a single threshold and once with a 2-by-2 ordered pattern. Explain which preserves average tone and which preserves clean regions.

Time controlled updates

Record several page-turn times for text pages and image-heavy pages under the same mode. Use median and range, not only the fastest trial. Do not claim the result applies to all devices.

Map front-light uniformity safely

With a camera locked to manual exposure and the device showing one blank tone, photograph from a fixed position. Compare a grid of sampled regions. State that the camera response and room reflections limit the result.

Test line length

Measure the number of characters or words per line at several text settings. Ask readers to report comfort without treating subjective preference as an objective medical result.

How Students Can Learn Progressively

Primary level: arrays and fractions

Count rows and columns, calculate total cells, shade fractions and compare page shapes.

Lower secondary: ratios and unit conversions

Use aspect ratio, pixels per inch, millimetres per pixel, percentages and energy units.

Upper secondary: functions and statistics

Study quantisation functions, interpolation, repeated timing, residual error and optimisation trade-offs.

Beyond school: fields and dynamical systems

Explore electrophoresis, material response, waveform control, colour science, typography and embedded display pipelines.

Keep the model boundary visible

Label an equation “simplified,” list what it ignores, and avoid converting a classroom fit into a manufacturer claim. This is mature mathematical modelling.

Parent and Student Guidance

Choose for the reading task

Screen size, weight, lighting, document support, navigation and comfort may matter more than a single maximum specification. Try the kinds of books, notes or diagrams actually used.

Adjust the environment

Reduce glare, use suitable ambient light, change front-light level gradually and keep a comfortable viewing distance. A brighter setting is not automatically a clearer one.

Allow for update behaviour

Occasional flashes or residual images can be part of the chosen refresh mode. Consult the device manual before assuming a fault.

Keep software and accessibility settings in view

Font, spacing, orientation, contrast mode and document structure can change usability without new hardware.

Treat comfort as real evidence

If reading causes persistent discomfort, stop and seek appropriate advice. A display specification is not a diagnosis or guarantee.

Frequently Asked Questions

What is E-Ink?

E Ink is a company and a commonly used name for its electrophoretic electronic-paper technology. Other electronic-paper approaches also exist.

Why can an image remain when power is removed?

The pigment arrangement can be bistable, so maintaining a static optical state does not require the same continuous drive as some other displays.

Why does the screen sometimes flash?

A waveform may drive pixels through intermediate states or perform a cleaning refresh to improve the final result.

What causes ghosting?

The previous particle arrangement can influence the next appearance, especially after partial updates. Temperature, waveform, content and panel condition can matter.

Is a colour E-Ink screen the same as an LCD?

No. Both use pixel grids, but their optical and electrical mechanisms differ. Colour E-Paper designs also differ from one another.

Does 300 ppi guarantee sharp text?

It describes pixel density. Font design, rasterisation, contrast, scaling and viewing distance also affect perceived sharpness.

Why are page turns slower than on some tablets?

Electrophoretic particles need controlled motion through a waveform, and the system may choose a slower transition to manage optical quality.

Which careers use this mathematics?

Display engineering, materials science, embedded software, graphic design, accessibility, publishing and product testing use related ideas. Mathematics supports these fields but does not guarantee a particular career.

Useful Next Reading

The Larger Lesson

An E-Ink page looks still, but it is the result of carefully managed change. Electric fields move charged pigments. Pixel coordinates place content. Quantisation turns continuous tones into finite states. Waveforms guide transitions. Region algorithms decide what to update, and energy accounting separates the display from the whole device.

That is why mathematics matters. It reveals that “paper-like” is not a magical property but a collection of measurable design choices. A student who asks which state, which coordinate, which time interval, which energy boundary and which viewing condition is learning how engineers turn materials into a dependable reading experience.

A Full Refresh-Scheduling Example

Imagine a note-taking page divided into a 12-by-16 grid of update tiles, giving 192 tiles. During one second of writing, the pen path changes 18 tiles. Updating the whole page would process all 192 tiles; updating the bounding rectangle around the stroke might cover 30 tiles; tracking the exact changed set would cover 18.

The exact-set method updates 90.6% fewer tiles than a full-page update because (192 − 18)/192 ≈ 0.906. The bounding rectangle updates 84.4% fewer. Yet the exact set may require more region commands and create discontinuities in waveform scheduling. The smallest changed area is not automatically the fastest complete policy.

Suppose a local partial update takes a fixed 4 milliseconds of command overhead plus 0.08 millisecond per tile. One 30-tile rectangle takes 4 + 30(0.08) = 6.4 milliseconds. If the 18 changed tiles are split across six separate regions of three tiles, total becomes 6[4 + 3(0.08)] = 25.44 milliseconds. Merging regions processes extra pixels but saves overhead.

Add ghosting to the policy

Assume a simplified residual score rises by 0.7 unit after each partial update and a cleaning refresh resets it nearly to zero. If a product team sets a 7-unit limit, the elementary model schedules cleaning after ten partial updates. But content matters: black-to-white edges may reveal residue sooner than mid-grey regions, and temperature can change behaviour.

A smarter controller can assign different costs to latency, flashing and residual error:

cost = a(time) + b(flash disturbance) + c(residual error).

The weights a, b and c are design choices. Note-taking might place a large weight on immediate response, while a finished reading page might prioritise optical cleanliness.

Why the example matters

This calculation shows that display control is not one contest for the fastest nominal refresh. It resembles delivery planning: fewer trips can cover extra distance, while many precise trips pay repeated setup costs. Mathematics makes the trade-off visible and testable.

Data Visualisation on Limited-Tone Displays

Use position before subtle shade

When only a few stable grey levels are available, charts should communicate through position, length, shape and labels before relying on gentle colour differences. A bar twice as long encodes a 2:1 ratio even if the two fills look similar.

Patterns need an appropriate scale

Hatching can distinguish categories, but very fine patterns may alias against the pixel grid or shimmer during partial updates. If diagonal lines repeat every four pixels on a 300-ppi display, their physical spacing is only about 0.339 millimetre. At ordinary viewing distance they may merge visually.

Animation is not always the right solution

A rapidly changing chart spends energy and can expose refresh artefacts. Small multiples—several static panels showing successive times—may communicate change more clearly on electronic paper. This turns time into space.

Accessibility benefits everyone

Direct labels, sufficient text size and redundant encodings help readers with colour-vision differences and also survive printing, screenshots and low-quality refresh modes. Designing for constraints can produce a more robust explanation.

Preserve meaning when the display changes mode

A chart might look different after switching from a quality refresh to a faster mode. Before release, test headings, thin lines, shaded regions and selected controls in each supported mode. If two categories merge into the same tone, add a pattern or direct label.

This is graceful degradation: the design remains understandable when colour depth, update quality or lighting is reduced. Mathematical structure—position, order and ratio—carries meaning even when the optical channel becomes weaker.

Keep a plain-text data table available too. It gives exact values, supports screen readers and remains usable if a patterned chart is hard to interpret. Visual elegance and numerical accessibility should reinforce one another.

Why Mathematics? | E-Ink Displays, Charged Particles, Pixel Grids and Refresh Waveforms