Flash Measurement System for Battery Cell and Tab Inspection

A battery cell is a difficult dimensional target for reasons that have nothing to do with magnification. The part is thin, the electrode foils are soft enough to be deformed by a stylus, the aluminium tabs are mirror-bright while the adjacent coating is matte, and the dimensions that matter sit in several tolerance classes at once: a cell outline, a tab pitch, a tab width, the position of a vent opening on a cap plate. A flash measurement system answers that combination by measuring the whole field in a single calibrated exposure. The stage does not travel during the measurement, nothing presses on the electrode stack, and every programmed feature in the image is found by software rather than by an operator moving a crosshair from edge to edge.

This page is written for the engineer who has to decide whether one-shot optical measurement fits a specific cell or module job, and for the buyer who has to read a specification without being misled by it. It starts with what the part actually asks for, runs the field arithmetic that decides fit, explains why the published accuracy and the published repeatability are not the same number, and states where a flash system stops and an inline measuring head, a confocal height sensor or a lens-level solution takes over.

What a battery cell asks an optical instrument to measure

The measurable list on a lithium-ion cell is a list of planar, edge-based dimensions. On a prismatic cell it is the outline length and width, the position and width of each electrode tab, the tab pitch, the height of the tab above the cap plane if that is fixtured as a silhouette, the spacing between the two terminals, and the location of the vent and the fill port on the cap plate. On a cylindrical cell it is the can diameter, the overall height, the terminal geometry on the cap, and the position of the weld nugget relative to the terminal centre. On a pouch cell it is the pouch outline, the seal width on each of the four sides, and the tab width and pitch on the two exit edges. On a module it is the busbar hole pitch, the bracket outline and the terminal-to-terminal spacing.

Three properties of that list shape the instrument choice. First, almost every item is an edge measured in the plane, which is exactly what a calibrated field does well; the exceptions are thickness, step height and the profile of a weld, and those are not 2D measurements at all. Second, the tolerance classes are heterogeneous. A cell outline is a millimetre-class dimension, a tab pitch is a tenth-of-a-millimetre-class dimension, and a tab width may be called out in a tolerance band an order of magnitude tighter than the outline. A single uncertainty figure quoted for the whole cell is therefore misleading in both directions, which is why the accuracy discussion further down is written per feature rather than per instrument. Third, the surface is optically inconsistent: bright aluminium or nickel tabs next to a dark matte coating, a translucent separator edge, a cap plate that may be blasted or plated.

The list the instrument does not cover matters just as much, and it is better written down before a quotation than discovered after acceptance. A standard flash system measures 2D dimensions. Cell thickness, cap height, step height and film or coating thickness are outside the 2D chain: height measurement on the IMAGE 3 series is a separate point-spectral confocal channel with a ±3 mm range, a 55 mm working distance and a 17 µm spot, standard on IMAGE 3 MAX and optional on MAX2. Anything that needs a cross-section, a weld penetration measurement or a film-stack composition is a different instrument category altogether.

Why a single exposure suits cell and tab work

Cell production is repetition. A lot contains thousands of nominally identical parts, and the same twelve dimensions are checked on each of them. That repetition is what makes a single-field instrument efficient: the measurement program is taught once on one cell, and every subsequent cell is measured by the same program in the same calibrated field, with no re-datuming, no operator judgement about which edge to pick, and no accumulated drift from an axis that travels between features. The IMAGE 3 series supports one-click measurement, automatic recognition and positioning, and multi-workpiece measurement, all of which exist precisely because the parts repeat.

The mechanical argument is stronger on a battery than on most parts. An electrode stack and a tab do not tolerate a contact probe, and a clamp that flattens a slightly domed cap plate changes the dimension being measured. A flash measurement system places no load on the part beyond its own weight, images from above and measures what the field can see. The published stage load for the IMAGE 3 series is 5 kg, which is far above the mass of any single cell or small module, so payload is never the constraint on a cell job. The constraints are field size, calibration, and the smallest feature the optics can resolve.

There is a scheduling benefit as well. A feature-by-feature instrument spends its time on motion and on re-datuming every region it visits; a single-field instrument spends its time once, on the exposure. For a quality lab that has to release a lot, or for an incoming inspection station that has to sample a shipment, the difference shows up as parts per hour rather than as micrometres, and it is usually the reason the one-shot route is chosen even when a travelling machine would be marginally more accurate on the single tightest dimension.

Does the part fit the field? Run this arithmetic first

A flash measurement system works inside one calibrated field, so the first check on any cell job is whether the features sit inside it. The IMAGE 3 series reaches a wide-field measuring range of up to 300 x 200 x 75 mm on the PRO and PRO-H configurations, model dependent, and a precision measuring range from 118 x 118 mm to 218 x 118 mm depending on configuration. Those two numbers describe two different jobs and confusing them is the most common way a battery project is specified wrongly. The wide field is the cell-level view: it captures the whole part, or several small parts, at a coarser pixel scale. The precision range is the tight-tolerance view: it captures a smaller region at a finer pixel scale, which is where tab width and tab pitch belong.

The arithmetic is simple enough to run before any quotation. An example pouch cell of 200 x 150 mm sits inside the 300 x 200 mm wide field with 100 mm and 50 mm of margin on the two axes respectively (derived), which is enough for a carrier edge and an alignment mark. An example prismatic cell of 150 x 100 mm sits inside it with 150 mm and 100 mm of margin (derived), and four such cells can be presented in one field if the fixture pitch allows it. A module plate of 420 x 180 mm does not fit in any orientation, because the 420 mm axis exceeds the 300 mm long axis (derived), so it has to be measured as a sampled set of windows or on a travelling machine. The same logic applies to the precision range: a 118 x 118 mm window covers a tab group and its cap-plate features comfortably, but it does not cover the whole cell outline at the same pixel scale.

Accuracy, repeatability and the 0.02L term

The published figures for the IMAGE 3 series are a repeatability of up to ±0.25 µm, model dependent, and a measuring accuracy from ±2 µm in the precision non-connected configuration to ±(1.5 + 0.02L) µm in the connected configuration, model dependent. The minimum display unit is 0.1 µm, which is a display resolution and not a statement that a 0.1 µm feature can be measured with a usable uncertainty; treating the display unit as an accuracy figure is the single most common misreading of this class of instrument.

The 0.02L term is the part that matters on a battery, because it grows with the measured length. At L = 100 mm the connected configuration works out to ±3.5 µm (derived), and at L = 200 mm to ±5.5 µm (derived). Those are whole-instrument budgets for a length measurement across the field, not per-edge noise. The practical consequence is a split between feature families: a tab width of a few millimetres sits at the low end of the L range and is limited mainly by edge quality and lighting, while a cell outline of 150 to 200 mm carries the full L-dependent term and should be specified with that number in hand rather than with the headline repeatability.

Repeatability and accuracy also differ by roughly one order of magnitude here: ±0.25 µm against ±2 µm is a factor of eight (derived). Repeatability describes how tightly the same edge repeats on the same part in the same position, which is what a gauge R&R study on a fixed cell will show. Accuracy describes how close that repeatable answer is to the true dimension, and it absorbs calibration, field distortion, optics and the length-dependent term. A cell supplier who qualifies an instrument on repeatability alone will pass the study and still ship parts whose tab pitch is outside the drawing, because the systematic part of the error never appears in a repeatability plot.

Two further effects belong in the budget even though they are not instrument errors. Thermal expansion scales with part size, so a large aluminium module plate measured in a cold inspection room and re-measured on a warm line will not read the same, and the difference is a property of the part rather than of the instrument. Fixturing repeatability matters in the same way: a cell that rests 0.2 mm higher on one presentation changes the working distance and, through it, the apparent magnification at the field edge. Both are controlled by procedure - soak the part, define the datum, present it the same way - rather than by buying a tighter specification.

Surfaces, fixturing and the edges that break detection

Battery parts are optically awkward in a specific way: the tab is a mirror, the coating is matte, and the separator is translucent. A mirror edge against a dark background is high contrast and easy; the same mirror edge against a bright plated cap plate is low contrast and drifts with the angle of the surface light. The IMAGE 3 series carries adjustable multi-angle surface light, coaxial light and parallel backlight, and the three are used for different features rather than as a general brightness control. Parallel backlight produces a silhouette and is the reliable choice for tab width and tab pitch, because it removes the surface reflectance from the measurement entirely. Coaxial light is for cap-plate features where the surface has to be seen rather than silhouetted. Multi-angle surface light is for the matte coating, where the edge is defined by a change in scattering rather than by a change in brightness.

Fixturing follows from that. The tab must not be clamped, because a clamp load changes both its width and its position; the cell should rest on a kinematic seat that defines the datum face, with the tab unsupported or supported from below at a known height. A translucent separator edge is the hardest feature on a pouch cell, because the edge seen in transmission is not a single line: it is a band whose apparent position moves with illumination angle. Where that matters, the drawing should be reviewed to see whether the separator edge is really a controlled dimension, or whether the seal width around it is the controlled dimension and the separator is simply present.

A final optical point concerns the cell thickness dimension itself. A prismatic cell that is 0.1 mm thicker in the middle than at the edge is still inside the wide field, but the part of it that is out of focus is not measurable at the same pixel scale. Depth of field in a single-exposure instrument is set by the optics, so a domed cap plate or a bowed pouch has to be either fixtured flat, measured with autofocus per region, or accepted as a feature that the 2D route cannot hold. The IMAGE 3 series supports auto focus, which helps on flat parts with local height differences but does not convert a 2D instrument into a 3D one.

Where a flash system stops, and what takes over

Thickness, step height and cap height belong to a different sensor. The 8055 point-spectral confocal channel on the IMAGE 3 series covers a ±3 mm range with a 55 mm working distance, a 17 µm spot and 0.02 %F.S., standard on MAX and optional on MAX2, and it turns the instrument into a 2D plus height station for cells that need both. It does not measure film thickness through a coating stack, and it does not replace a cross-section.

Line-speed inspection belongs to an inline measuring head rather than to a flash system. The HM series telecentric measurement heads cover φ15 mm, φ40 mm, φ65 mm and φ120 mm fields with a repeatability of ±0.05 µm to ±0.2 µm, and they are built to sit over a conveyor or a transfer station and measure every part rather than a sample. The division of labour is straightforward: a flash system is the offline or at-line instrument that establishes the measurement program, qualifies the lot and covers a whole cell footprint; a head is the inline gauge that holds one or two critical dimensions at production rate. The selection logic for the heads is set out in the HM series measuring head selection guide.

Two adjacent intents are worth separating explicitly. Weld inspection on a battery pack is a visual and often three-dimensional job, and the optics for it are discussed in zoom lens for EV battery weld inspection. Lens-level selection for electrode and tab measurement - which telecentric lens, at what magnification and working distance - is a different decision from which instrument to buy, and the reasoning there belongs to telecentric lens for battery electrode and tab measurement rather than to this page. What this page settles is the instrument-level question: whether a single calibrated exposure can hold the cell dimensions at all.

A specification checklist before quotation

Six items decide most battery measurement projects, and each of them can be answered before a price is quoted. Write the feature list with its drawing tolerance for each item, and separate 2D dimensions from height and thickness. Take the largest footprint on the list and check it against both the wide-field and the precision measuring range, because a part can fit the first and not the second. Decide which features need silhouette illumination and which need surface illumination, since that determines the fixture. State whether the part is presented one at a time or in a tray, because multi-workpiece measurement changes the effective cycle time. State the sampling rate the line needs, and if it is every part, compare the flash route against an inline head instead of forcing the flash route to run faster. Finally, name the surfaces that are mirror-bright and the ones that are translucent, because those two are where a technically correct specification still fails at commissioning.

The same six items are what a supplier needs in order to quote honestly, and they are also the reason two battery projects with apparently identical feature lists can need different instruments. A cell program dominated by outline dimensions and a program dominated by tab pitch look similar on paper, but the first is limited by field size and the second by illumination and edge definition. Writing them down separately, with the tolerance attached to each, removes most of the ambiguity before the first site visit. Where the list already exists in a drawing, the useful exercise is to mark which features lie inside one tab group and which span the cell outline: that split maps directly onto the precision range and the wide field, and it tells the integrator whether one program or two is needed.

The configuration that carries the described measurement chain, including the dual-magnification bi-telecentric optics, the adjustable surface and coaxial illumination and the parallel backlight, is the IMAGE 3 instant vision measuring system. Its published specification is the source of every instrument figure quoted on this page, and the per-model differences are stated there rather than assumed here.

Frequently asked questions

Can a flash measurement system measure battery cell thickness?

Not as a standard 2D configuration. The 2D measurement chain on the IMAGE 3 series covers planar dimensions such as outline, tab width, tab pitch and cap-plate feature position. Thickness and step height require the 8055 point-spectral confocal channel, which has a ±3 mm range, a 55 mm working distance and a 17 µm spot, and which is standard on IMAGE 3 MAX and optional on MAX2. A cell thickness callout should therefore be specified against that channel, or against a dedicated thickness gauge, rather than against the 2D accuracy figure.

What accuracy should I expect on a 100 mm battery part?

With the connected configuration the published formula is ±(1.5 + 0.02L) µm, which at L = 100 mm works out to ±3.5 µm (derived). That is a whole-instrument length budget, so it applies to a dimension measured across the field rather than to a short tab width measured inside a small window, where the L term is much smaller. The precision non-connected configuration is published as ±2 µm, and repeatability is published as up to ±0.25 µm, model dependent, which is a different quantity from accuracy.

How many cells can be measured in one field?

It depends on the cell footprint and on which range the features need. The wide-field measuring range reaches up to 300 x 200 mm on PRO and PRO-H configurations, model dependent, and the precision measuring range is 118 x 118 mm to 218 x 118 mm. A φ46 mm cylindrical cell fits the 118 x 118 mm precision window many times over (derived), while a 200 x 150 mm pouch cell occupies most of the wide field on its own. Multi-workpiece measurement is supported, so tray presentation is a fixture decision rather than a software limit.

When is an inline measuring head a better choice than a flash system?

When the requirement is to measure every part at line rate rather than to qualify a lot. The HM series heads cover φ15 mm, φ40 mm, φ65 mm and φ120 mm fields with repeatability from ±0.05 µm to ±0.2 µm, and they are designed to sit over a transfer station. A flash system remains the better instrument when the whole cell footprint has to be seen in one exposure, when several features across the part have to be measured together, or when the program is still being established and the part mix is not yet fixed.