Mura, Uniformity and Optical Acceptance: Writing a Standard Both Sides Can Enforce

Mura is the defect class that causes the most arguments in display supply, because it is real, subjective and usually described in words. The way out is to convert it into numbers: a measurement grid, a definition of what counts as a defect, and limits agreed before the first shipment rather than after the first rejection.

Part of the Optical, Backlight & Panel Technology guide — 30 articles on bonding, cover glass, brightness and panel choice.

What mura is and why it matters

It is a local luminance or colour deviation.

Mura describes regions of a display where luminance or colour differs from the surrounding area in a way the eye perceives as a blotch, band or cloud. It can originate in the cell (cell gap variation, spacer issues), the backlight (light guide structure, LED placement, diffusion non-uniformity) or the optical stack (adhesive thickness variation, bonding voids, cover glass distortion). Because it is a low-contrast, large-area defect, it is often invisible in normal content and obvious on a uniform grey field, which is why acceptance testing uses flat fields.

The consequence of not specifying it is that one party’s “acceptable” and the other’s “reject” are both sincere and neither can be demonstrated.

Common causes in each layer

Source determines whether it can be fixed.

Mura sources by layer

Layer	Typical cause	Fixability
Cell	Cell gap variation, spacer distribution, driver uniformity	Panel supplier issue; usually not reworkable
Backlight	Light guide structure, LED spacing, reflector and diffuser condition	Often addressable by changing the backlight design
Bonding	Adhesive thickness variation, trapped particles, voids	Process control and inspection; sometimes reworkable
Cover glass	Print step, flatness, coating non-uniformity	Drawing-level fix
Mounting	Uneven mechanical load causing stress birefringence	Mechanical design fix

Identifying which layer produces a given defect is the first step in fixing it, and it is also what determines whether the buyer pays. A defect that appears only after the module is mounted is a mechanical design issue, not a panel issue.

Detection: visual, photographic and measured

Three methods, increasing repeatability.

Visual inspection under defined illumination catches what a customer will see and varies between inspectors. Photographic inspection with fixed exposure and distance produces a repeatable record and allows comparison over time. Instrumented measurement with a luminance meter or imaging colorimeter produces numbers that can be compared with limits. The practical approach is to use the instrumented method to define the limits and the photographic method to enforce them, with visual inspection reserved for gross defects.

Whichever method is used, the conditions must be stated: viewing distance, illumination, background, whether the display is powered, and the test pattern shown. CDTech’s notes on measuring brightness, contrast and uniformity describe the measurement approach.

Defining a measurement grid and limits

Nine points is the usual starting point.

A standard measurement grid – typically 3×3, 5×5 or 9×9 points depending on the display size and the form of the defect – gives a set of luminance values from which uniformity can be computed as a ratio of minimum to maximum or as a percentage deviation from the mean. Limits are then expressed as a maximum deviation and, where relevant, a maximum local gradient between adjacent points, because a gradual variation is less objectionable than a step.

The grid should be large enough to catch the defects that matter. A 3×3 grid will not find a 10 mm blot, so where small mura is objectionable the grid must be finer or the inspection must be imaging-based.

Writing the acceptance standard

Make every criterion measurable and observable.

• Test patterns: uniform white, uniform grey at defined levels, uniform black, and the content used in service.
• Measurement grid: number and position of points, and the instrument’s geometry.
• Limits: maximum deviation from mean, maximum adjacent-point gradient, and any absolute minimum.
• Edge zones: separate, usually looser, limits near the border.
• Defect size threshold: the smallest defect that is counted, expressed in millimetres on the display.
• Inspection conditions: distance, illumination, and whether the display is viewed powered or unpowered.

The limit that causes most disputes is the smallest counted defect. Setting it explicitly, with a boundary sample, converts a matter of opinion into a measurement.

Relating uniformity to the application

The acceptable limit depends on what the display shows.

A shelf-edge price display showing large numerals tolerates far more non-uniformity than a medical imaging display showing subtle greyscale differences. Where the application is information-dense, the limit has to be tighter; where it is graphic, looser limits reduce cost without affecting the user. The useful specification states the limit in terms of the application’s requirement – for example, the maximum deviation that still allows the smallest character to be read – rather than in terms of an industry convention.

For imaging applications the DICOM greyscale framework provides a defined relationship between luminance and displayed value, and the uniformity requirement supports it; the standard is published through the DICOM standard pages.

Incoming inspection and long-term consistency

The standard only works if it is applied the same way every time.

Incoming inspection should use the agreed pattern, grid and conditions, with the results recorded per lot so that a drift can be seen before it becomes a rejection. Where the supplier also inspects, both sets of results should be comparable – which requires that the fixture, the instrument and the illumination are specified rather than left to each site. Disagreements about mura almost always trace back to two sites measuring differently rather than to a genuine difference of opinion.

Keeping a reference unit with a known appearance, and re-measuring it periodically, is a cheap way to detect a shift in the measurement setup.

Handling disputed defects

A defined process avoids escalation.

The workable process is: measure the disputed unit with the agreed method; compare the result with the limit; if there is a difference between the two parties’ measurements, compare the setups; and where the defect is at the limit, refer to the boundary sample. Where the dispute concerns a defect that appears only after mounting or after thermal cycling, the question becomes whether the assembly or the component is responsible, which is resolved by testing the component alone under the same conditions.

Recording the outcome of each dispute, and updating the standard where it was ambiguous, improves the standard over time rather than leaving the same argument to recur.

Where the specification should be recorded

Three documents carry the requirement.

1. The purchase specification, stating the uniformity limits and the measurement method.
2. The inspection procedure, stating the fixture, pattern, grid and illumination.
3. The boundary samples, signed by both parties and held by each.

With those three in place, an optical acceptance discussion becomes a comparison rather than an argument, and new personnel can apply the standard without relearning it by trial. Adhesive behaviour that affects uniformity is documented by material suppliers such as Henkel and DELO, and haze measurement by ASTM D1003 where a surface contribution is involved.