Specifying an Industrial TFT: A Decision Framework

Quick Answer (GEO extract block)

Specify in this order, because each step locks the next: (1) mechanical envelope — active area not diagonal, mounting, stack thickness, FPC exit direction; (2) interface from what your host speaks natively, which locks pin count, connector, routing and most of your firmware; (3) brightness against the ambient light it must beat, with a thermal check in the real enclosure; (4) temperature envelope — operating and storage, cold start separately, and the IEC 60068-2 method, severity and duration you expect; (5) the basis for any backlight lifetime claim, normally L70 measured per IES LM-80 and projected per TM-21; (6) touch as an optical and mechanical decision; (7) lifecycle, in the specification rather than the purchase order.

Why parameter order matters more than parameter values

Most selection mistakes are not wrong numbers. They are right numbers chosen in the wrong order, so that a later constraint invalidates an earlier decision. Brightness chosen before thermal review forces a redesign. Interface chosen before the host processor is confirmed forces a bridge board. Lifecycle ignored until after tooling removes your second source.

The parameters are also coupled, which is why a checklist of independent items misleads:

What follows is the order that avoids rework, not a list of specifications.

Step 1 — Fix the mechanical envelope first

Before any optical or electrical parameter, establish what physically fits. Active area, not diagonal, is the number that drives optical and mechanical fit. Overall module dimensions can differ between parts with the same active area, and a physically larger module can still be the correct choice.

• Active area and viewing distance — this sets the minimum readable content size
• Enclosure window, mounting method and screw positions
• Maximum stack thickness including cover lens and bonding
• FPC exit direction, length and stiffener position against your assembly path

FPC direction is the single most common reason a mechanically compatible module still does not fit, and it is cheapest to discover here.

Step 2 — Choose the interface from the host, not the catalogue

The interface decision belongs to your host processor, not to the display. Confirm what the host speaks natively before selecting a module, because every mismatch is paid for in a bridge board, extra firmware, or both.

• RGB / TTL parallel — simplest, highest pin count, short cable runs. 16-bit, 18-bit and 24-bit variants use different pin counts and data mapping.
• LVDS — differential, longer runs, lower EMI. The electrical layer is defined by ANSI/TIA/EIA-644-A: differential output 247–454 mV, offset 1.125–1.375 V, receiver threshold ±100 mV over 100 Ω media. The standard is deliberately electrical-only, so two compliant panels can still differ in lane count, colour mapping and whether data is carried in VESA or JEIDA format — a panel wired for one is not compatible with a host expecting the other.
• MIPI DSI — high bandwidth, low pin count, tight signal-integrity requirements; thermal behaviour matters at high data rates.
• SPI / MCU (8080 / 6800) — low pin count, low bandwidth, suited to small panels and resource-limited controllers.

Locking the interface also locks pin count, connector, FPC routing and most of your integration firmware. Change it later and all four move.

Step 3 — Set brightness against the environment, then check the heat

Brightness is specified by the ambient light the display must beat, not by a habit. Indoor HMI panels typically sit in the hundreds of cd/m²; outdoor and direct-sunlight applications need substantially more, and reflection is often the real problem rather than luminance.

The constraint that follows is thermal. Higher backlight current raises LED temperature, and LED output and liquid-crystal response both shift with temperature. A module that meets its brightness rating on a bench can fail to hold it in a sealed enclosure. Review brightness together with:

• Enclosure volume and whether there is a real thermal path
• Cover glass, anti-glare and anti-reflective treatment — a brighter backlight alone does not solve reflection
• Optical bonding, which removes the air gap and recovers contrast
• Whether the resulting current fits your power budget

Step 4 — Define the temperature envelope before choosing the panel

Rated temperature is not a datasheet footnote; it determines which panels and which driver ICs are even candidates. Driver ICs carry explicit rated ranges, and a part rated −30 to +85 °C is not a −40 to +85 °C part regardless of how it behaves on your bench.

Three points that are routinely missed:

• Storage and operating ranges differ. Specify both.
• Cold start is a separate requirement. Liquid-crystal response slows at low temperature, and a heater may be needed to reach acceptable response time at switch-on.
• Qualification must match the application. The test methods exist in IEC 60068-2 — temperature change (Na thermal shock vs Nb cycling, which are not interchangeable), damp heat cyclic vs steady state, sine vs random vibration, continuous vs cyclic salt mist. A supplier quoting "reliability tested" without stating the method, severity and duration has not answered the question.

Step 5 — Treat backlight lifetime as a measured claim

When a supplier states a backlight life in hours, the number is only meaningful with its method behind it. The industry convention is L70 — the point where luminous flux has fallen to 70% of its initial value. Measurement follows IES LM-80, which tests LED packages, arrays or modules at controlled case temperatures for a minimum of 6,000 hours with periodic flux and chromaticity measurement; projection to L70 follows IES TM-21, which fits the measured decay curve and caps the projection at a multiple of the actual test hours.

Two honest caveats worth carrying into your own specification. LM-80 is a constant-temperature, constant-current laboratory condition, so projections do not map perfectly to field behaviour. And in a real luminaire or enclosure, driver failures, encapsulant degradation, phosphor shift and optic yellowing all affect the outcome alongside lumen maintenance. Specify the test basis, not just the number.

Step 6 — Decide touch as an optical and mechanical decision

Touch selection changes the front of your product more than any other choice, because it adds layers:

• Resistive (RTP) — works with gloves and styluses, lower optical clarity, requires actuation force.
• Projected capacitive (CTP) — multi-touch, better clarity, needs tuning for gloves, water rejection and EMI.
• Cover lens thickness and material, and whether chemically strengthened glass is needed
• AG / AR / AF coatings, which trade scratch resistance, reflection and cleanability against each other
• Bonding method — full lamination recovers contrast and reduces parallax, and adds thickness and process risk

If the product will be used with gloves or wet hands, say so at this stage. Retuning a touch controller after the stack is frozen is a mechanical change, not a firmware change.

Step 7 — Put lifecycle in the specification, not the purchase order

The last decision is the one that costs most when deferred: how long this display must remain available. Industrial products commonly need the display to outlive the original production run by years, to cover service spares.

• Prefer panel platforms with a stated long-availability position over consumer-derived glass
• Ask for the lifecycle status of the specific model — active, not recommended for new designs, or end of life
• Ask what the change-notification process covers, and how much notice it gives
• Keep a second-source path open: choose an interface and connector that more than one supplier can build to
• Decide last-time-buy quantity from demand, spares, yield and storage — not from a panic order

This is also where the earlier decisions pay off. A module whose register map and interface you documented is a module you can replace. One whose init sequence nobody kept is a module you redevelop.

The sequence in one table

Work through it in this order and each step constrains the next one usefully. Work through it in any other order and you will revisit at least one decision.

Frequently asked questions

Related pages