16×4 LCD vs touchscreen HMI in industrial panels: when each one belongs

The design engineer opens the RFQ and reads: "industrial panel with colour touchscreen HMI, 7 inches minimum". It is the fifth time this semester he sees that text. The panel in question has eight DC channels to monitor, three local alarms and a Modbus network that pushes everything up to SCADA. The HMI will sit on the same screen 99 percent of the time. The cost of an embedded-Linux HMI with dedicated 24 V supply and quarterly security patching is three times the cost of the 16×4 LCD that would solve the problem. And it adds one more IP address to the cybersecurity inventory. This article is the material that engineer needs in hand to push back on the RFQ with technical arguments, instead of simply complying out of inertia.

What a 16×4 LCD actually delivers

A 16-column by 4-line character LCD is one of the most mature technologies in industrial automation. Sixty-four ASCII characters on screen, LED backlight, HD44780 or equivalent controller, parallel or I²C interface to the microcontroller. There is no operating system, no window manager, no network stack behind it. The device firmware draws character by character.

Immediate local readout. The operator looks at the screen and sees voltage, current, alarm status in a font legible at one metre. No boot, no splash screen, no "please wait, initialising". Information is there as soon as the panel is energised, in under 200 ms.

Three physical key navigation. Left, right, enter cover 100 percent of local configuration use cases: scrolling channels, opening setpoint menus, confirming values. Mechanical keys with defined travel, tactile feedback, no capacitive calibration required, no ghost touch from dirt or gloves.

Very low BOM cost. A certified industrial 16×4 LCD module costs between 8 and 25 US dollars depending on volume and temperature range. Adding three tactile keys, connector and bracket, the interface subassembly stays below 40 dollars in the finished product. A 7-inch industrial touchscreen HMI starts at 350 dollars and easily passes 1,200 in Linux/Ethernet versions.

Very high MTBF. Character LCDs have no moving parts, no large electrolytic capacitor, no complex switching power supply. Typical industrial MTBF exceeds 100,000 hours, against 40,000 to 60,000 hours for commercial touchscreen HMIs.

No operating system. This is the most underestimated advantage. There is no Linux kernel, no TCP/IP stack, no graphics library carrying CVEs. The LCD attack surface equals zero: it displays what the device firmware told it to display, and the device firmware is signed, audited and exposes a declared Modbus map.

What a touchscreen HMI delivers that an LCD does not

There are scenarios where the touchscreen HMI is the technically correct choice, and pretending otherwise would be dishonest. Worth listing what it genuinely adds.

Rich graphical interface. Coloured mimic diagrams, status animations, real-time trend charts, gauge indicators. In process panels with twenty or more correlated variables, visual representation helps the operator understand system state without reading number by number.

Configurable visual and audible alarms. Alarm pop-ups with visual priority, alarm history filtered by severity, local acknowledgment with operator logging. All of this can be done on an LCD with well-written firmware, but the HMI's visual granularity is higher.

Local history. The HMI has mass storage (SD card, eMMC) and can retain days or weeks of log even without SCADA available. For panels in remote sites without continuous connectivity, that is real value.

Native connectivity. Ethernet, optional Wi-Fi, USB for configuration thumb drive, direct integration with Modbus TCP, OPC UA, MQTT. In projects that need a local protocol gateway, the HMI already ships with that function embedded.

Customisation without reflashing the device. The integrator changes screens, adds variables, modifies alarms via the HMI's project software, without touching the field device firmware. In a large fleet with per-customer personalisation, this dramatically reduces engineering cycles.

Total cost: BOM, integration, training, spares

Sticker price is just the entry of the bill. Total cost across the typical 10 years of an industrial panel includes much more.

Cost component	16×4 LCD + 3 keys	7" Linux touchscreen HMI
BOM (display + interface)	30–40 USD	350–1,200 USD
Screen engineering (NRE)	Already in firmware	20–80 h per project
Panel integration (cutout, mounting, IP)	Standard DIN hole	Custom cutout, sealing, dedicated 24 V supply
Operator training	4-page manual	Manual plus navigation training
Security patching	Not applicable	Quarterly minimum, CVE-driven
Spare parts (10 years)	Trivial stock	Specific display, end-of-life in 5–7 years
Touch calibration	Does not exist	Every 1–2 years in dirty environments
Field replacement	15 minutes, no special tools	1–2 h, recommissioning
Estimated 10-year total cost	80–120 USD	1,800–4,500 USD

The figures above reflect averages for industrial DC monitoring, telecom and substation panels. In complex process panels with 50+ variables and per-shift recipes, the math flips and the HMI pays back in engineering cycle.

Reliability in real industrial environment

The panel environment is rarely the datasheet environment. Climate-controlled rooms exist in sales decks; in the field we find telecom shelters at 55 °C inside, battery rooms with acid vapour, substations with vibration and direct solar UV through the window.

IP rating. A 16×4 LCD behind a polycarbonate window with a gasket easily reaches IP54, IP65 with modest effort. IP65 frontal touchscreen HMIs are common, but capacitive touch loses sensitivity under heavy protection film and fails with internal condensation.

Vibration. The LCD parallel connector handles IEC 60068-2-6 vibration with margin. Touchscreens with large glass have a mechanical failure point at the laminate edge and at the internal flat-flex connection.

Temperature. Industrial character LCDs typically operate from -20 to +70 °C in industrial grade and -30 to +85 °C in extended grade. Commercial touchscreen HMIs sit at -10 to +50 °C; industrial versions climb to -20 to +60 °C with dedicated supply, at double the cost.

UV and ageing. Passive character LCDs barely lose contrast under UV. IPS and capacitive screens age visibly: capacitive loses touch precision, IPS develops hot spots and uneven backlight after 3 to 5 years in direct sunlight.

Touch cycles. Industrial touchscreens are specified at 1 to 5 million touches at the same point. In panels where the operator taps the "ack alarm" zone 50 times a day, that spot saturates in 5 to 10 years. A quality industrial mechanical key exceeds 10 million cycles.

Backlight life. LCD 16×4 LEDs reach 50,000 hours half-life. Low-cost touchscreen HMI backlights fall to 20,000–30,000 hours. For a 24/7 panel, 50,000 hours equals 5.7 continuous years; 20,000 equals 2.3.

Compliance and cybersecurity

This section is where the conversation tilts toward LCD for any new panel from 2024 onward. Standards have shifted, and an RFQ that requests "HMI" without qualifying opens a vulnerability door the customer will discover in the first audit.

Attack surface. A commercial Linux HMI has, as a minimum: kernel, libc, BusyBox, configuration web server, SSH client, TCP/IP stack, graphics library (Qt or similar), OPC UA middleware. Each one is a CVE source. In 2025 alone, more than 200 relevant CVEs were catalogued in typical embedded Linux HMI stacks.

Patch management. IEC 62443-4-1 requires a formal vulnerability management process. For a Linux HMI, that means monitoring published CVEs, validating applicability, generating a signed patch package, distributing to the installed base, evidencing installation. It costs recurring engineering. For an LCD without OS, the obligation does not exist because the surface does not exist.

OT inventory identification. Every Ethernet HMI gets an IP address, MAC, hostname and enters the OT inventory. The auditor asks: how do you confirm firmware integrity? How do you detect tampered firmware? Who has root access? How do you roll out the latest patch? For the device LCD, the answer is: the device firmware already answers all of that, and the LCD is not a network device.

Zone segregation. IEC 62443-3-2 demands zone and conduit segregation. Adding a Linux HMI to the field zone means expanding the zone or creating a dedicated conduit. Adding an LCD to the device does not change zonification.

SL simplification. Reaching SL2 on a device without OS is a firmware design exercise. Reaching SL2 on a device with embedded Linux HMI requires Linux hardening, extra processes and supplier audit on the HMI vendor. Certification cost grows 2 to 5 times.

Decision matrix by application

This is not a religious choice. It is an application choice. The matrix below comes from direct observation of hundreds of projects in telecom, energy, industry and data centre.

Application	Recommendation	Why
Telecom site (tower, rooftop)	16×4 LCD on device + central SCADA	Panel without frequent local operation; everything managed remotely via Modbus/SNMP. HMI would be dead cost.
Primary substation	Depends	General control panel: HMI makes sense for mimic and alarms. DC auxiliary panel: LCD is enough.
Industrial process panel	LCD if local I/O; HMI if frequent setpoint/recipe	Daily recipe change justifies HMI. "Set-and-forget" panels do not.
Industrial DC monitoring panel	16×4 LCD	Variables are voltage/current per channel, low interaction, high MTBF criticality.
Data centre	HMI on lead rack + LCD on nodes	Consolidated view at room entrance; granular per-device LCD monitoring on each rack.
Railway control panel	SIL-certified HMI	Information volume and operational demand call for HMI; LCD complements on auxiliaries.
Production / test bench	HMI	Operator interacts constantly, screens change per product model, customisation is an asset.
Remote pumping skid	LCD	Sporadic technical visit, basic local readout, central SCADA runs the rest.

How the AEM-60DC8 handles this decision

The LRI AEM-60DC8 — Industrial DC Monitoring Platform — was designed with a 16×4 LCD + 3 keys as a deliberate decision, not as a cost cut. The reasoning is as follows.

Target application is industrial DC monitoring. The eight isolated 0–60 V DC channels exist to measure busbar, battery strings, rectifier output. Local interaction is voltage reading, derived current reading (when shunt is present), Modbus address configuration, baudrate configuration and alarm thresholds. All of those tasks fit comfortably in a 4-level menu tree with 3 keys and 16×4 display.

Predictable customer cost. The platform stays in an accessible price band for large panel fleets. An integrated HMI would double the final product price without adding capability to the real product use case.

MTBF compatible with 24/7 panels. Operating 100,000+ hours without a display event in an industrial panel is compatible with the expected life of the monitored rectifier and battery bank. It makes no sense to put a 40,000-hour MTBF HMI inside a system that must last 15 to 20 years.

Zero CVE on the display. The product cybersecurity goal is IEC 62443-4-2 SL2, with Ed25519-signed firmware and a documented 147 holding registers map. Keeping the display out of OS scope lets the entire security discussion happen around the Cortex-M0+ firmware on the STM32G0B0RE, which is controllable, auditable and has a patch cycle under our direct control. Putting Linux there would expand the attack surface disproportionately to the UX benefit.

Modbus RTU server for HMI integration when the customer wants it. When a panel really demands a process HMI, the recommended path is for the HMI to talk to the AEM-60DC8 via RS-485 in Modbus RTU at any standard baudrate (4800/9600/19200/38400/57600/115200 bps). The HMI becomes the integrator's responsibility, in the brand and model they prefer, and the AEM-60DC8 remains a lean field device with its local LCD for independent diagnostics. When the HMI is down (and it will be, at some point), the operator still has a local readout.

When to add HMI on top. On sites where a system HMI already exists (substation control room, plant mimic, local supervisor), the AEM-60DC8 enters as a subordinate Modbus device, mapped into the existing screens. The gateway between device RS-485 and HMI Ethernet is typically an industrial converter already present in the panel. No additional engineering on the device side.

Frequently asked questions

1. Isn't a 16×4 LCD too old a technology for a 2026 product? Mature technology is not the same as obsolete. Character LCDs remain the best cost-to-MTBF-to-attack-surface ratio for local readout of few parameters. New technology is not a virtue per se; it is a virtue when it solves a real problem.

2. How does the operator see historical trends if the LCD only shows instantaneous values? Historical trending is a SCADA function, not a field device function. The AEM-60DC8 delivers instantaneous value plus minimal statistics in 147 holding registers, and the SCADA stores, plots and alarms. Trying to do trending on a 16×4 LCD forces the wrong device to do the right device's work.

3. Can I have a touchscreen HMI and a 16×4 LCD in the same panel? Yes, and that combination is very common. The HMI acts as the general panel interface; the AEM-60DC8 LCD acts as independent device diagnostics, useful when the HMI is unavailable, in commissioning, or under maintenance.

4. If the LCD fails, can it be replaced in the field? On the AEM-60DC8 v1.03 the display is an integrated PCB component. Field replacement is per unit, not per loose display. Given the compact unit and high subassembly MTBF, the maintenance strategy is shelf spare, not field repair.

5. Is there a future AEM-60DC8 with integrated HMI? Not on the roadmap. The LRI product line serves lean field devices, Secure by Design, with Modbus integration to any HMI the customer chooses. Adding an integrated HMI would mischaracterise the product as a field device and expand the attack surface we currently keep under control.