Why I Rejected an All-in-One HMI-PLC and Chose a Modular Setup (And What It Cost Us)

Back in late 2023, I was reviewing a proposal for a mid-size packaging line upgrade. The project was straightforward: replace an aging relay-logic panel with a modern PLC-based control system. The client wanted an HMI for operator interface, analog inputs for temperature sensors, and Modbus communication to existing drives. The vendor—a well-regarded system integrator—submitted a design using an HMI PLC all-in-one unit. On paper it looked clean: one device, one programming environment, lower cabinet space. But as the person who signs off on every deliverable before it reaches our customers, something felt off.

The All-in-One Pitch That Almost Slipped Through

The proposal specified an integrated unit combining a 7-inch HMI and a compact PLC with transistor outputs. It included a Modbus IO module for expansion and claimed to handle the eight analog inputs via the built-in I/O. Total hardware cost: $3,400. Impressively neat.

From the outside, it looks like vendors just need to work faster for rush orders. The reality is rush orders often require completely different workflows and dedicated resources. But here, there was no rush. The client had given us a reasonable timeline. The integrator simply defaulted to the all-in-one because it was their standard solution.

"It's simpler, cheaper, and we can program it with the same ladder diagram PLC tools," the project manager explained during our kickoff. "Plus, the Modbus IO module is daisy-chained right from the HMI port."

Simpler? Yes. Cheaper? On first glance. But cheaper at what cost?

The Turning Point: Reading the Fine Print

I flagged three issues during our Q4 2023 quality audit:

1. Analog input isolation. The all-in-one's analog inputs module had no galvanic isolation between channels. The datasheet (which I pulled from the manufacturer's site) clearly stated "non-isolated" for the eight channels. The client's temperature sensors ran across a 50-meter cable run near variable frequency drives. Noise injection was almost certain.

2. Modbus IO module limits. The Modbus IO module supplied was only a 4-slot base. The proposal listed two analog input modules (4 channels each) and one digital output module for valve control—three modules, leaving one slot. No room for future expansion. And the module's power supply was shared with the HMI backlight, meaning if the HMI failed, the entire remote IO went dark.

3. Transistor output capacity. The PLC's transistor outputs were rated at 0.5A per channel, sinking only. The solenoid valves on the packaging line drew 0.7A each at 24VDC. The integrator had planned to use a relay interface to step up—adding complexity and a potential failure point.

I pulled the specs and called the integrator lead. "Look, I'm not saying the all-in-one can't work. I'm saying it's not designed for this application. If we ship this, the first field failure will cost us a $22,000 redo and delay the client's production launch by two weeks."

The Solution: A Modular Approach That Hurt Our Budget but Saved Our Reputation

We went back to the drawing board. Replaced the all-in-one with a separate 10-inch HMI (same brand, ABB), a medium-range PLC with transistor outputs (0.8A per channel—direct drive for the solenoids), and a separate Modbus IO module chassis with isolated analog inputs. The analog inputs module had 2500V isolation between channels. The total hardware cost jumped to $4,950. That's $1,550 more—about 45% higher than the original quote.

People assume the lowest quote means the vendor is more efficient. What they don't see is which costs are being hidden or deferred. In this case, the deferred cost was the risk of a field failure that would have dwarfed the hardware savings.

The integrator pushed back: "We've shipped this all-in-one on ten projects without issues." They were right—they had. But those were packaging lines in clean, climate-controlled environments with short sensor runs. This client's facility was a dusty warehouse with temperature extremes and electrical noise.

I rejected the original proposal. The revision passed after I certified the new design against our internal standard (based on IEC 61131-2, clause 6.3 for environmental conditions). We paid the integrator an extra $800 for redesign hours. Total project cost increase: $2,350.

What Happened Next

The system was commissioned in February 2024. During acceptance testing, we simulated electrical noise on the analog input lines. The modular system passed flawlessly—the isolated inputs showed less than 0.5% deviation. The all-in-one, if we had used it, would have seen ±8% swing, enough to cause temperature control oscillations.

Six months into operation, the client reported zero HMI or PLC downtime. They later added a Modbus temperature scanner module for extra sensors—the 7-slot chassis we specified had room. Exactly what we needed. Simple.

That quality issue cost us a $22,000 redo and delayed our launch—wait, no. That was a different project. Here, we avoided the redo entirely. The only cost was the upfront $2,350, which we explained to the client as an investment in reliability. They accepted it without question after seeing the test results.

Lessons Learned (the Hard Way, but Not Too Hard)

Here's the thing: most engineers don't think about IO isolation until it's too late. The temptation to simplify—to pick the all-in-one, the integrated widget, the "just works" package—is strong. But simplification ignores the reality of industrial environments.

I ran a blind test with our engineering team: same control requirements, two proposals—modular vs all-in-one. 83% identified the modular design as "more professional" without knowing the cost difference. The cost increase was roughly $2,350 per system. On our 50-unit annual volume, that's $117,500 for measurably better perception and actual reliability.

Is the premium option worth it? Sometimes. Depends on context. When the client's brand is on the line—when their production line stoppage means lost revenue and angry customers—the calculation changes.

What would I do differently? I'd involve the integrator earlier in the spec review process. But in the end, sticking to the spec isn't just checking boxes. It's protecting the reputation that customers pay for. Our brand is built on equipment that works the first time, every time. That $2,350 difference? It's the price of being trustworthy.

"The $50 difference per project translated to noticeably better client retention." – Actually, in our case it was $2,350 per system. The principle holds: quality perception drives repeat business.

Note: Pricing data as of early 2024. Current ABB components pricing may vary. Always consult the latest technical documentation for isolation specifications and ratings.