The Upgrade Project That Started It All
In September 2023, my plant manager asked me to upgrade the sensor system on our bottle-filling line. The old photo-eyes were ancient—unreliable, covered in sticky syrup residue, and prone to false triggers. We were stopping the line two or three times a shift just to wipe lenses and realign brackets. Honestly, any upgrade would be an improvement, I figured.
I'd been in controls maintenance for six years at that point. Maybe that was the problem. I had just enough experience to think I knew what I was doing, and not enough to double-check myself.
Let me set the context, because it matters: we're a mid-size beverage bottling plant in Ohio running three shifts. Maintenance team of six. We keep inventory lean, so there's no drawer full of spare sensors back in the shop. When something breaks, we order it, and we wait.
The plan seemed simple enough. Replace the photo-eyes on the filler line, then add a vision sensor at the label-inspection station. Two product categories, one supplier decision, and done.
It didn't go that way.
Comparing Omron and Keyence (Badly)
Here's where the trouble started. I did what most of us do: I Googled “sensor comparison,” opened a few spec sheets, and went down a rabbit hole.
The two names that kept coming up were Omron and Keyence. Both make high-quality sensors. And on paper, Omron looked like the obvious choice—cheaper, comparable response times, faster shipping. Keyence looked expensive. I wasn't just comparing prices, though. I built a spreadsheet with sensing distance, response time, IP rating, cable length. The numbers were nearly identical across both brands.
What I didn't do was read the manuals. Not the Omron manual. Not the Keyence manual. The spec sheets gave me the cherry-picked numbers; the manuals would have shown me how those sensors actually install, how they wire into a machine, and how they talk to our PLC. That's where the differences lived.
There's an old way of thinking that “sensors are all basically the same.” That might've been true thirty years ago, when photoelectric sensors were little more than light-activated relays. Today's units have microprocessors, communication protocols, configurable outputs, and entire software ecosystems behind them. Comparing them by price per unit like you'd buy a relay from a drawer is a mistake. I made that mistake.
I placed the order: twelve Omron photo-eyes and one Keyence vision sensor. Around $4,300 total. Maybe $4,150, I'd have to check the PO. The order arrived in four days. Everything looked fine in the box.
The Day It All Fell Apart
Installation day was a Monday. I'd scheduled eight hours of line downtime, thinking I was being generous. I wasn't.
First problem: mounting brackets. The Omron photo-eyes used a different bracket pattern than our old units. The dimension sheet had the numbers right there, but I didn't map them to the physical reality of the machine. Two hours lost in the machine shop fabricating adapter plates.
Second problem—the expensive one—was the vision sensor. The Keyence model I ordered needed a signal converter to communicate with our PLC. I saw “signal converter required” in the spec sheet. What I didn't do was check which converter. Keyence makes something like fourteen different models in that category, each with its own input/output logic, wiring scheme, and supported PLC families.
I ordered the first one that showed up in the search results.
When we wired everything together, nothing happened. No communication. No response. I stood there with my multimeter, checking voltage at the sensor, checking voltage at the converter, checking voltage at the PLC input card. All fine. The converter just wasn't speaking the same language as anything else on that line.
That's when I should've stopped and called someone. I didn't.
The Multimeter Fuse Moment
In my defense—there is no defense. I have a Fluke 87-V. I know what a capacitor does. I just skipped the step.
While troubleshooting the wiring, I decided to test the old power supply feeding part of the line. There was a capacitor across the output terminals, and I wanted to rule it out. I set the Fluke to DC voltage, touched the leads across the capacitor terminals—and didn't discharge it first.
There's a correct way to test a capacitor with a Fluke multimeter, or any multimeter: discharge the cap first with a resistor or insulated tool, then set the meter to capacitance mode, then probe carefully. Fluke's own safety documentation is crystal clear about this. Their fuses are designed to protect the meter and the user from arc flash, but the person holding the probes is the real first line of defense.
I knew this. I've known it since trade school. I just... didn't do it.
I told myself it never mattered before. And it hadn't—I'd tested capacitors in de-energized circuits a hundred times and they were always drained. That's the kind of luck that runs out eventually. When it did, it went pop inside the meter. The display went dark.
The high-energy multimeter fuse did exactly what it was designed to do. It's a $20 part, and worth every cent. But in that moment, I was less grateful than embarrassed. The second-shift electrician walked in while I was staring at the dead meter and said, “You're supposed to discharge those, right?”
He was right. He's still right.
The Recovery (and the Real Cost)
The final tally on that project, for anyone who thinks skipping the manual saves time:
- $240 for fabricated adapter plates
- $780 for the correct signal converter, with expedited shipping
- One week of rescheduled downtime on a production line that couldn't afford it
- One $20 multimeter fuse (it's never about the $20)
- One permanently dented ego
The worst part? The Keyence manual for that vision sensor has a compatibility chart in the first twenty pages. It lists the exact signal converter models for each sensor and PLC combination. I could've solved the problem in thirty minutes at my desk instead of a week on the floor. Same story with the Omron brackets—there's a small note on the dimension sheet pointing to compatible bracket kits.
The sensors themselves were fine. They did exactly what they claimed. The problem was me, and my refusal to open the PDF files.
We finally completed the retrofit on a Saturday. Factory brackets this time, the right signal converter, and a full afternoon of commissioning with the sensor's configuration software. It worked. It's still working today.
What I Actually Learned
This story is self-deprecating enough to be funny, but the lessons are real. Three of them, in order of pain:
Read the manual before you recommend anything. Spec sheets are for shortlisting. Manuals are for committing. I now download the full manual as soon as a product makes the shortlist, and I read the installation and compatibility sections before I write a PO. Keyence manuals are actually well-structured—installation requirements, wiring diagrams, troubleshooting, parts lists, all in predictable sections. I learned that after the fact.
Comparisons need to be comparisons of applications, not numbers. The Omron vs Keyence question isn't answered by putting spec columns side by side. It's answered by walking the line, looking at the mounting space, the wiring path, the PLC model, and the people who'll operate it. Both companies make excellent sensors. But “excellent” doesn't mean much if the bracket doesn't fit your machine, or the signal converter doesn't match your PLC.
Know the boundary of your expertise. I'm a controls technician. I'm comfortable with PLCs, sensors, and wiring. I'm not an electrical engineer, and pretending I fully understood the capacitor safety side was me crossing a line I wasn't qualified to cross. “I don't know” costs less than “I guessed.” The vendor who says “this isn't our strength—here's who does it better” earns my trust a lot faster than the one who says “we can do everything.” Sometimes the best move is to call the second-shift electrician. Turns out they know things.
One more thought. This approach worked for us, but our situation was pretty specific: mid-size plant, standard industrial voltage, predictable maintenance cycles, and a maintenance team big enough to catch each other's mistakes. If you're running a continuous-process facility where an hour of downtime costs five figures, your tolerance for this kind of learning experience is a lot lower.
So read the manual. Discharge the capacitor. And if you're not sure which signal converter to order, ask someone before the parts arrive.
The manual was there the whole time. I just didn't open it.
