Your SICK Encoder Didn't Just Fail. Something Killed It.

At 7:12 a.m. on a Thursday, I answered a call from a maintenance planner who was home sick. He was standing over his kitchen island with a laptop open, running a failure investigation from home because his line had gone down at 6:40 and nobody at the plant could explain why yet. The labeler had dropped position feedback. The SICK encoder on the infeed starwheel had gone quiet, and every bottle after that had been kicked out or mislabeled. At 18,000 bottles an hour, that is not downtime. That is a slow-motion disaster.

By 8:20, someone at that plant was searching for sick encoder distributors. We had a genuine replacement in stock, got it onto a courier before lunch, and by mid-afternoon the line was running again. Problem solved. Six weeks later I took the same call—same line, same starwheel, same SICK encoder. That second call is what this article is about.

In my role coordinating emergency parts for manufacturers and labs, I have handled more than 200 rush orders over the last six years. I am not the engineer who finds the fault. I am the person who sees what happens after the first replacement fails. That vantage point changes how you look at breakdowns.

What looks like a delivery problem is usually a detection problem

From the outside, an emergency encoder order looks like a logistics problem. A part failed. Get another one there fast. That is the surface story, and it is half true. The other half is less convenient: most components don't just fail. An encoder dies because something killed it—heat, contamination, vibration, a power fault—and those things usually develop for weeks before the part finally gives up.

Replacing a dead encoder is the right way to restart a line. It is not the right way to prevent the next stop. The same environment that killed the first unit is already working on the second one. So now I ask an uncomfortable question before we ship: what was different this time?

Based on our internal data from 200+ rush jobs, about one in four replacement parts is truly at the end of its natural life. The rest are victims. And the machines almost always leave evidence.

The dead encoder was telling the truth about the motor

Back to that labeler. When the second encoder went down, the plant understandably blamed the part. Instead of shipping a third encoder immediately, we asked them to take a few readings first. A technician walked out with a thermal camera and shot the drive motor. The rear bearing housing was running about 20°C hotter than every other motor on that line.

The bearing was starting to fail. It was vibrating at high frequency, and that vibration had been hammering the new encoder's internal bearing for six weeks until it gave up. From the outside, it looked like a bad SICK encoder. The encoder was just the messenger.

The motor bearing took 45 minutes to change. That encoder has been running fine ever since.

Fluke vs FLIR thermal cameras: the brand question misses the point

That labeler story is not rare. I have seen loose connections, failing contactors, overloaded cabinet fans and misaligned gearboxes show up on a thermal image weeks before they turned into an emergency. Maintenance managers constantly ask me about Fluke vs FLIR thermal cameras, and honestly, both make cameras that would have caught that motor bearing.

Here is what I tell them: Fluke cameras sit naturally in an electrician's ecosystem and are built for life inside panels. FLIR offers a wider range of price points and more mature analysis software. Pick either one. The brand decision matters far less than whether the camera actually leaves the office.

A thermal camera that lives in a manager's desk drawer is worthless. NFPA 70B, the electrical maintenance standard, has long treated periodic infrared scanning as a normal part of electrical maintenance—because heat writes the report long before the breakdown happens. The trick is reading it on a Tuesday, not after the 2 a.m. shutdown.

Dirty power doesn't care what your meter says

Thermal catches heat. Power problems are sneakier. At another plant, a SICK encoder on a cartoner would lose count a couple of times a week. They replaced the encoder, then the cable, then the PLC input card. Still faulted.

We asked them to log the supply voltage over a full shift instead of watching it live. A human cannot stare at a meter for eight hours and catch a dip that lasts a few milliseconds. The log showed the 24 VDC supply sagging every time the machine accelerated—too short to see, but long enough to drop below what the input card considered a reliable logic level.

The cause was an aging power supply with a failing capacitor. It had been cooking inside a hot cabinet for years. Replacing the supply fixed it. The encoder was fine all along.

When we ask for field readings like these, one question matters: is that meter true RMS? Modern lines are full of variable-frequency drives and switching supplies that distort the voltage waveform. A meter that assumes a clean sine wave can be off by 10 percent or more, and a false in-spec reading sends you chasing a new encoder that was never the problem. In most plants, the logging meter that comes out for this job is the Fluke 289 true RMS multimeter, because it can record min/max and average readings unattended for a full shift. That capability is worth more than the resolution of any single reading.

Laboratory equipment lies in different units

The same pattern shows up in laboratories, just with different vocabulary. A pharma QC lab once asked us to rush a detector lamp for their chromatography system because the baseline noise was getting worse every week. They installed the lamp. The noise stayed. They were about to order a new column when someone thought to check the electrical supply.

Their chromatography instrument shared a wall circuit with a −80°C freezer. Every time the freezer compressor cycled, it dragged the voltage down just enough to show up as baseline wander. They moved the freezer to another circuit. The baseline cleaned up, the lamp was fine, and the column they almost ordered is still sitting in stock.

Same story, different instrument: the symptom gets blamed, the environment gets ignored.

The gray market encoder is the most expensive discount you will ever buy

All of this assumes the replacement part is genuine. When your line is down and you are searching for sick encoder distributors at 9 p.m., the search results will include listings that look too good to be true. Some of them are.

A maintenance manager at a steel mill once told me he could buy SICK encoders for 35 percent less from an online seller. He bought four. Two failed within six months. The cheap price did not include the cost of pulling the machine apart twice, nor the second round of emergency freight, nor the credibility he lost with his own production manager.

I am not going to name platforms or competitors. I will give you one rule: a genuine SICK encoder has an OEM part number, and an authorized distributor can be verified through SICK's official distributor locator on sick.com. That check takes two minutes. Gray market units often arrive with no serial traceability, no factory warranty, and no one to call when they fail.

The vendors I trust list all charges up front—part price, freight, duties, expected delivery window—before I commit. The ones who add fees after the fact are the same ones who disappear when a warranty claim happens. The lowest quote is rarely the cheapest purchase.

What changed after we got tired of watching failure reruns

Our company still does rush orders. That is the business. But we stopped treating the rush order as the whole solution. Four changes came out of the patterns I have seen:

1. Before we sell the same replacement part twice for the same machine, we ask for basic readings: temperature, supply voltage, motor current, and what changed before the failure. If the customer cannot provide them, we send a short checklist.
2. Every serious maintenance team we work with now owns two tools before they invest in any fancy condition monitoring: a thermal camera and a logging true RMS multimeter. Those two instruments have solved more repeat failures than any predictive software I have seen.
3. Critical machines get a spare encoder or sensor in the cabinet. Not the whole range—just the one part that stops production when it fails.
4. We build relationships with authorized distributors before the emergency, not during it. You want to know their after-hours policy before you need it at 11 p.m., not while your line is down.

None of this prevents every breakdown. Machines still wear out. Bearings still fail. But the failures that surprise us are almost never the first failure. They are the ones we refuse to investigate.

So the next time a machine stops and someone says it needs a new SICK encoder, they might be right. But before you bolt it in, ask the same question we ask now: what killed the last one?

The machine already knows. It has been writing the answer in heat, vibration and voltage for weeks. It is worth reading before you order a rerun.