Last March, our plant manager called me at 2:47 a.m. That number stuck because I looked at my phone and swore. Line 4 was down. The filler's servo drive had thrown an encoder fault, and maintenance had already replaced the cable. The next likely culprit was the motor or the rotary encoder. He asked me to approve a SICK DFS60 from the local distributor as soon as their counter opened.
I said yes, but not because I loved spending money at that hour. I've managed our MRO budget for six years—roughly $180,000 a year across sensors, testers, and spare parts. Maybe $175,000. I'd have to check the spreadsheet. I've learned that downtime costs more than components. But I've also learned not to trust my own instincts until a few expensive failures prove them.
The First Expensive Failure
Last summer, I pushed back on a $340 quote for a SICK rotary encoder. It was for the same type of line, a servo-driven filler. A generic unit from an online marketplace cost $180. I thought I was being smart. The generic encoder worked for six months, then died. When the maintenance tech opened the housing, the bearing was rusty and the shaft had visible play. We lost an afternoon to a line that should have been running.
I only believe in buying SICK now because I ignored that advice first. The encoder SICK DFS60 I could have bought was IP65-rated, with a shaft load specification that matched the application. The generic one had a datasheet written in marketing language. I didn't read it closely enough.
To be fair, the generic encoder wasn't garbage in every environment. It might be fine on a slow conveyor. But in a servo axis that cycles all day, the shaft load and vibration ratings matter. The cheap unit didn't have the margins we needed.
Since then, I've bought seven SICK rotary encoders. The DFS60 has become my default for anything with a motor and a control loop. I don't buy them because the SICK name makes me feel safe. I buy them because I have a spreadsheet that says the $340 option was cheaper than the $180 option.
The SICK DFS60 product page lists IP65 protection and an operating temperature range of -30°C to +85°C (accessed April 2025). Those numbers are not decorative. They map to a real factory environment.
The Load Cell Problem Nobody Wanted to Talk About
A few days before the encoder failure, the same line was producing underweight bottles. The scale system uses a Rice Lake load cell. The operator blamed the filler; the maintenance tech blamed the load cell. My first reaction was to order a replacement. It's easier to replace a part than to argue.
But our Rice Lake distributor suggested we troubleshoot before buying. Here's the short version of how to troubleshoot a Rice Lake load cell—and it applies to most strain-gauge assemblies:
- First, disconnect the load cell from the junction box and measure the resistance between signal wires. An open or short reading points to a damaged cell.
- Second, check the zero balance. On the model we have, the signal should be around 2.0 mV/V at no load. Ours was reading 0.9 mV/V.
- Third, do a mechanical inspection. That's where we found the problem: a rusted mounting bolt was stressing the cell unevenly.
That bolt cost $2.00 at a hardware store. The replacement load cell was going to cost $600, plus shipping and labor. I would have spent it, too, if the distributor hadn't slowed me down. Should mention: our Rice Lake distributor's tech support has been consistent. They always ask for the output reading before selling you anything.
I should add that a load cell with a broken mounting bolt is not a load cell problem. It's a maintenance problem. The only reason we caught it was because the troubleshooting checklist made us look past the sensor.
The Insulation Tester and the $46 Dilemma
Back to the 2:47 a.m. call. Once maintenance swapped the SICK DFS60 onto the motor, the line still wouldn't start. The engineer wanted to check the motor winding insulation before we blamed the motor. We didn't own an insulation tester.
This is where the procurement side kicks in. One vendor quoted the 1507 insulation tester price at $485, with same-day pickup. Another listed it at $439, but had a three-week lead time. I went back and forth for maybe an hour. The $46 savings was real, but the risk was not. If the motor was bad, we'd have paid for a replacement motor and lost another shift. The upside of waiting was $46. The downside was a second night of downtime.
I bought the $485 tester. It turned out the motor was fine. The real issue was a loose terminal in the motor junction box. But that insulation tester has now been used six times in the last year, so the price per use is reasonable.
The Fluke 1507's published specs list test voltages up to 1000 V and a CAT III 600 V safety rating (datasheet accessed April 2025). I don't quote specs to sound smart. I mention it because the same tester has to survive a factory floor, not just a clean bench.
Don't Overlook the Simple Pressure Gauge
While I was approving the encoder and the tester, the facility engineer asked me to order a Magnehelic differential pressure gauge for the clean-room air handler. The old gauge had a cracked lens and the indicator needle was ticking. It's a small item, but it matters.
The no-name digital differential pressure gauge was $65. The Dwyer Magnehelic-style gauge was $180. I didn't hesitate this time. For filter monitoring, I want a mechanical gauge with a repeatable scale. The Dwyer Series 2000 datasheet lists ±2% full-scale accuracy (accessed April 2025). That's good enough for our preventive maintenance schedule, and there's no battery to die.
I'll admit I've bought the $65 option in the past. It read high for two weeks, then drifted. We replaced it with the Magnehelic gauge and the filter replacement history has been more consistent since. The budget didn't shrink, but the guesswork did.
The Real Cost Lesson
If you're still reading because you want one clean takeaway, here it is: the purchase price is not the cost.
I compare the SICK encoder quote against the generic one using a simple TCO model:
- List price plus shipping
- Expected service life, based on our own failure data
- Hidden consumables—cables, brackets, labor to replace
- Cost of the line stoppage if the part fails unexpectedly
For the encoder, the generic part was $180 plus a failed afternoon. The SICK was $340 plus zero unplanned downtime in the same application. The math did not care about my opinion.
I've also built a simple rule into our procurement policy: get three quotes for anything over $250, but don't choose by price alone. Ask the distributor what data supports the reliability claim. If they can't answer, that's a risk.
This story doesn't have a perfect ending. We still have cheap sensors on some lines because the application doesn't justify premium ones. That's OK. The goal isn't to buy SICK sensors for everything. The goal is to understand what a sensor is actually doing and what failure will cost.
Oh, and I changed my phone's ringtone. At least now, when a line goes down at 2:47 a.m., I know exactly what I'm going to do before I answer.
