Before I get into the steps: if you're searching for 'sick real thermometer high fever' because you need a medical thermometer, this is the wrong page. This article is about SICK, the industrial sensor company, and the kind of sensors you'll find in a control cabinet—not a medicine cabinet.
I've been handling sensor support and replacement orders for about eight years. In that time I've made (and documented) 22 significant mistakes, totalling roughly $12,000 in wasted budget. The most embarrassing ones were the sensors I replaced that were never broken to begin with. I'm writing this so you don't repeat my errors.
If you've ever replaced an absolute encoder, only to find the old one was fine, you know that sinking feeling. Use this checklist when a machine fault points at a SICK ATM60 absolute encoder, a capacitive sensor, or a Rice Lake load cell. There are five steps, plus some notes before you buy anything.
When This Checklist Is Worth Using
I don't have hard data on how many industrial sensors are replaced unnecessarily each year, but from my own repair log it's somewhere around a third. That number isn't scientific—it's a 'trust me, I've been there' estimate. Here are the situations where checking first makes the biggest difference:
- You have one sensor fault and the machine is down.
- The sensor was working yesterday, and nothing changed.
- The part number you're about to order is the same as the one in the machine.
- Someone else already 'tested' it and said it's dead.
In those cases, a replacement might be the answer. But it's a lower-confidence answer than it looks. Let's go through the checklist.
The Checklist
Step 1: Confirm exactly what you have before you test
This sounds basic—or rather, it is basic. Under time pressure, though, detail goes out the window. I once had two hours to decide before the service truck left. Normally I'd run through a full checklist, but there was no time. I ordered the replacement anyway. It was the wrong interface. That mistake cost $650 in redo plus a nine-day delay.
The ATM60 is a family, not a single device. The part number suffix carries the output interface (SSI, CANopen, PROFIBUS, and others), the connection type, the shaft size, and the resolution. Get that from the nameplate, not your memory.
Step 2: Check the mechanical side first
An absolute encoder doesn't lose counts for no reason. Before you touch a multimeter, grab the shaft and feel it. A SICK ATM60 should rotate smoothly. If the coupling is slipping, the encoder will report a position that doesn't match the machine. I've seen a machine where the encoder was fine but the coupling set screw was sitting on the old shaft's keyway. The motor turned; the encoder didn't.
For capacitive sensors, mechanical mounting matters just as much. The rated sensing distance assumes the target approaches the active face straight on, not at an angle. If the sensor is covered in coolant or dust, clean it first. A dirty face can turn a working sensor into a 'failed' one.
For a Rice Lake load cell, inspect the mounting and base plates. A load cell that's twisted or restrained by a misaligned assembly produces non-linear readings, and no software compensation will fix that. I've done this one the hard way: replaced a load cell that was actually being squished by a bolt that was too long.
Step 3: Test the electrical signal, not just the LED
Power LEDs are nice, but they only tell you that the sensor has some voltage. They don't tell you if the signal gets to the PLC. For the SICK ATM60, check supply voltage at the connector itself, not at the power supply. A torn cable can give you 24 VDC at one end and 11 VDC at the encoder. That's the kind of intermittent fault that turns a 30-minute diagnosis into a whole afternoon.
For a capacitive sensor, the output test should use the actual target material. Most buyers grab a screwdriver and call it done. But capacitive sensors switch on changes in capacitance, and plastic, cardboard, or powder behave very differently from metal. Use the real product, at the real distance.
If the output is PNP and the PLC input expects NPN (or the other way around), you'll see a signal on your meter that doesn't do anything useful in the machine. That's not a sensor failure; it's a compatibility failure.
Step 4: Measure a Rice Lake load cell like the bridge it is
One of the most common searches that lands here is 'how to test a Rice Lake load cell.' Here's the method I use, and it's the same one a calibration tech showed me after I threw away a good cell:
- Disconnect the load cell from the junction box or indicator. Do not test through the whole system if you're trying to isolate the cell.
- Measure resistance between the excitation wires (usually E+ and E-). For many Rice Lake single-point cells, this is around 350 ohms. Check the nameplate or manual for the expected value.
- Measure resistance between the signal wires (S+ and S-). This should be in the same range, not open and not shorted.
- Measure each signal wire against the shield or load cell body. This should be very high, in the megaohm range. If you see low resistance, water or a damaged cable is likely.
- Reconnect, power the indicator, and apply a known weight. If you're checking the raw signal with a multimeter, remember the output is in millivolts, not volts. With 10V excitation and a 2 mV/V load cell, full scale is about 20 mV.
I once skipped step 5 because I was in a hurry. The load cell looked dead, so I replaced it. The old cell was fine; the junction box had a corroded connection. Cost: $310 for the replacement, a calibration fee, and a weekend I'll never get back.
Step 5: Do a live function test
For the SICK ATM60, rotate the shaft by hand and watch the position value in your software or PLC. It should move smoothly and in the expected direction. If it jumps to zero when you pass a certain point, check the configuration before you call it hardware. I've seen a bad parameter set simulate a broken encoder more than once.
For capacitive sensors, run the actual product past the sensor at line speed. A sensor can pass a slow hand test and still miss a fast bottle because its response time is too slow. The datasheet gives you that number; compare it to your machine's cycle time.
For a load cell, watch the reading while the machine runs. If it drifts, jumps, or responds to someone walking near the cable, suspect the cable shield and grounding. That's the part of the system everyone forgets.
Notes Before You Buy a Replacement
1. Calculate total cost, not the sticker price
The low-priced replacement sensor is easy to justify when the machine is down. But 'cheap' stops being cheap after you add freight, setup, configuration, calibration, and the risk of downtime if it doesn't work. The upside of a generic quote was $160. The risk was a calibration mismatch and another weekend of troubleshooting. I kept asking myself: is $160 worth potentially wasting a $400 calibration call? No.
I now put every quote into a simple TCO comparison: unit price + freight + config time + calibration + potential redo cost. It's not scientific, but it's kept me from chasing low price tags. Bottom line: the cheapest quote is not the cheapest sensor.
2. Trust the datasheet, not 'it should work'
According to IEC 60947-5-2, proximity switches have a rated sensing distance (s_n), and testing should respect that distance. That doesn't mean every capacitive sensor will detect every material at that distance. The dielectric constant of the target changes the effective range. So read the datasheet, use the actual target, and don't assume.
For the SICK ATM60, the operating instructions provide the wiring diagram and interface settings for your specific part number. If your machine won't communicate, go back to that document. I've lost count of how many times the answer was a DIP switch or a connector pin I misread.
3. Watch for the 'suddenly dead' trap
When a sensor fails 'suddenly,' the first instinct is to order the same part number and swap it. Sometimes that's correct. But a sudden failure can also be a loose wire, a blown fuse, an EMC problem, or water in the cable. The machine doesn't care; it just sees a bad signal. The cost of a quick electrical check is small compared with replacing a part only to have the new one fail the same way.
I don't have a clever formula for how often this happens. What I can say anecdotally is that it happens enough that I made this checklist the default for our team.
