We Blamed the Load Cell. It Wasn't the Load Cell.

It was a Tuesday in March 2024 when our biggest customer called about short weights. Three pallets in a row were under by roughly 3 kg each. In your head, that may not sound like much. In our world—thousands of shipments a month, tight order tolerances, a customer who audits our process—it was the kind of call that makes you set down your coffee and start taking notes.

The first suspect was obvious. Our floor scale on Line 3 was a Rice Lake weighing systems model 120. It had been there for years, through production overruns and late-night maintenance shifts. I assumed it was a scale problem because that's what the complaint pointed to. I also assumed the scale was in decent shape because it's a Rice Lake. That second assumption, I learned, was exactly the wrong kind of trust.

Suspect Number One: the Scale

We did what most maintenance teams do first: put test weights on the platform and watched the display bounce. It didn't settle. That pointed toward the load cell or the cable. So we pulled the rice lake load cell manual off the shelf in the maintenance office.

The troubleshooting section is fairly practical. I want to say it says to verify the junction box and cable before condemning the sensor—paraphrasing, but that's the gist. One line stuck with me: “Unstable readings are usually traced to a bad connection or damaged cable before the load cell itself.” We should have listened to that line earlier.

Our senior technician, Danny, has been at this plant for about 11 years. He's not the type to grab the first tool in the cart. He looked at the readings, walked back to the tool room, and came out with two different multimeters.

“Let's compare meters,” he said. “If they don't agree, we're chasing a ghost.”

I hadn't thought about that. What I soon learned is that a multimeter comparison isn't just about whether both meters read 1.2 mV. It's about resolution, accuracy class, and whether the leads are shielded. Danny's handheld meter read 1.6 mV at the load cell output. The older shop meter read 1.2 mV. Same test point. Same probe position. Different results.

Let me rephrase that: 0.4 mV might sound tiny, but on a 300 kg load cell with 2 mV/V sensitivity, that's tens of kilograms of apparent error. Enough to explain the complaint entirely.

Where the Multimeter Led Us Astray

At that point I was ready to buy a better multimeter and call it a day. Danny had a different instinct. He borrowed a thermal imaging camera from the panel shop and spent a few minutes scanning the junction box on the scale.

That scan changed everything. One terminal block was slightly warmer than the others. Not visibly so. But clearly hot enough to indicate a poor connection under current. We opened the box, and the screw on that terminal was loose. Really loose. The kind of loose that makes intermittent contact when the scale vibrates with a fork truck driving past.

So the load cell was fine all along. The cable was probably fine. The problem was a connection that cost twenty cents to fix but took us three days to find, because the meter was telling us what we wanted to hear: “it's the expensive part.”

That was my real lesson in this whole mess. It wasn't a brand failure. It was a verification failure.

Insulation Tester vs Megger: Just Use the Right Tool

Once we had the junction box open, we also noticed some greenish crud on the cable jacket near the conduit. That raised the question of moisture ingress. Here's where I have to admit my personal knowledge runs out. I'm not an electrical engineer, so I can't walk you through the current-leakage theory behind insulation testing. What I can tell you from a quality-manager perspective: when moisture gets into a scale cable, it can cause the exact same unstable-reading symptom as a bad load cell—and a standard multimeter in continuity mode won't catch it.

That's when the insulation tester vs megger conversation finally made sense to me. In our shop, people use these terms almost interchangeably. The instrument applies a voltage to the cable and measures insulation resistance. Danny set his to 500 V and got a reading below 1 MΩ. That's a red flag. A cable in good condition should read much higher, from what I understand. Water had been wicking its way under the jacket.

If we had stopped at the multimeter comparison, we'd have replaced a perfectly good load cell and still had a wet cable in the ground. Sometimes the difference between two tools is not one being “better.” It's one asking a different question.

Fixing It Was the Easy Part

The actual repair was almost boring. We clipped out the damaged section of cable, put in a new gland, torqued the loose terminal, and ran a full calibration routine. The display went to zero and stayed there. Total work at the scale: about two hours. Total diagnostic time before that: about three days.

And while we were at it, we double-checked our tolerance assumptions against the 2024 edition of NIST Handbook 44, since this scale fed directly into a commercial shipment process. That handbook is updated regularly, so verify current requirements at nist.gov if you're doing the same—but it's a solid reference for acceptance tolerances. As of the 2024 edition, acceptance tolerances are tied to the scale's number of divisions, not a single blanket percentage. I had that backward, and the handbook corrected me.

The cost calculation that stayed with me wasn't the overtime. It was the rush replacement inventory we had to ship to the customer, the Line 3 downtime, and the awkward phone call where our account manager had to explain why a process we said was under control wasn't. That's what inefficiency really costs. It never shows up on the purchase order for the repair.

After that fix, I ran a small exercise with the quality team: same junction box, same loose connection, but with three different starting tools. One person started with the multimeter. Another started with the thermal camera. The third just opened the junction box and looked. The camera took eleven minutes to reach the right answer. The multimeter path took two days, because the meter didn't suggest looking at the connection. The visual check found it in four minutes, but only because we already knew where to look. Efficiency isn't always about fancy equipment. It's about knowing what question to ask first.

What I'd Do Differently Today

I still trust the Rice Lake name, but now I verify the whole chain: connections, cable, sensor, and instrument. If you're staring at an unstable weight display from any brand, here's the order I'd work through:

• Check the junction box first—loose terminals and corrosion are more common than sensor failure.
• Look at the cable path for pinch points, crushed spots, or moisture around the conduit.
• Do a quick multimeter comparison if there's any doubt about the test tool itself.
• If the scale is in a wet or vibrating environment, use a thermal imaging camera and an insulation tester. They answer questions a multimeter can't.

And keep the manual—the rice lake load cell manual, or whatever manual matches your equipment—close to the scale, not in a binder on the third floor. I'm fairly sure that binder on the third floor is now where we keep old work orders, but at least the PDF is on everyone's tablet.

Looking back, the thing I was most wrong about was not “the scale is broken.” It was “the brand guarantees the details.” No brand does that. The details are on you—or, in my case, on the quality team that catches them before the scale does.