Scale Drift at 2 p.m.? It's Probably Not the Load Cell

The Problem That Starts With a Picture

At 1:47 p.m. last June, the scale on a packaging line started adding 35 pounds to every bin. Not consistently — sometimes 30, sometimes 50. The indicator jumped like it was trying to be a slot machine. The operator's first call? 'Load cell's dead. Need a new one.'

I get that reaction. I've made it myself. More than once.

Search for an image of a Rice Lake Survivor weighing system and you'll see a clean stainless installation. It looks bulletproof. But that image never shows the junction box, the cable glands, or the 25-year-old disconnect that was definitely not bulletproof. That's where the real problem starts.

What most people don't realize is that 'load cell failure' is often the easiest diagnosis to write down, not the most accurate one. The sensor is the most expensive component in the loop, so it gets blamed first.

The Surface Problem: The Scale Drifts

The symptom feels simple: weight readings drift up or down. Actually, in my notes they usually drift high. Zero won't settle. Calibration won't hold. Sometimes the problem takes a break in the morning and comes back after lunch.

That '2 p.m. drift' is a clue, not a curse. It means heat is doing something. Heat expands terminals, changes cable resistance, and pushes insulation resistance down at the exact moment the production manager is watching.

Every one of those symptoms points at the load cell, right? Not always. In my first year (2017), I was gonna trust the error code and order a replacement cell without checking anything else. My lead made me open the junction box first. The original cell was fine. The cable gland was letting in water.

That was my first lesson: the sensor is the victim, not the cause.

The Deeper Problem: Three Things Behind the Drift

When I stopped treating the load cell as the only suspect, three root causes showed up again and again.

1. Moisture, Not Metal Fatigue

A load cell output is tiny: typically 2 mV per volt of excitation. On a 10 V system, that's 20 millivolts at full scale. Any moisture in a junction box creates a parallel resistance path. You're not measuring the cell anymore. You're measuring the leak.

People think moisture damages the load cell. Actually, moisture is the failure. The cell is just the first component to show the symptom. The assumption that a broken sensor causes the drift is backwards.

In August 2022, I opened a junction box with water dripping out of the lid. The 'failed' cell tested fine once it was dried out. The $620 replacement I'd already ordered went back.

2. Loose Connections and the C8 Compact Thermal Imaging Camera

A loose terminal screw can do more damage than a cracked load cell. The connection adds resistance, resistance creates heat, and heat changes the resistance. It's a feedback loop that makes the reading wander.

That's where the C8 compact thermal imaging camera changed how I work. On one call, I scanned the junction box and saw a terminal block running at 48°C while the rest of the box was at 29°C. The connection looked tight. It wasn't. Ten seconds with the camera saved a $1,100 service charge.

I probably would have replaced the cell, then the cable, then the indicator, and finally, a month later, opened the box. Instead, I re-terminated the wire and the reading settled down immediately.

3. The Weighing Indicator Calculation Nobody Verified

Here's something the industry doesn't print in the brochure: the indicator is not a dumb display. The weighing indicator calculation on a Rice Lake system converts a millivolt signal into a weight using a calibration slope. If someone enters the wrong capacity, decimal point, or gravity factor, the display can be stable and completely wrong.

I once found a 500 lb capacity setting on a 5,000 lb scale. The indicator was repeating perfectly — it was just repeating a tenth of the actual weight. Nobody checked the setup because 'the indicator is fine.' The indicator was the problem.

The phrase 'weighing indicator calculation Rice Lake' shows up in our tech support tickets more than you'd think. Most of the time, it's not the math that's wrong. It's a forgotten setting from a previous job.

Rice Lake's technical documentation says the same thing in different words: the span setting is based on the cell's rated output in mV/V, not on a guessed number. When that setting is wrong, the system is accurate for zero and only zero.

What Misdiagnosis Actually Costs

In 2021, I went back and forth for a day: replace the cell or pull the junction box apart. The replacement was faster on paper. But my gut said the cell wasn't the problem. My gut was right, and I still ordered the cell because the owner wanted speed.

After I ordered it, I kept second-guessing. What if it wasn't the cell? The two-day wait was stressful. I didn't relax until I opened the box and saw the water.

That mistake cost $760 in parts and freight (not including my pride) plus a 3-day production delay. Over eight years, I've documented $28,400 in wasted budget from solving the surface problem instead of the deep one. That doesn't count the credibility hit every time a customer called back.

According to NIST Handbook 44, Section 2.20, a commercial scale must repeat within tolerance and return to zero when unloaded. That standard doesn't care which part you replace. It cares that the system performs. You can't reach that point by replacing a cell that tests good on the bench.

The Checklist That Finally Stuck

The solution isn't a secret software trick. It's a sequence. It's boring. And it stops almost every unnecessary load cell order I used to make.

1. Isolate the load cell. Disconnect it from the junction box. Here's how to use an Extech multimeter for this: set it to DC millivolts, apply 10 V DC excitation to the red and black wires, then read the signal between green and white. A healthy 2 mV/V cell outputs about 20 mV at full scale. Half scale should read about 10 mV. If the output is stable and proportional, the cell is not the problem.
2. Check the environment. Look for water, corrosion, and loose terminals. Scan the junction box and cable path with a C8 compact thermal imaging camera. Hot spots are bad connections. Then use a signal analyzer to check for noise. I once saw a 60 Hz hum that made a Rice Lake indicator roll by 20 pounds. No amount of calibration could fix that.
3. Verify the indicator setup. Confirm capacity, division size, and calibration settings in the indicator menu. Compare them to the nameplate. Then perform a span check with test weights. The Rice Lake weighing indicator calculation is simple arithmetic; the wrong number in a setup field ruins it.

I do not trust a replacement decision until I see the mV/V number. Neither should you. And no, I'm not gonna tell you that load cells never fail. They do. But not as often as the paperwork says.

Since September 2024, this checklist has caught 47 potential errors at my plant. Not all were load cells. Some were cable shields. Some were indicator settings. One was a junction box that had been drinking rainwater for years.

Is the load cell ever the culprit? Sure. But replace one after you've isolated it, not because the operator has a hunch. The shiny image of a Rice Lake Survivor weighing system doesn't show the part that fails. The checklist does.