The Checkweigher Was Rejecting Good Jars. Loose Tamper-Evident Bands Told Us Why.

I’m a QC supervisor at a contract packaging plant. We fill glass jars and bottles for food and supplement brands, and one of my jobs is making sure the instruments that guard the line are actually the right instruments. After seven years in this role, I’ve personally documented 11 significant equipment mistakes, totaling roughly $15,000 in wasted budget. This is the one I use for new-hire training.

I still kick myself for not calling an instrument specialist on day one. The whole investigation would probably have taken one afternoon instead of four days.

Tuesday, 6:40 a.m., April 2022. I heard Line 3’s checkweigher from the break room before I saw it: rapid beep, reject gate, clink into the bin. Rapid beep, reject gate, clink. A checkweigher that is happily rejecting product makes this rhythm. I walked over expecting a jammed reject gate. What I found was worse: about three jars a minute were being sent to the reject bin, and every one of them weighed fine when we checked it by hand on a calibrated bench scale.

The same week, a customer sent us a photo. The jar they received had a tamper-evident band hanging loose around the cap. It hadn’t broken—it had never shrunk down in the first place. So now we had an impossible combination: false underweight rejects and loose seal bands on the same line. The checkweigher and tamper evident failures looked unrelated. I decided, incorrectly, that they were.

The portable oscilloscope I bought by bandwidth

I reasoned that false rejects usually mean the scale is getting bad timing signals. If I could watch the position sensor signal live, I would see the noise. What I needed, I announced with great confidence, was a portable oscilloscope. So I bought one: a handheld, two-channel, 100 MHz unit with a tempting price and a delivery time of two days.

The first hour was beautiful. The second hour was not. I tried to measure a point on a control circuit that wasn’t referenced to earth ground. The oscilloscope’s ground clip effectively shorted that circuit to ground, and I cooked an output fuse on a PLC card. The fuse cost about 40 cents. The lost morning cost a lot more.

What I learned: bandwidth is not the number that matters if you can’t connect the probe safely. Input isolation and IEC 61010-1 measurement category ratings matter far more when you’re probing floating control circuits. I checked neither.

The scope is still in our toolbox and gets used regularly. It wasn’t a bad purchase. It was a purchase made before I asked the question that should come first: can I connect this to the point I need to measure?

The IC thermal camera with the wrong working distance

While the maintenance team reterminated wiring, I moved to my next theory. Intermittent faults on boards are often heat-related. A component warms up, expands, and fails until it cools. I needed an IC thermal camera to find the hot spot. So I bought one.

This time I did compare specifications: detector resolution, thermal sensitivity, price. What I did not check was the minimum focus distance. When I aimed the camera at the suspect board from twenty centimeters away, the board turned into a warm, useless blur. There were plenty of pixels, but the lens could not focus close enough to resolve the small ICs I was hunting.

Board-level thermal work comes down to spot size, not just pixel count. If the camera’s spot at your working distance is bigger than the component, you are measuring the component plus everything around it. That is how you aim at a failing IC and see nothing unusual. I learned that after the purchase, not before.

In both cases, the lesson was not “tools are bad.” It was “define the measurement problem first, then pick the tool.”

The Mitutoyo vs Starrett calipers argument

Somewhere in the middle of the chaos, two of my techs started arguing about the band position on the jar. One was using a Mitutoyo digital caliper. The other had a Starrett. Their readings disagreed by about 0.02 mm, and that somehow became a full Mitutoyo vs Starrett calipers debate.

I checked both calipers against the same gauge block. Both were within tolerance. The real difference was measuring technique and the fact that shrink wrap does not hold still for a caliper. We spent two hours debating brands of tools that were not the problem.

It’s the same trap I had already fallen into twice: obsess over tools and specs before understanding the defect. The actual defect was still sitting in the line, waiting for us to look in the right place.

The Evident scientific microscope that ended the hunt

On day three, our senior maintenance tech finally said the sentence that changed everything: “This smells like a cracked solder joint. We can’t see cracked solder joints with the naked eye.”

I called an instrument distributor and asked for a microscope demo. A specialist from Evident named Dana called back. She asked questions I was not prepared for: What exactly fails? What is the failure mode? How big is the area of interest? What lighting do you have?

Then she told me something I did not expect: “We build very nice metallurgical microscopes. I would not sell you one for this job. A cracked solder joint on a through-hole pin is best seen at 20x or 30x with oblique light, which is stereo microscope territory. And if your problem was under a BGA package, you would need X-ray inspection—which is not what Evident does.”

That honesty earned the sale more than any spec sheet could. A supplier who tells you what you do not need is a supplier you can trust with everything else.

A loaner Evident scientific microscope showed up the next day—a stereo zoom model, not the metallurgical system Dana had talked me out of. Manuel placed the suspect board on the stage, tilted a gooseneck lamp until the light raked across the solder joints, and looked at 20x.

He found it in about twenty minutes. A crescent-shaped crack around one pin of the connector that carries the position signal to the bander and the checkweigher. When he pressed gently on the connector, the crack opened, and the signal on our portable oscilloscope dropped out. Release the pressure, and the signal came back. That explained everything: when the signal vanished, the bander applied heat at the wrong moment, leaving loose tamper evident bands, and the scale triggered while jars were not fully positioned, producing false underweight rejects.

We reflowed the joint, added strain relief, and put the board back in service. The line ran the rest of the week without a single false reject.

What I keep on the checklist now

We still bought that stereo microscope. It was the right tool for the question, and we use it for failure analysis, seal inspections, and connector checks. But the purchase was not the real fix. The real fix was a process that I now force myself to follow before any equipment order:

• Write down the suspected failure mode and what evidence would confirm it. If you cannot name the evidence, you are not ready to buy a tool.
• Check the electrical environment and safety ratings first. For a portable oscilloscope, that means input isolation and IEC 61010-1 category, not just bandwidth and sample rate.
• Confirm the working distance you can actually achieve. An IC thermal camera is only useful if its spot size at your working distance is smaller than the component you need to inspect.
• Ask a specialist to test your theory before you spend. A straight answer like “that is not the right tool for this” is valuable information, not a weakness.
• When two instruments disagree, verify them against the same standard before arguing about brands. The Mitutoyo vs Starrett calipers debate taught us that.

This whole episode happened in spring 2022. Prices and models have moved on since then, so verify current specs before ordering. I still use the portable oscilloscope and the IC thermal camera. They are good tools. They just were not the tools that solved this problem—and that failure taught me more than any successful purchase ever has.