Three Test Setup Mistakes That Cost Me Thousands—and What the Keithley DMM6500 Taught Me

I've been running a small independent test bench for six years now. The Keithley DMM6500 has sat on that bench for almost all of that time, and I've put it through situations that would make a metrology engineer wince. But the DMM6500 wasn't the problem. Most of the time, I was.

This isn't going to be one of those "here's how great my gear is" posts. I want to talk about failures, because I've had a few real ones. Four years of doing this, and I've racked up three major mistakes worth sharing. They happened in 2018, 2019, and 2023. Direct costs north of $1,100, plus delays and embarrassment that took the real total higher. Each mistake taught me something that I still use every single week.

Mistake #1: The SICK DBS60 Encoder, or Why Pinouts Are Non-Negotiable

In 2018, I'd just unboxed my Keithley DMM6500 bench digital multimeter. I'd saved up for months. The 6.5-digit resolution was serious overkill for the repair work I was mostly doing, but I wanted it anyway. I was that guy.

A week later, a machine builder dropped off four SICK DBS60 encoders and asked me to verify the outputs before they went into a bottling line.

That should've been straightforward. Power them up, count pulses, check quadrature. But the DBS60 series comes in a bunch of variants—push-pull, HTL, TTL—and I didn't confirm which one I had before wiring it. In my defense, it looked like a standard M12 connector. I'd wired plenty of those. So I hooked up power and measurement leads based on how I assumed it worked.

About twelve seconds later, the encoder made a quiet, sad pop. The DMM6500's input protection saved the meter itself, but the encoder was dead. A $245 mistake. Plus the look on my client's face when I told him what happened.

Here's the part that still stings: the correct pinout was in the datasheet. I had the datasheet open on my screen. I just didn't check it because I assumed it was standard. To be fair, the DMM6500 did exactly what I asked it to do—it measured. I measured the wrong pins with the wrong expectations, and destroyed a component in the process.

The production line was waiting, and I had maybe twenty minutes to verify those encoders before the client's deadline. Twenty minutes. Normally I'd triple-check the wiring, but there was no time. I went with my gut. That's exactly how it happened.

That experience created the first rule on my checklist: verify the pinout from the datasheet before connecting anything. Not "if you have time." Not "if it's a new part." Every time.

Mistake #2: Weighing Scale Load Cells, or Know What You're Actually Measuring

By 2019, I'd added a Keithley LCR meter to the bench. That same year, I landed a contract to verify load cells that were going into industrial weighing scales.

A load cell is basically a strain gauge bridge. To verify it properly, you measure input and output resistance, bridge balance, and sometimes insulation resistance. I knew this. I'd read about it. I was confident.

I set up the LCR meter, picked 1 kHz as the test frequency—seemed reasonable enough—and worked through a batch of twenty cells. The numbers landed within a few ohms of nominal. I wrote the report and signed off.

Then came the actual calibration run. The complete weighing system was off by about 4%. Four percent on a scale that's supposed to meet Class III accuracy isn't "close enough." It's a fail.

Here's what I got wrong: my test frequency. The LCR meter at 1 kHz was measuring reactive effects in the load cell's equivalent circuit that don't matter at the DC or low-frequency excitation the scale actually uses. I was measuring something—the readings were real—but not the thing that mattered.

That mistake cost my client a week of delay and me roughly $900 in retesting. The credibility damage stung even worse. A mentor of mine summed it up: "You had the right instrument. You asked it the wrong question." That's still the best description of that whole disaster.

The second rule on my checklist: understand the device's real-world operation before choosing a measurement setup. The Keithley LCR meter can measure accurately across its entire frequency range. It's my job to pick the setting that's meaningful for the specific device.

Mistake #3: The Zeiss vs Global Dental Microscope Request, or Why I Almost Turned Down My Favorite Project

In 2023, a dental equipment distributor emailed me. Their customers were complaining about inconsistent light output in two dental microscope brands they carried—Zeiss and Global. They wanted to understand the electrical difference between them.

My first instinct was to decline. I'm not a dental equipment specialist, and I told them so. Lenses, ergonomics, optical quality? Not my lane. I can't speak to any of that.

But I asked one follow-up question: "What exactly is failing?"

The answer turned out to be electrical. These microscopes use LED light sources driven by switching regulators, and if the driver has poor transient behavior, you get flicker and brightness drift. You don't need an optics lab to measure that. You need a multimeter that can log current over time and catch fast transients.

So I took the job. The distributor shipped me one of each microscope, and I spent a week profiling their power behavior. Using the Keithley DMM6500's digitizing and logging features, I measured current draw on both units during warm-up, under load, and over extended operation.

The results were genuinely useful. The Zeiss unit drew a turn-on current spike roughly 40% higher than the Global unit's peak. Not enough to trip a breaker, but enough to stress the LEDs and possibly the driver circuit over time. The Global unit went the other direction: lower peak current, but a longer settling time before the current stabilized.

Both designs had trade-offs. But the conversation changed completely once the distributor had the actual curves. Instead of relying on marketing claims about Zeiss vs Global dental microscope quality, they showed their customers measured data and let them decide.

That's the right way to handle a comparison. I didn't declare a winner, and neither did they. We gave people real information and trusted them to make a smart choice.

If you ask me, that's what customer education should look like. I'd rather spend ten minutes explaining what the numbers mean than deal with a disappointed client six months later. An informed customer asks better questions and makes faster decisions. The dental microscope project was the clearest example of that principle I've ever experienced.

To be clear about my limits: I demonstrated measurable electrical differences in how the two microscope brands handle power. But which one has better optics or more comfortable ergonomics? That's outside my expertise. A dentist's hands and eyes are the real instruments there, and I'd recommend a clinical evaluation before choosing.

The Checklist That Finally Stopped the Bleeding

After the load cell incident, I started writing things down. My checklist now lives in a physical notebook, and it has exactly three rules:

1. Verify pinouts before connecting. Datasheet first. Always.
2. Understand the device's real-world operation. Is the LCR test frequency meaningful for this circuit? Is this a bridge? What are the actual operating conditions?
3. If a project sounds outside your lane, ask whether the core question is measurable. If it is, take the job. You can learn the rest.

In the past eighteen months, I've caught 47 potential errors using this checklist. That number sounds like bragging, but it's not meant that way. The point is that I was bad before, and a simple process fixed most of it.

I'd argue that most test equipment problems aren't equipment problems. They're setup problems. Granted, that's not always true—I've seen a defective probe and a misbehaving fixture. But more often than not, the person between the chair and the bench is the weakest link.

If you're starting out with a Keithley DMM6500 or a Keithley LCR meter, you already have instruments that are more accurate than you'll need on most days. They'll measure almost anything you ask them to, and they'll do it honestly. If your data looks wrong, the instrument is not lying. It's telling you that something about your setup is wrong.

Personally, I wouldn't trade my DMM6500 for a cheaper alternative. Not because the cheaper meters can't measure—they can—but because I've learned to trust this one. After you've tested an expensive component the wrong way and watched it pop, that trust is worth a lot.