I review test procedures for a living. Roughly 200 unique products pass through our lab each year, and my job is to make sure the measurement setup—not just the instrument—can be trusted. In Q2 2024, we rejected 8,000 units of a supposedly tested-OK product. The vendor's paperwork was perfect. Calibration certificates were current. Test reports were signed and stamped.
The parts were still bad.
That failure cost us around $22,000 in rework and delayed a launch by three weeks. The root cause wasn't lazy workmanship or a broken fixture. It was a selection problem: someone bought the right brand of instrument for the wrong measurement at the wrong point in the process.
This article is about what that failure—and several others—taught me. None of it is about bashing a brand. It's about a mistake I see engineers make over and over: treating the instrument as the answer instead of the measurement as the answer.
The Surface Problem: We Buy by Brand Instinct
Ask a room of engineers what to buy for electrical testing, and you'll hear names. Fluke. Keysight. Keithley. Tektronix. Then everyone nods and orders from the familiar online store.
I get it. These are genuinely strong brands. A Fluke multimeter is a field workhorse. Keithley's bench instruments are the reference in low-level DC measurement. If you're a working engineer thinking "I just need a reliable multimeter," brand recognition is a reasonable shortcut.
But here's what I've observed in practice: the same brand instinct that gets you the right tool for one job gets you the wrong tool for another.
Last year, a new engineer joined our team. His task: verify contact resistance in a new connector design. He walked over to the bench, picked up one of our Keithley multimeters—a good one, honestly—and started measuring. Readings were stable. Everything looked fine. When I reviewed his setup, I asked what test current he was using. He pointed at the DMM's range setting.
The problem: he was measuring a milliohm-level contact with a test current that was far too low, using a 2-wire measurement instead of a 4-wire Kelvin configuration. The instrument was fine. The measurement approach was wrong. The data looked perfect—until we re-ran it on a proper micro-ohmmeter and the values shifted by 18%.
Now, that's an easy fix once you know. But it illustrates the real issue: when people ask "which multimeter should I buy?" they're asking the wrong question first. The right question is: what exactly am I measuring, and at what uncertainty?
The Deeper Problem: Spec Sheets Don't Tell the Whole Story
Here's something vendors won't tell you: a spec sheet describes what an instrument can do under ideal conditions. It does not describe what happens inside your actual measurement scenario.
Take what people mean when they say "6.5 digit multimeter." A Keithley 2010 or 2000 delivers 6.5-digit resolution. Most engineers read that as "extremely accurate." And it is—when used properly. But resolution is not accuracy. Accuracy is not precision. And neither tells you your real-world uncertainty unless you account for the test leads, the thermal EMF at the connection points, the input impedance of the device under test, and the calibration history of the instrument.
Measurements are a chain. The instrument is one link. Test leads are two more links. The connector you're probing is another. Fixture, temperature, humidity, ground loops—each adds a small error. If any link is weak, your instrument was never the thing that saved you.
I only fully believed this after ignoring it myself. In 2022, I specified a new bench DMM for a critical lot of reference boards. I read the datasheet. I checked the specs. Excellent. I approved the purchase. Then, on a whim, I ran a comparative check against an older unit already in the lab—same boards, same leads, same room. The readings differed by 0.013%. Within spec for both instruments. But our production requirement was tighter than that, and the old unit was drifting because no one had planned an interim calibration between its annual cycles.
That decision cost us a week of validation work and a tense call with production.
What Wrong Tool Selection Actually Costs
Let me be direct about what bad instrument selection costs beyond your budget.
That $22,000 batch failure? It happened because a supplier used a bench multimeter to verify a dimensional feature. I won't name them. What matters is that a bore micrometer—a precision instrument designed specifically for measuring internal diameters—was the right tool. A DMM, even a high-resolution one, measures electrical parameters. It cannot see dimensional variation. The part tolerance was ±0.01 mm. The supplier shipped parts with dimensions outside that tolerance, and no electrical measurement would ever catch it.
I remember being frustrated, but honestly, I also understood. Engineers reach for the instrument they have on hand, not the instrument the measurement requires. It's a constraint of habit, not malice.
Then there was the thermal case. A prototype power supply would intermittently overheat. On the bench, everything looked normal: voltages correct, currents correct, no obvious failures. We were close to shipping it. Then a colleague brought in a one pro LT thermal imaging camera—the FLIR model that attaches to a phone. I'd dismissed it as a toy. But when we scanned the board, there it was: a hot spot at the gate driver running 40°C above the rest of the board. No multimeter would ever have caught that. We fixed the layout and the problem disappeared.
The thermal camera didn't replace the multimeter. It answered a different question entirely. But the team's instinct—which was calibrated around "buy a good multimeter"—would have shipped a defective unit.
The Question Nobody Asks First
So what's the actual problem? It's not brand. It's not price. It's not even instrument quality.
It's that people choose an instrument before they decide what they need to learn about their product.
A bore micrometer answers one question: what is the physical diameter of this hole? A multimeter answers another: what is the voltage, current, or resistance at this point? A thermal camera answers a third: where is heat being generated, and is the pattern normal?
These instruments aren't substitutes. They're different tools for different questions. And yet I regularly see teams pick equipment by asking "what brand should I get?" instead of "what do I need to measure, at what uncertainty, in what environment?"
What Actually Works
The approach that has served us well in recent years:
- Write the measurement requirement first. Not the brand. Not the budget. What parameter, what range, what tolerance, under what conditions. One page of requirements.
- Build an uncertainty budget. Don't buy a 6.5-digit instrument because it sounds impressive. Buy it because your uncertainty calculation says you need it.
- Pick the instrument class that answers the question. Electrical? A DMM or SMU. Dimensional? A bore micrometer. Thermal? A thermal imaging camera.
- Only then choose the brand. At this point, brand matters for calibration support, software ecosystem, and long-term reliability.
For my team, this is where Keithley comes in. Not because "Keithley is the best" in an absolute sense, but because when we define a requirement like "measure 1 pA with a noise floor under 10 fA," Keithley is one of the few brands that can document that performance repeatably. That's why Keithley multimeters—the model 2010, 2000, and DMM6500—are common in standards laboratories and semiconductor characterization. It's not branding. It's a demonstrated ability to hold specifications through calibration cycles.
And for those Googling "where to buy fluke multimeter"? That's still the wrong question. Fluke makes excellent field multimeters, and you can order one from plenty of reputable distributors. But be sure you're buying it because it fits your measurement requirement—not because it's the default answer everyone gives.
Bottom Line
The best instrument in the world won't help you if it's answering the wrong question.
I still kick myself for the assumptions I made in 2022. If I'd challenged our tool selection earlier, we'd have saved five figures and a lot of stress. But I also know now that the fix isn't buying a better brand. It's deciding what you need to measure, at what uncertainty, before you open the catalog.
Trust the instrument that matches the measurement. Not the brand that matches your habit.
