48 Hours to Go – And My Multimeter Was Lying to Me
Last month, a client called with a classic emergency: a capacitive sensor prototype needed full characterization data in 48 hours, or they’d trigger a $50,000 penalty clause. I grabbed a trusty Fluke 87V off the shelf—the same one I’d used for years to check voltage on control panels. Confident, I started measuring the sensor’s capacitance drift. Four hours later, the numbers were all over the place. Seriously all over the place—variations of ±15 pF when the spec called for ±2 pF tolerance.
I blamed the sensor at first. Then I blamed the wiring. I even blamed my afternoon coffee. But the real culprit? My “good enough” multimeter.
The Real Problem Isn’t What You Think
Here’s the thing most engineers miss: a standard DMM is designed for general voltage, current, and resistance checks. It’s built to tell you if 12 V is really 12 V, not to resolve pico-Farad changes against a background of electrical noise. The Fluke 87V has a capacitance range with a resolution of 1 pF (0.1 nF in some modes). For a sensor that changes by 0.5 pF, that’s like measuring rainfall with a bucket—you catch rain, but you can’t tell how much.
What I didn’t realize at the time: the real issue was noise floor and auto‑ranging hysteresis. The multimeter kept switching ranges because the signal hovered right at the boundary, introducing errors that looked like sensor drift. Let me rephrase that: the instrument itself was creating the problem I was trying to measure. (Should mention: I later verified this with a proper LCR meter—zero drift in the sensor.)
People think expensive precision instruments are a luxury. The reality? Cheap instruments are the luxury you can’t afford when deadlines matter. The causation runs the other way: you don’t buy Keithley because you have money; you buy Keithley because you don’t have time to waste.
The Cost of “Good Enough” – A Quick Calculation
Let’s put a number on it. That first failed test cost me:
- 6 hours of wasted testing and troubleshooting → $600 at my shop rate
- 1 hour on the phone with the client explaining delays → eroded trust (hard to price)
- 2 hours driving to a colleague’s lab to borrow a Keithley 2000 DMM with proper shielding → $40 gas + 2 hours lost
- $80 in rush shipping for a Keithley 2000 multimeter software license (I needed the logging capability) from a local distributor
Total non‑recurring cost for using the “cheap” meter: $720 + 9 hours. The Keithley 2000 DMM itself costs around $1,200 new. So I burned 60% of its purchase price in one failed test. That’s the real TCO story—your labor and deadline risk can easily dwarf the instrument cost.
I still kick myself for not doing this math sooner. If I’d bought the Keithley unit six months earlier, I’d have saved two previous rush jobs that went nearly as badly.
The Fix: Keithley Precision + Software = No‑Brainer
Once I had the Keithley 2000 on the bench, the measurement became trivial. The 6½‑digit resolution gave me stable, repeatable capacitance readings down to 0.01 pF. The Keithley 2000 multimeter software (free with the instrument, by the way) let me log data at 1‑second intervals and export directly to CSV. The client approved the sensor data within 30 minutes of my second test run.
Bottom line: when you’re racing a deadline, precision is not a feature—it’s a survival tool. The Keithley 2700 multimeter (same core measurement engine, more channels) would have worked equally well. And for capacitance‑sensitive work, the Keithley LCR meters are a game‑changer.
Oh, and I should add: the Fluke 87V is still a great meter for voltage checks on the production floor. It’s just not the right tool for high‑precision low‑level measurements. That’s not a criticism of Fluke—it’s a category difference. But when you’re trying to test a capacitive sensor for a high‑value client, you need the right tool for the job. The total cost of choosing the wrong one is way more than the price difference.
Prices as of April 2025; verify current rates. My story is real; names and penalty clauses are masked to protect the innocent (and the guilty).
