When a measurement suddenly looks wrong, don't start with the repair ticket. Start with the setup. After 12 years and more than 200 urgent measurement calls—many of them same-day turnarounds for semiconductor, process, and utility customers—my own call records say roughly 70% of the so-called broken instruments I was asked to fix were not broken at all. The meter was fine. The pressure transmitter was fine. The real fault was a loose shield wire, a blocked impulse line, a fixed-emissivity misunderstanding, or a unit conversion nobody had double-checked. This article is the field manual I wish every customer had before calling me.
Why I distrust 'the instrument is broken'
It took me about three years and 150-odd field calls to learn something that still sounds backward to newer engineers: the more precise your instrument is, the easier it is to fool. A 7½-digit Keithley 2001 multimeter will happily show you errors that a cheap handheld meter never even notices—thermal voltages from a warm connector, ground-loop currents, or a guard wire that isn't actually guarding anything. The instrument isn't lying. It's telling you the truth about a sloppy setup.
I remember one March 2024 case well. A lab called 36 hours before a scheduled audit with a Keithley 2001 multimeter that kept drifting on low-current readings. The unit had just come back from calibration, so the engineer was ready to send it out again. I asked one question: what was the shield wire on his test fixture connected to? It had come loose at the connector. After twenty minutes and a fresh termination, the 2001 held its reading, the audit was saved, and the customer avoided a several-thousand-dollar repair bill for a meter that had never been sick.
That story isn't rare. Last quarter alone we processed 47 rush jobs with a 95% on-time rate, and the same pattern keeps showing up. Before you spend money on recalibration or replacement, run through the four cases below.
The four instruments that generate the most false alarms
1. Precision multimeters (Keithley 2000, 2001, 2010, 2110, and DMM6500)
A good Keithley multimeter does not fail as often as people think. What fails is the measurement path. If your readings are unstable, out of range, or just different, check the boring stuff first:
- Warm-up. According to Keithley product manuals, high-resolution multimeters need time to settle before they meet their published accuracy spec. I typically allow an hour—check your model's manual for the exact requirement. I've seen meters 'drift' for 20 minutes and then behave perfectly.
- Input setup. Front vs. rear input, 2-wire vs. 4-wire, or the wrong current jack can change the reading exactly like a hardware fault.
- Cables and connectors. If readings move when you wiggle a cable, you've found the problem. Test leads are the weakest link in any precision setup. Swap them first. It's a $20 fix, not a $4,000 repair.
- Guard and shield. On low-current and high-resistance measurements, an open or miswired guard connection creates exactly the slow drift that sends Keithley 2001s and 2010s to the repair bench.
Verify model-specific requirements in the free documentation on Keithley's official site (tek.com/keithley, accessed April 2025).
2. Pressure transmitters (4–20 mA loop)
Pressure transmitter calls usually start with 'it reads high.' The first questions should not be about the transmitter. Is the isolation valve open? Is the impulse line blocked, frozen, or full of moisture? Is the sensing diaphragm clean? A transmitter measures whatever pressure is actually at its diaphragm. If the line to the diaphragm is plugged, it is measuring trapped pressure—not the process. That isn't a transmitter failure; that's a plumbing failure.
If the process side is clean, check the loop itself. For a standard 4–20 mA transmitter, measure the loop current with a good multimeter and convert: pressure = (measured current − 4 mA) ÷ 16 mA × full-scale range. A 0–100 psi transmitter sending 12 mA should read 50 psi. If the current says 50% but the control screen says 80%, the transmitter may be fine. The scaling or configuration is wrong.
3. The compact 62 Mini IR thermometer and its cousins
Every electrical toolbag has one: the compact 62 Mini IR thermometer or something shaped like it. It's fast, but it's also one of the most misapplied instruments I know. Most of these pocket units assume a fixed emissivity around 0.95. That works well on painted, taped, or matte surfaces. It does not work on shiny metal, which can read far below its true temperature.
People also forget the distance-to-spot relationship. A typical 8:1 unit at 24 inches measures an area about 3 inches across. If you aim it at a 1-inch-wide pipe from three feet away, you're measuring whatever is around the pipe too. That's not a bad sensor; that's bad geometry.
The quick fix for shiny surfaces: put a piece of matte black tape or flat-black paint on the target, let it equalize for a minute, and measure that. In one pump-room call, the IR gun said 160°F on a stainless line while a contact sensor said 220°F. The team suspected a failing pump. After we taped the pipe, the gun agreed with the contact sensor within a few degrees. Both instruments had been correct all along—they were just looking at different surfaces.
4. How to read a Sensus digital water meter (without breaking anything)
I've also taken calls from facilities that start with a billing question: how do I read a Sensus digital water meter? Usually the display shows total consumption. The reason people misread these meters isn't the LCD; it's the units.
- Open the meter lid carefully. Don't cut or break any utility seal.
- Find the digital display. It may need a moment to light up or cycle through screens.
- Write down every digit, including leading zeros and the decimal point.
- Check the unit label on or near the display. If it says cubic feet and your bill is in gallons, multiply by 7.48. Always compare the same units before assuming something is wrong.
- Check for a flow or leak indicator while nothing is running. If it keeps moving, you have a leak—not necessarily a bad meter.
A silent toilet flapper can add 50 to 200 gallons a day. One 'impossible' high water bill I investigated turned out to be exactly that. The Sensus meter was doing its job. Nobody wanted to believe the toilet was the culprit. Exact display behavior varies by model, so confirm the details in the official Sensus/Xylem manual for your specific meter.
The 20-minute drill I use before suspecting hardware
When I'm triaging a rush case, I don't start with the instrument. I run a simple order: source first, then path, then setup, then instrument. First, is the input actually what you think it is? Second, is the signal path sound—cable, connectors, guard, valves, impulse line? Third, are the range, units, and scaling correct? Fourth, is the environment causing trouble—heat, EMI, vibration, moisture? Only after those four do I allow myself to suspect the instrument itself.
That drill catches the causes behind most of my 70% figure. It also keeps me honest, because sometimes the instrument really is dead.
When it really is the instrument
Now the flip side. Some readings really are hardware failures. A multimeter that was connected to an overvoltage, a pressure transmitter with a ruptured diaphragm, a cracked IR lens, or a physically damaged water meter—no amount of setup checking will fix those. If the instrument fails against a known-good source, or if it has physical damage, stop using it and send it to a qualified repair or calibration service. For Keithley products, Keithley's official calibration and service channels are the right starting point. For other brands, use the manufacturer's own service network.
Looking back at the calls I mishandled early in my career, the expensive ones were almost always the ones where I trusted an initial assumption. So here's my final, unglamorous advice: check the boring stuff first. The broken wire, the unit setting, the blocked line. Your Keithley multimeter, pressure transmitter, IR thermometer, or water meter has been trying to tell you the truth all along. Give it a chance to prove it before you replace it.
