Last August, a senior tech on our team closed a ticket on a 3000W inverter generator installation. His report looked clean: 119.8V at the generator terminals, 119.6V at the transfer switch, continuity confirmed on the control circuit. Two weeks later the customer called back. The ATS was still dropping loads during outages.
I pulled the report during our Q3 2024 review. The readings were real. The meter wasn't broken. What was missing was context. He'd used an averaging meter on an inverter output, and the number on the screen wasn't telling him what he thought it was telling him.
That single ticket turned into a $1,100 controller replacement and about a week of callback time. Not because anyone was careless. Because the test method—not the tool, the method—had a gap nobody had written down.
The reading looks fine. That's the problem.
Most of us learn early that a multimeter reads volts. Then we spend years trusting that number without asking a harder question: volts according to what math?
Here's the part that doesn't get taught enough. A standard averaging multimeter rectifies the AC waveform, takes the average, and scales the result to display what the RMS value would be if the waveform were a clean sine wave. On a clean sine, that's accurate. On anything else—inverter output, VFD terminals, dimmer circuits, a generator carrying a reactive load—the display drifts from reality. Sometimes by 20%. Sometimes more.
From the outside, a 119.6V reading looks like proof the circuit is healthy. The reality is it might be the average of a waveform that's anything but clean.
Three places this shows up in real service work
1. Inverter generator output under load
A 3000W inverter generator with a clean resistive load usually holds THD under a few percent, and most averaging meters will read close enough. Add a reactive load—a compressor, a pump motor, a battery charger—and the waveform distorts. The meter still shows a steady number. The number just isn't the true RMS voltage anymore.
If you're using that reading to decide whether the generator is ready for transfer, you're making a decision on bad data. A True RMS meter—the Fluke 114 true RMS multimeter is the one we spec for general electrical work—reads the actual heating-equivalent RMS value regardless of waveform shape. According to Fluke's published specifications (fluke.com, accessed January 2025), the 114 is rated CAT III 600V with a LoZ mode and VChek, which matters for the next two cases.
2. Contactor coil testing and ghost voltage
Here's a scenario I've seen three times in the last eighteen months. A tech isolates a contactor coil circuit, puts a high-impedance meter across the coil terminals, and reads 78V. The circuit is supposedly dead. The coil is supposedly bad. So they replace the coil.
What they actually measured was capacitive coupling—ghost voltage from an adjacent energized conductor. The high input impedance of a standard meter (typically 10MΩ) is enough to complete a circuit through stray capacitance and produce a phantom reading. A LoZ mode drops the input impedance to around 3kΩ and collapses that reading to near zero on a dead circuit. The coil was probably fine both times.
This is a measurement-method problem, not a parts problem. And it's expensive because it hides inside a clean-looking report.
3. ATS commissioning and transfer timing
Most techs understand how an automatic transfer switch works—it monitors utility voltage, starts the generator on loss, waits for stabilization, transfers the load, then reverses the sequence when utility returns. What gets missed is that every step in that sequence is a voltage decision. If the voltage sensor is fed by a distorted waveform or the control circuit is carrying ghost voltage, the switch does the wrong thing at the right time, or the right thing at the wrong time.
On a recent commissioning, our tech reported 119.4V on the utility sense line and 119.3V on the generator sense line—within 0.1V, so the controller should have transferred cleanly. It didn't. The actual issue was that the utility sense line was reading a distorted signal from a nearby VFD, and the averaging meter smoothed it into something that looked normal. A True RMS meter showed 124.7V on that same line. That's a 5V gap, and it was enough to keep the controller from arming the transfer sequence correctly.
What this actually costs
I'm not going to pretend this is a crisis in every shop. To be fair, averaging meters have their place—on clean utility-fed circuits, they're accurate and they're cheap, and for a lot of routine checks that's enough. I get why people run them.
But when the reading feeds a decision that has a dollar figure attached, the gap matters. In our shop, the numbers from 2024 look like this:
- Three unnecessary contactor coil replacements: about $340 in parts, plus four hours of labor
- One ATS controller swap that didn't solve the problem: $1,100 in parts, plus a callback visit
- One generator commissioning that passed on paper and failed in the field: $600 in return-trip labor
That's roughly $2,000 in 2024 on problems that a $150 meter upgrade and a one-page test procedure would have caught. The math isn't complicated.
And the number that doesn't show up on an invoice: the customer who called back after a "clean" commissioning. That one costs you the next job.
I knew we should have standardized meter models per job type a year before we actually did it. I thought, "techs know their own tools." That was the quarter we ate the $1,100 controller swap.
What actually fixed it
We didn't replace every meter in the shop. That would have been an overreaction, and honestly a budget fight I wasn't going to win. What we did instead was narrower:
First, we mapped meters to job types. General troubleshooting stays on the existing averaging meters. Anything that touches inverter output, VFD terminals, or ATS sensing lines runs on a True RMS meter. The Fluke 114 is what we spec for most of that work—no current measurement, but that's rarely the gap. Our automotive bay techs run a Fluke mechanic multimeter for the dwell and duty-cycle side of the same principle.
Second, we added a line to the test report template. Every reading now records the meter model and the waveform condition. That sounds bureaucratic, and it is. But when a callback happens, I can tell within two minutes whether the reading was trustworthy or not. Before, that took an hour of guessing.
Third, we added a LoZ check to every dead-circuit verification. It takes about four seconds. It has killed at least two ghost-voltage false alarms since we started doing it.
That's it. No new testing philosophy, no six-figure equipment budget. Just closing the gap between what the meter displays and what the circuit is actually doing.
The part I'm still not sure about
I'm not 100% sure this fully solves the ATS sensing-line problem. True RMS gets you a better number, but it doesn't get you the waveform shape or the transient behavior. If your controllers are tripping on transients that a handheld meter never sees, you need a scope or a power quality analyzer, and that's a different conversation.
This worked for us, but our situation is a mid-size electrical contractor doing mostly commercial and light industrial service. If you're running a high-volume residential operation with clean utility power and no inverter load, the calculus is probably different. Your mileage may vary.
What I can say with confidence: if your test reports show clean readings on inverter output, contactor coil circuits, or ATS sensing lines, and the equipment still misbehaves, the meter isn't lying. It's answering a different question than the one you asked. Knowing which question you're actually asking is the whole job.