Most engineers already know to reach for a true RMS meter when working with non-sinusoidal waveforms. What far fewer know is that “True RMS” on the label doesn’t guarantee accurate readings on every signal your facility generates. An averaging meter can read 40% low on a VFD output. An unfiltered true RMS meter can read 20–30% *high* on that same signal. Both instruments are technically working correctly. Both will mislead you.
This guide covers how the two measurement approaches actually work, where each one fails, and — most importantly — what the spec sheet leaves out about the instrument you already own.
Key Takeaways:
- Average-responding meters assume a pure sine wave and read up to 40% low or 10% high on distorted waveforms.
- A true RMS meter *without* a low-pass filter over-reads VFD PWM output by 20–30% by capturing the high-frequency carrier.
- Crest factor — not just meter type — determines real-world accuracy. Most meters de-rate their CF spec at lower input levels.
- Power quality distortion costs U.S. industry $15–24 billion annually (EPRI/CEIDS), and wrong meter selection is a direct contributor.
Understanding these limitations is critical when choosing the right instrument. Explore a range of precision test equipment designed for accurate measurement across distorted and non-linear signals.
What’s the Actual Difference Between True RMS and Average Responding Multimeters?
Average-responding meters measure what they say: the rectified average of the AC waveform. Then they multiply that average by 1.11 — a constant derived from the ratio of RMS to average for a pure 60 Hz sine wave — to estimate an RMS value. It’s a fast, inexpensive computation that was accurate enough in the era when power lines delivered genuinely sinusoidal current.
True RMS meters do the math directly. They sample the waveform instantaneously, square each sample, average those squares over time, and take the square root of the result. That’s the full mathematical definition of root-mean-square — no waveform-shape assumptions, no fixed constants, no implicit dependence on the signal being a sine wave.
For a clean sine wave, both methods produce identical output. That’s the entire basis of the averaging shortcut. The problem is that modern industrial loads rarely produce a clean sine wave — and the 1.11 multiplier doesn’t know that.
| Category | True RMS Meter | Average-Responding Meter |
|---|---|---|
| Measurement method | Sample → square → average → root | Rectify → average → × 1.11 |
| Accurate on sine waves | ✓ Yes | ✓ Yes |
| Accurate on distorted waveforms | ✓ Yes (within CF and bandwidth spec) | ✗ No — assumes sine shape |
| Accurate on VFD PWM output | ⚠ Only with low-pass filter (see below) | ✗ Reads 10–40% low |
| Crest factor rated | Yes (typically CF 3.0–5.0) | No (assumes CF 1.414) |
| Typical price premium | +$30–$150 over comparable averaging model | Baseline |
| Best for | VFDs, switching supplies, non-linear loads, power quality | Resistive loads, regulated DC circuits, basic continuity |
How Average-Responding Meters Fail on Real Waveforms
The problem lives in the 1.11 multiplier. It’s derived from the mathematical relationship between RMS and average for one specific waveform shape: a pure sine wave. The moment the waveform deviates from that shape, the ratio changes — but the meter keeps using the same constant, regardless.
How far off can it get? In 2025, Fluke’s technical documentation documented that an average-responding meter can read up to 40% low or 10% high on distorted waveforms. The direction and magnitude depend on whether the waveform carries more energy near its peaks — which drives a high read — or near its flat regions, which drives a low read. Either way, the meter doesn’t know it’s wrong. It reports a confident, stable number.
| Waveform Type | Typical Error (Average-Responding Meter) | Notes |
|---|---|---|
| Pure 60 Hz Sine (reference) | 0% | Baseline (accurate) |
| Triangular wave (CF 1.73) | ~4% | Slight under-reading |
| Half-wave rectified sine | ~11% | Noticeable error |
| PC / IT equipment (CF 2–3) | 15–20% | Common non-linear loads |
| VFD PWM output (typical) | 10–40% | Highly variable, often significant under-reading |
| Switching PSU, no PFC (CF >3) | Up to 40% | High crest factor causes large errors |
| Highly distorted loads (CF 3–5) | Up to 50% | Worst-case scenarios |
According to Fluke’s VFD measurement documentation (February 2025), an averaging meter on a VFD output typically reads 10–40% low. That’s not a calibration offset you can account for. A 25% under-read means you conclude a motor is seeing 300V when it’s actually receiving 375V — or you clear a fault that genuinely exists because the reading looks within range.
In 2025, Fluke’s technical documentation confirmed that average-responding multimeters can read up to 40% low or 10% high on non-sinusoidal waveforms, because they multiply the rectified average by a constant (1.11) derived from a pure sine wave. When the waveform shape changes — as it does with VFDs, switching power supplies, and non-linear loads — the constant becomes wrong and the error becomes systematic.
What the Spec Sheet Doesn’t Tell You About “True RMS”
So you’ve bought a true RMS meter. The box says TRUE RMS. Problem solved?
Not quite. Here’s what the spec sheet leaves out.
The low-pass filter problem on VFD outputs
Variable frequency drives generate PWM output — a high-frequency switching signal that encodes the fundamental frequency inside a series of rapid pulses. A true RMS meter measures everything it captures: the fundamental and the carrier frequency. On a typical VFD output, Fluke’s own technical data shows that an unfiltered true RMS meter reads **20–30% higher** than the actual fundamental voltage. That’s the opposite direction of an averaging meter’s error — and it’s equally misleading.
The fix is a meter with a built-in low-pass filter (LPF) that strips the carrier frequency before the RMS computation runs. The Fluke 87V’s LPF mode exists precisely for this application. Without it, you’re choosing between two wrong answers: the averaging meter that reads low and the unfiltered true RMS meter that reads high.
The crest factor de-rating problem
Meter specs list a maximum crest factor — commonly CF 3.0 or CF 5.0 — but almost none state that this rating applies only at the meter’s full-scale input range. Drop to half-scale, and the effective crest factor headroom typically falls with it. A meter rated CF 6 at full scale may only handle CF 3 when the input is at 50% of range. If you’re measuring a 120V signal on a 600V range — a common practice for margin — you’re at 20% of scale, and your actual CF limit is well below the number on the data sheet.
The bandwidth problem
RMS accuracy on harmonic-rich waveforms requires the meter to respond faithfully up through the highest significant harmonic component. A 6-pulse VFD generates substantial current at the 5th harmonic (300 Hz), 7th (420 Hz), 11th (660 Hz), and 13th (780 Hz). If the meter’s frequency response rolls off before those frequencies, it under-counts harmonic energy — even as a “True RMS” instrument. Bandwidth specs matter, not just the RMS label.
When waveform shape and harmonic content matter, instruments with sufficient bandwidth are essential. Advanced oscilloscopes allow engineers to visualize and analyze high-frequency components that standard meters may miss.
Crest Factor: The Number That Exposes the Problem
Crest factor is the ratio of a waveform’s peak value to its RMS value. For a pure sine wave, it’s always 1.414. When a waveform distorts — when current gets pulled in sharp spikes rather than smooth curves — the crest factor rises. The higher it climbs, the more an averaging meter’s fixed 1.11 multiplier breaks down.
According to EC&M’s technical analysis common industrial loads sit well above the sine-wave baseline:
| Load Type | Crest Factor (CF) | Notes |
|---|---|---|
| Resistive heater (pure sine) | 1.41 | Baseline sine wave |
| PFC-corrected equipment | ~1.45 | Near-sinusoidal waveform |
| Fluorescent lamp (electronic ballast) | 1.5–2.0 | Moderate distortion |
| VFD (6-pulse, standard) | 2.5–3.5 | Often near meter CF limits |
| PC / IT equipment (no PFC) | 2.0–3.0 | Common non-linear load |
| Switching PSU (no PFC) | >3.0 | Exceeds many meter limits |
| Extreme non-linear loads | Up to 5.0 | High crest factor, challenging to measure |
A standard 6-pulse VFD — the most common drive topology in industrial facilities — sits at crest factor 2.5–3.5 on the input side. A switching power supply without power factor correction runs above CF 3.0 consistently. These aren’t niche edge cases. They’re what engineers measure every day in motor control, automation, and power electronics environments.
What does this mean in practice? For Keysight true RMS instruments, published crest factor error tables show additional measurement uncertainty of 0.15% at CF 2–3, rising to 0.40% at CF 4–5. Those figures apply to true RMS instruments at full scale. Averaging meters have no crest factor specification at all — they don’t acknowledge that the problem exists.
Where These Errors Show Up in Real Facilities
The mismeasurement problem isn’t theoretical. It’s happening on facility floors, in test labs, and in field service applications where technicians assume their meter is telling the truth.
These environments require reliable instrumentation. Explore electrical test equipment built for industrial diagnostics, troubleshooting, and validation.
Variable frequency drives
VFDs are everywhere: HVAC air handlers, conveyor systems, pump stations, CNC machines, robotic motion systems. They’re also among the worst waveform offenders. In April 2026, Industrial Monitor Direct published analysis showing that a standard 6-pulse rectifier VFD generates input current with total harmonic distortion (THD-i) of approximately 30% under typical field conditions, with the 5th harmonic alone running at roughly 20% of the fundamental current.
Put an averaging meter on that drive’s output and you’re likely reading 10–40% low. Put an unfiltered true RMS meter on it and you’re reading 20–30% high. Neither instrument tells you what the motor is actually receiving.
For VFD applications, using the right instrument is critical. Specialized true RMS multimeters and power analyzers with low-pass filtering help isolate the fundamental signal from high-frequency switching noise.
Switching power supplies
Every modern control system, PLC, servo drive, and embedded computer runs off a switching power supply. They draw current in pulses, generating crest factors consistently above 3.0. An averaging meter treating that pulse current as sinusoidal will under-read — which means load balance calculations, branch circuit loading figures, and overcurrent protection sizing are all built on incorrect current measurements.
Grid-level power quality
In November 2024, Whisker Labs published monitoring data showing that 38% of homes on the ComEd (Chicago) grid exceeded the IEEE 519 voltage THD limit of 8% Whisker Labs . That’s the supply voltage alone — before facility loads add their own distortion on top. Industrial facilities running VFDs, arc welders, UPS systems, and unfiltered rectifiers compound an already-distorted grid. If your measurement tools can’t capture the actual waveform content, your power quality baseline is unreliable.
The Real Cost of Measuring Wrong
Wrong measurements don’t just produce bad data. They produce bad decisions — and bad decisions in electrical systems carry a quantifiable financial cost.
Power quality phenomena — the category that includes harmonic distortion — cost U.S. industry an estimated $15–24 billion annually, based on EPRI/CEIDS research covering 985 establishments (VECTO System, citing Lawrence Berkeley National Laboratory data). That’s separate from the $104–$164 billion attributable to outages. The mechanisms are direct: in April 2026, Energy Control Systems documented that high harmonic levels increase transformer losses by 10–15% annually and raise motor operating temperatures 15–20°C above normal. Every 10°C rise in winding temperature cuts motor insulation life in half.
If harmonics go undetected and uncorrected, motors and transformers face a 30–50% reduction in operating lifespan. For facilities running process-critical equipment, that’s a capital replacement budget problem and a production schedule problem — not just a maintenance footnote.
| Harmonic Order | Frequency (Hz) | % of Fundamental Current | Notes |
|---|---|---|---|
| 5th | 300 Hz | 20% | Dominant harmonic |
| 7th | 420 Hz | 14.2% | Significant contribution |
| 11th | 660 Hz | 9.1% | Moderate level |
| 13th | 780 Hz | 7.7% | Moderate level |
| 17th | 1020 Hz | 5.9% | Lower contribution |
| 19th | 1140 Hz | 5.3% | Lower contribution |
Who Needs Which Meter?
The answer isn’t simply “always buy true RMS.” It’s more precise than that — and more nuanced than the label implies.
When processing calibration records and measurement audits across customers in electronics manufacturing, power systems, defense, and automotive test environments, one pattern repeats consistently: mismeasurement problems are almost never caused by engineers who deliberately chose the wrong meter type. They’re caused by field technicians using whatever meter is in the bag, on signals they didn’t expect to be non-sinusoidal.
True RMS with low-pass filter capability is the minimum for these applications:
- Motor drive and VFD commissioning, troubleshooting, and output verification
- Power quality analysis and THD baseline documentation
- Switching power supply design validation or production test
- UPS system load testing and battery runtime verification
- Semiconductor fabrication power monitoring (process tools draw highly non-linear current)
- Aerospace and defense test environments where power quality affects avionics and guidance system performance
Averaging meters remain appropriate when:
- Testing purely resistive DC circuits and battery voltages
- Verifying transformer ratios or regulated supply rails under clean, controlled conditions
- Measuring utility mains voltage at a residential service entry where distortion is low and 5–10% accuracy is acceptable
The calibration dimension
Whichever meter type you use, calibration status matters independently of RMS methodology. A true RMS meter with an expired calibration certificate may read worse than a freshly calibrated averaging meter on the same sine-wave signal. RMS measurement capability and metrological traceability are separate requirements — and both matter.
Frequently Asked Questions
Can an averaging multimeter damage equipment through misreading?
Not directly — a meter doesn’t influence the circuit it measures. The risk is indirect: a 40% under-read on distorted current leads you to conclude a circuit is operating normally when it’s overloaded. That false clearance delays corrective action, allowing harmonics to continue stressing insulation and increasing transformer losses by 10–15% annually.
If my meter says “True RMS,” does it always read VFD output correctly?
No. True RMS meters without a low-pass filter over-read VFD PWM output voltage by 20–30% because they capture the high-frequency carrier alongside the fundamental. You need a meter with a selectable low-pass filter function specifically designed for drive output measurement — the Fluke 87V’s LPF mode is the standard reference for this.
What crest factor rating should I look for in an industrial true RMS meter?
For VFDs and switching power supplies, CF 3.0 at full scale is a baseline minimum. But verify whether the spec de-rates at lower input levels — many instruments rated CF 6.0 at full scale drop to CF 3.0 at half scale. Always check your actual operating input level against the CF spec at that scale position.
Does meter bandwidth affect true RMS accuracy on harmonic loads?
Yes, significantly. A 6-pulse VFD generates current at the 11th harmonic (660 Hz) and 13th harmonic (780 Hz) at levels of 9.1% and 7.7% of fundamental respectively. If your meter’s bandwidth rolls off before those frequencies, it under-counts harmonic energy — even as a true RMS instrument. For power quality work, verify bandwidth specs before selecting a meter.
How often should a true RMS multimeter be calibrated?
Most manufacturers specify annual calibration for instruments in industrial measurement environments. The interval should reflect the instrument’s published drift specification, usage conditions, and the accuracy tolerance of the measurements. An instrument past its calibration date carries unknown uncertainty regardless of its measurement architecture — and no amount of “True RMS” circuitry compensates for a lapsed calibration certificate.
Conclusion
The true RMS vs. average-responding distinction is real and consequential. But buying the right label isn’t the whole answer. The spec sheet tells you which computation method the meter uses. It doesn’t tell you whether the low-pass filter you need for VFD work is present, whether the crest factor rating applies at your actual input level, or whether the bandwidth is sufficient for the harmonic orders your loads generate.
