What True RMS Actually Changes

Why this matters

True RMS is sold as a general upgrade and it is not one. On a clean sine wave an averaging meter and a true-RMS meter agree exactly, by design, and the extra capability buys you nothing. On the current feeding a variable-speed drive, an LED driver bank or a switch-mode power supply, the same two instruments can differ by enough to reverse your conclusion about whether a machine is overloaded. What decides it is not the instrument, it is the shape of the waveform you happen to be pointing at, and that shape can be benign for one quantity and severe for another at the same set of terminals.

Before either of these readings

Both readings below are taken live, at a panel, and sit behind the energized-work gate at 29 CFR 1910.333(a)(1), which requires de-energizing before work on or near live parts unless that is infeasible and treats a test that can only be performed energized as one of those cases. Instrument, leads and tips must all be rated at or above the measurement category and voltage of that location under IEC 61010-1, which binds through the product's listing, and under 29 CFR 1910.334(c)(3), which binds you. Wear the electrical protective equipment 29 CFR 1910.335(a) requires for the exposure, work inside the arc-flash boundary and PPE your employer's electrical safety program sets under NFPA 70E-2021 in the edition it adopts, keep the free hand out of the enclosure, and stand out of the line of the opening.

The current reading below is a clamp reading, which needs one conductor alone in the jaw. If the conductors are bundled and you would have to separate them, that is no longer a test that can only be done energized: de-energize, lock out under 29 CFR 1910.333(b)(2), or 29 CFR 1926.417 on construction, and prove dead with the live-dead-live sequence in NFPA 70E-2021, 120.5 before separating them.

What RMS actually means

Root mean square is not a mathematical nicety, it is the heating-equivalent value. The RMS current in a conductor is the DC current that would heat that conductor the same amount. That is why every thermal rating, every conductor ampacity, every motor nameplate and every overload device is written in RMS terms: heating is what damages things, and RMS is the number that predicts heating.

So when a meter reports a current that is not the RMS value, it is not slightly imprecise. It is reporting a quantity that does not correspond to the thing the equipment ratings are about.

How the two meter types get their number

A true-RMS instrument computes the square, the mean and the root, so it produces the heating-equivalent value regardless of shape.

An averaging-responding, RMS-calibrated instrument rectifies the signal, measures the average of the rectified value, then multiplies by a fixed constant. That constant is the form factor of a sine wave: for a sine, the RMS value is 0.7071 of peak and the rectified mean is 0.6366 of peak, so the ratio is 1.111. Multiply the measured mean by 1.111 and you get the correct RMS value for a sine, exactly.

Which is the whole story. The averaging instrument is not approximate on a sine, it is exact. It is exact because the constant was chosen for that one shape, and it is wrong on every other shape by a predictable factor.

The gate, and the formula that predicts the error

Form factor is RMS divided by rectified mean, and it is a property of the waveform. Then for an averaging-responding instrument:

Displayed value divided by the true RMS value = 1.111 divided by the waveform's true form factor.

Read the direction off that. A peaky waveform, current drawn in short high pulses by a rectifier front end, has a form factor above 1.111, so the ratio is less than one and the averaging meter reads low. A flat-topped waveform has a form factor below 1.111, so the meter reads high: on an ideal square wave the form factor is 1.000 and the averaging meter reads 11.1 percent above the truth. Both errors are real and the low one is not universal, which is worth holding onto, because "averaging meters read low on distorted waveforms" is the half-remembered version that leaves a tech unable to explain a reading that came in high.

The gate: choose the meter type per measured quantity, not per circuit and not per instrument. If the quantity is shaped by a nonlinear load, true RMS is the only instrument that answers the question. If the waveform is an undistorted sine, either instrument is exact and the choice does not matter.

What true RMS does not fix

Three limits, all of which live in the instrument's own specification rather than in a general rule:

• Crest factor limit. Crest factor is peak divided by RMS; a sine is 1.414 and distorted currents run higher. Every true-RMS instrument has a specified maximum crest factor, above which its own accuracy degrades or it clips. A waveform peakier than the specification is not measured correctly by a true-RMS meter either.
• Bandwidth. Distortion means harmonic content at multiples of the fundamental. An instrument's rated bandwidth sets how much of that content it can see, and energy above the bandwidth is simply absent from the calculation. A true-RMS meter with a narrow bandwidth still under-reports a harmonic-rich current.
• The DC component. Most AC ranges are AC-coupled and deliberately exclude any DC offset. Where a circuit carries both, you need an instrument that offers an AC-plus-DC RMS function, and you need to select it.

None of these are reasons to prefer an averaging meter. They are reasons to read the specification rather than the marking on the case.

One gate, two outcomes: the same panel, the same minute

A drive-fed motor on a 480 V branch. Nameplate current on the drive's input is 14.0 A. The tech has an averaging-responding instrument in hand.

Outcome one: the voltage reading. Supply voltage at a panel is normally close to sinusoidal, because the source impedance ahead of it is low and the distortion drawn by loads downstream shows up far more in current than in voltage. Its form factor is essentially the sine's 1.111, so the ratio 1.111 divided by 1.111 is 1.000 and the averaging instrument is exact. It reads 478 V; a true-RMS instrument on the same terminals reads 477 V, and the one-volt difference is instrument specification, not waveform. The instrument in hand is fit for this reading.

Outcome two: the current reading. The drive's input current is drawn in short pulses as the rectifier charges its bus. Model it as a rectangular pulse train conducting for 30 percent of the cycle, which is an idealization of that current and not a measurement of it, but it lets the arithmetic close. For a rectangular pulse train of duty fraction D and peak Ip, the RMS value is Ip times the square root of D and the rectified mean is Ip times D, so the form factor is one over the square root of D. At D of 0.30, that is one over 0.5477, or 1.826.

Say the peak is 30 A. True RMS is 30 times 0.5477, or 16.4 A. Rectified mean is 30 times 0.30, or 9.0 A. The averaging instrument multiplies that mean by its fixed 1.111 and displays 10.0 A.

Now apply the gate's formula as a check: 1.111 divided by 1.826 is 0.609, and 0.609 times 16.4 A is 10.0 A. The two routes agree, which is what you want before you trust either.

The two conclusions. At 10.0 A against a 14.0 A nameplate, the drive is at 71 percent of rating and there is nothing to report. At 16.4 A against the same nameplate, it is 17 percent over, which is a finding: the conductors and the protective device are being heated by 16.4 A regardless of what the display says, because heating follows RMS. Same terminals, same minute, same instrument, and it was fit for the voltage and unfit for the current.

A qualification on the true-RMS number too. The crest factor of that waveform is 30 divided by 16.4, or 1.83. That is inside the crest factor limit of ordinary true-RMS field instruments, but it is a value to check against your own instrument's specification rather than assume, and a more sharply peaked front end will go higher.

What would flip this. A resistive load. Strip heaters, incandescent lighting, a resistance element: the current follows the sinusoidal voltage, the form factor is 1.111, and the averaging instrument is exact on the current too. The distinction is not the size of the load or the voltage of the system, it is whether the load draws current in proportion to the instantaneous voltage.

The failure mode. A shop with averaging instruments surveys a building full of electronic loads and finds every circuit comfortably loaded. Nothing in the readings is erratic, nothing is obviously wrong, and the numbers are all reproducible, because a wrong constant applied consistently produces consistently wrong numbers. The evidence that something is off does not come from the meter, it comes from conductors and terminations running hotter than the recorded currents can explain. When measured current and observed heating disagree, the measurement is the thing to doubt.

How to verify you got this right

• Name the load before you name the instrument. Rectifier front end, switch-mode supply, electronic ballast, drive, dimmer: true RMS. Motor across the line, heater, incandescent: either.
• Read both quantities the same way and compare. If an averaging and a true-RMS instrument agree on voltage and disagree on current at the same terminals, that is the expected result, not a fault in either one.
• Compute the implied form factor when two instruments disagree. Divide 1.111 by the ratio of the displayed value to the true-RMS value. If the answer is near 1.111, the difference was not waveform and you should look at the instruments.
• Check the crest factor against your instrument's specification before quoting a true-RMS number on a badly distorted current.
• Confirm the coupling. If the circuit can carry a DC component and your range is AC-coupled, the reading excludes it by design.

References

• 29 CFR 1910.333(a)(1) and 1910.335(a) - the energized-work gate and required electrical protective equipment for a live panel reading
• 29 CFR 1910.334(c)(3) - the requirement that instruments be rated for the circuits they are connected to
• 29 CFR 1910.333(b)(2), 29 CFR 1926.417 and NFPA 70E-2021, 120.5 - electrical lockout in general industry and construction and the live-dead-live proving sequence, binding through the employer's electrical safety program
• IEC 61010-1 - measurement category and voltage ratings, binding through the product listing
• Instrument manufacturer specification - crest factor limit, bandwidth, and whether the range is AC-coupled or AC-plus-DC