A technical question that comes up now and then and deserves a clear answer: «my Shelly EM (the Gen1) reads noticeably less than the utility meter, especially with the house idling». Almost always the wiring is fine and the real culprit is something else: measurement at very low current. Let's walk through it with a real case, because the diagnosis is more interesting than it looks.
The clue is in the reactive power
The customer who wrote in had almost solved it himself. These are his live values, single-phase, with the 50 A clamp on the main line:
- At idle (~188 W, 1.4 A): reactive ≈ −247 var, power factor ≈ 0.61.
- With the oven (~2164 W, ~10 A): reactive ≈ −318 var, power factor ≈ 0.99.
Notice one detail: the current rises seven-fold, yet the reactive power barely moves (from 247 to 318 var). If it were a phase error, the reactive would grow with the current (it would reach ~1700 var at 10 A). Since it stays almost fixed, it isn't a phase error: it's an offset. The meter «sees» about 1 A of reactive current that isn't really there.
At 10 A that offset is negligible (that's why the power factor reads 0.99 and the reading is good). But at 1.4 A it eats most of the measurement, drags the power factor down and makes active power read too low. And since your house —server, router, network gear— sits at 1-1.5 A all day, the daily total ends up ~35 % below the utility meter.
Why it fails precisely at low current
Two things meet in the worst possible place:
You're at 3 % of the clamp's range. 1.4 A on a clamp built for 50 A is the bottom of the scale. There, any clamp's relative error is largest: the noise and offset of the measurement circuit itself weigh as much as the signal you're trying to read.
And your baseline is electronic. A server, a router or a switch use switch-mode power supplies: they don't draw a clean sine current, but short spikes with harmonics and a low or leading power factor —that's why your reactive reads negative—. A simple meter measures that pulsed current worse, and at low load it really shows. Low current plus an electronic load is, literally, the worst case for a 50 A clamp.
To be clear, because it's everyone's first suspicion: your wiring is correct. Clamp on the live, the meter powered from that same phase, and the arrow toward the loads. It's not a wiring fault or a reversed clamp: it's the physical limit of measuring 1 A with a clamp designed for 50.
Can you calibrate the Gen1?
Here the answer is straightforward: the EM Gen1 has no phase or offset calibration, not in the web UI nor the API. That's why you can't find the option: it doesn't exist.
The calibration methods that appear in Shelly's documentation (the PhaseToPhaseCalib ones) belong to the Pro / Gen2 meters, not the EM Gen1. The only DIY route is to reflash third-party firmware (Tasmota or ESPHome), but that calibrates gain with a purely resistive load —it doesn't cancel the low-current reactive offset— and it voids the warranty. And updating the official firmware won't fix it either: it isn't a software bug, it's a hardware limit of the clamp at low current.
How to measure your electronic baseline well
Answering your third question, from least to most:
- Set expectations. The utility meter is the legal reference and will always win at low load. The Shelly EM is a consumption monitor, not a certified billing meter: it's for spotting trends and automating, not for matching your bill to the watt.
- Move up to a Pro meter. For a whole single-phase home, the Shelly Pro EM-50 (2×50 A) is Class B, measures in 4 quadrants and does support calibration. Its measurement front-end behaves far better at low current and with non-linear loads than the Gen1. Note: no 50 A-clamp meter is billing-grade when reading 1 A, but the step up in quality is real.
- Meter the rack at the socket. If what you want is to quantify your electronic baseline accurately, a metered plug like the Plug S Gen3 measures the server and the network at the outlet, accurate down to a few watts, without fighting a clamp's bottom of scale.
The combination that works best in your case: keep the EM for the whole-home picture and add a metered plug at the rack. That way you get the fine figure for your electronic baseline and the overall view, each where it measures well.
FAQ
Is it normal for the EM Gen1 to under-read at low current?
Yes. At 1-1.5 A you're at 2-3 % of the 50 A clamp's range, where relative error is largest and a reactive offset dominates the reading. Electronic loads make it worse. At medium and high currents the same unit measures well.
Why does my reactive read negative?
The switch-mode supplies in electronic gear often present a leading (capacitive) power factor, and the meter's low-current offset adds to that. The combination gives that almost-constant negative reactive.
Can I correct it from the app or API?
Not on the Gen1: it exposes no phase or offset calibration. That option lives on the Pro/Gen2 meters, not this one.
Would a different clamp solve it?
The problem isn't the clamp itself, but measuring 1 A on one built for 50. Shelly doesn't offer a much lower-range clamp for the EM, and switching to a third-party CT leaves you without factory calibration. It's more effective to meter that load at the socket or move to a Pro meter.
Will the Pro EM-50 match the utility meter to the watt?
It'll get much closer, especially at low load, but the utility meter remains the legal reference. For home monitoring, the Pro EM-50 is a clear step up in accuracy.
And if my installation were three-phase?
The equivalent would be the Shelly Pro 3EM, also Class B and with calibration. In your single-phase case, the Pro EM-50 is the right fit.
Summary
- Under-reading at low load is expected on the EM Gen1: a reactive offset dominates when you work at 2-3 % of the clamp's range.
- It isn't a wiring or phase error, and the Gen1 has no user calibration to correct it.
- For accuracy at low load: meter the rack with a metered plug, or move to a Pro EM-50 (Class B, 4-quadrant, calibration).
- The utility meter is and remains the billing reference.
Got numbers that don't add up? Send them over (V, A, W, var and power factor at idle and under load) and we'll tell you whether it's expected or whether another meter fits better. Technical support and shipping available for your market.
Leave a comment