A smart plug that reports watts feels like a measurement instrument. It is closer to a well-behaved guess. For a resistive heater the guess is good. For the devices people actually want to measure, phone chargers, standby loads, LED drivers and anything with a switching supply, the number can be wrong by enough to change the conclusion you draw from it.
What the metering chip actually measures
A plug metering chip samples voltage and current, multiplies them, and integrates over time. In a load where voltage and current are sinusoidal and in phase, that arithmetic gives real power in watts, and the result is accurate.
Mains voltage is close enough to sinusoidal. Current is not, once the load contains a rectifier.
A switching supply draws current in short pulses near the voltage peak rather than smoothly across the cycle. The current waveform is spiky and it is not in phase with the voltage. Cheap metering front ends sample slowly, and a spiky waveform sampled slowly loses most of its shape. What remains is an approximation that can sit well below the true figure.
This is why the same plug reads a heater accurately and a laptop charger badly. The heater is a resistor. The charger is not.
Power factor is the part people skip
Real power in watts is what you are billed for as a domestic customer. Apparent power in volt-amperes is voltage multiplied by current without the phase correction. The ratio between them is the power factor.
A load with a poor power factor draws more current than its wattage suggests. The plug measures current, and if it reports a figure derived from current alone, it overstates. If it tries to compute real power without proper phase handling, it understates. Both errors come from the same missing term.
Switching supplies have power factors that vary with load. A charger at full output behaves differently from the same charger idling with nothing attached. So the error is not a fixed offset you can subtract. It moves with the operating point, which is what makes the reading so hard to trust.
The resolution floor hides standby draws
Metering chips have a minimum current they can resolve. Below it they report zero.
Standby consumption is often below that floor. A device drawing a fraction of a watt through a plug with a one watt resolution reads as nothing at all. The display says zero, the device is drawing power, and the user concludes the device is free to leave plugged in.
Multiply that across a house and the conclusion is wrong in the direction that matters. The whole point of measuring standby is that the numbers are small, and small numbers are exactly where low-cost metering gives up.
This is also why summing plug readings rarely reconciles with the utility meter. Each small load rounds toward zero, the errors accumulate in one direction, and the total comes out low.
Calibration is done at one point
A metering plug is calibrated during manufacture against a known load, usually a resistive one at a moderate power level. Accuracy is specified near that point and degrades away from it.
That is a sensible way to build a cheap product. It does mean the stated accuracy applies to conditions similar to the calibration conditions, and a plug specified at plus or minus two percent near a hundred watts may be considerably worse at five watts with a poor power factor.
Vendors who publish a curve rather than a single number are telling you something useful. Vendors who publish one number are quoting the best case their chip can achieve, which is usually a resistive load at the middle of the range.
How to use the reading without fooling yourself
Use plug metering to rank loads, not to account for them. A plug that says the heater is the biggest thing in the room and the phone charger is small is correct even if both numbers are off. Relative comparisons survive the error. Absolute totals do not.
Measure at the wall for a full day rather than reading a number once. Cycling loads, thermostats and compressors vary, and a single sample tells you almost nothing about a fridge.
For anything below a few watts, plug metering will under-read. If you need the real figure, a meter designed for standby measurement is a different instrument with a different current range, and no smart plug replaces it.
Do not use a plug to sanity-check your electricity bill. The bill comes from a revenue-grade meter with a defined accuracy class. A consumer plug is not in that class, and a mismatch between them is expected rather than alarming.
For sourcing, the useful questions are which metering chip is used, whether it does true RMS on both channels, what the specified accuracy is at low current rather than at mid range, and whether the vendor publishes error against load. A plug built around a proper energy metering IC costs a little more and produces numbers you can reason about.
The smart plug manufacturer range specifies metering components per model, including which parts do true RMS and what accuracy holds at low current. Where the requirement is Tuya ecosystem compatibility, the Tuya smart plug range covers the same metering question within the Smart Life platform.
If the goal is whole-home measurement rather than per-appliance ranking, the smart switch board range with per-gang energy monitoring measures at the plate instead of the socket. For volume programmes where metering accuracy is a specification rather than a feature, XingLiao manufacturing can confirm chip selection and calibration method per model.
The short version
A plug reports real power computed from sampled waveforms. Spiky loads, poor power factor and a resolution floor all push that number away from the truth, and the worst errors land on the small standby loads people most want to measure. Use the readings to rank, not to total, and check what chip is inside before you trust a specification.
Top comments (0)