A 12 V, 2 A wall adapter draws 24 W when the product on the end of it is working and something between 0.05 W and 0.5 W when the product is switched off or unplugged. That second number is what a customs broker, a retailer's compliance desk or a state energy regulator will ask about, and it is the one most buyers never read. Wall adapter no load power consumption and average efficiency are the two figures that decide whether a design can legally be sold in the United States and the European Union at all, and they sit on the datasheet next to each other for a reason: they are measured in the same test, against the same regulation, and they fail for different reasons.

This article explains what each number means, which limit applies to which wattage, how the test is run, why a design can pass one figure and fail the other, and what to write into your purchase specification so the sample you approve is the adapter that ships.

What the two numbers measure

No-load power is the input power the adapter draws from the wall with nothing connected to its output. The output is open-circuit; the primary side is still switching, the control IC is still biased, and the X-capacitor across the mains still needs discharging. Every one of those consumes energy, and the regulations cap the total.

Average active efficiency is the arithmetic mean of efficiency measured at four output loads — 25%, 50%, 75% and 100% of nameplate output current — with the adapter warmed up and the input at nominal mains voltage. That is why datasheets call it "four-point" average efficiency. A fifth point, efficiency at 10% of rated load, was added by the EU regulation that took effect in 2020 and is a separate pass/fail line.

Both figures are functions of the nameplate output power, not of what the connected product consumes. A 12 V, 1.5 A adapter is an 18 W adapter for regulatory purposes even if it spends its life powering a 6 W router. Over-specifying the adapter moves it up the efficiency curve and, at the 49 W boundary, into a different no-load band.

Which limits apply, and where

The two regulations a U.S. or EU buyer will meet are the U.S. Department of Energy standard for external power supplies, universally called DoE Level VI (10 CFR 430.32(w), mandatory for products manufactured on or after 10 February 2016), and the EU Ecodesign regulation (EU) 2019/1782, which replaced Regulation 278/2009 from 1 April 2020 and is what people mean by "ErP" in an adapter context.

The table shows the limits for the most common case — a single-voltage-output adapter with nominal output of 6 V or more ("basic-voltage" in DoE language). Low-voltage supplies (below 6 V at 550 mA or more, which catches most 5 V USB chargers) have their own slightly lower efficiency lines. Confirm the formula against the current published text before copying it into a contract.

Nameplate output P (W) DoE Level VI average efficiency (min) DoE Level VI no-load (max) EU 2019/1782 average efficiency (min) EU 2019/1782 efficiency at 10% load (min) EU 2019/1782 no-load (max)
P ≤ 1 0.5·P + 0.16 0.100 W 0.5·P + 0.16 0.5·P + 0.06 0.10 W
1 < P ≤ 49 0.071·ln(P) − 0.0014·P + 0.67 0.100 W 0.071·ln(P) − 0.0014·P + 0.67 0.071·ln(P) − 0.00115·P + 0.57 0.10 W
49 < P ≤ 250 0.880 0.210 W 0.880 0.79 0.21 W
P > 250 0.875 0.500 W not in scope — —

Two things to notice. First, the four-point average lines are identical between the U.S. and EU for single-voltage supplies — deliberately — which is why a competent factory builds one design for both markets. Second, the EU adds the 10%-load line and the U.S. does not. A design tuned to squeak past the four-point average with a controller that only becomes efficient above 20% load can carry a valid DoE Level VI mark and still fail an EU test report. If you sell on both sides of the Atlantic, ask for the 10% figure even on U.S.-only stock.

Worked example: a 12 V, 3 A adapter (36 W). The four-point average minimum is 0.071·ln(36) − 0.0014·36 + 0.67 ≈ 0.254 − 0.050 + 0.67 ≈ 0.874, or 87.4%. The 10%-load minimum under the EU text is 0.071·ln(36) − 0.00115·36 + 0.57 ≈ 0.783. No-load must stay under 0.10 W. A datasheet that says "efficiency ≥ 85%" for that adapter is telling you it does not meet either regulation, however good the rest of the sheet looks.

California's Title 20 rules align with the DoE standard but add a listing requirement in the CEC appliance database; Canada's regulations are harmonised with DoE Level VI; the UK carried the 2019/1782 technical requirements into its own ecodesign rules, with the same numbers and its own conformity route.

How the test is actually run — and why that changes the result

The test procedure (10 CFR 430 Appendix Z in the U.S.; IEC 62301 measurement practice referenced by the EU text) has details that decide whether the number on the datasheet means anything:

  • Input voltage. DoE tests at 115 V, 60 Hz. The EU tests at 230 V, 50 Hz. A universal-input adapter is a different circuit at those two voltages: the switching losses in the primary MOSFET rise with input voltage, and the X-capacitor's discharge resistor dissipates four times as much at 230 V as at 115 V. A no-load figure quoted without the input voltage is close to meaningless, and a figure measured at 115 V will typically be lower than the same unit at 230 V. When a supplier gives you one number, ask which voltage.
  • Warm-up and stabilisation. The unit runs at full load until thermally stable before efficiency points are recorded, and the no-load reading is taken over a defined averaging window. A factory that reads efficiency two minutes after plug-in on a cold unit is reporting a number that will drift down in the lab.
  • Adaptive and multi-voltage supplies. USB-C Power Delivery chargers that negotiate 5 V to 20 V are tested differently from a fixed 12 V adapter: the DoE procedure tests adaptive supplies at their highest and lowest nameplate voltages and requires compliance at both, and the EU text has its own treatment of multiple-voltage outputs. Ask the test lab, not the factory, which condition applies to your PD charger.
  • Instrumentation. No-load power at 0.08 W with a high current crest factor needs a power analyser rated for it; a general-purpose bench meter reads low. If the report does not name the instrument, treat the number as indicative.

Where the milliwatts go: what an engineer sees in a no-load budget

A 0.100 W no-load ceiling is a tight budget once you list what has to fit inside it. These are the items an engineer trades off, and they explain why two adapters with the same rating can differ by a factor of three at no-load:

  1. X-capacitor discharge. Safety standards require the capacitor across the mains input to discharge to a safe voltage within a fixed time after the plug is pulled. The cheap way is a bleeder resistor. A 1 MΩ bleeder across 230 V dissipates roughly V²/R = 0.053 W — more than half the entire no-load budget, before the converter has switched once. The better way is a discharge IC that only conducts when the mains is gone. Which one the design uses is not on the datasheet; it is on the schematic, and it is a fair question.
  2. Controller start-up and bias. Older designs bled the controller supply continuously from the high-voltage rail through a start-up resistor; modern controllers use a high-voltage start-up switch that turns off once the auxiliary winding takes over, and drop into burst mode at no-load so the transformer switches only a few times per second.
  3. Feedback network. The optocoupler and shunt regulator on the secondary side draw a steady bias current even at no-load, typically a few milliamps at the output voltage. At 12 V, 3 mA is 0.036 W. Designs that hit 0.05 W at no-load are the ones that have starved this network deliberately.
  4. Indicator LED. A power-on LED at 2 mA from a 12 V rail through a resistor costs 0.024 W — a quarter of the budget for a light.

The four-point average has a different set of levers: synchronous rectification on the secondary instead of a Schottky diode (worth several percentage points at 5 V outputs, less at 20 V), transformer winding design, MOSFET on-resistance, and the choice between fixed-frequency, quasi-resonant or active-clamp flyback topology. Those choices show up in the bill of materials and in the price, which is the honest reason a compliant adapter costs more than a non-compliant one that looks identical.

Reading a datasheet or test report critically

When a supplier sends you a datasheet, check these lines before anything else:

Datasheet line What a good entry looks like What a weak entry looks like
Efficiency "Average active efficiency ≥ 87.5% at 115 V and 230 V; 10% load ≥ 79%" "Efficiency ≥ 85%" or "High efficiency"
No-load power "≤ 0.075 W at 230 V AC" "Low standby" or a number without a voltage
Compliance "DoE Level VI (10 CFR 430.32(w)); EU 2019/1782" "Energy Star" (not applicable to EPS since 2010) or "Level V"
Test condition Warm-up stated, instrument named No test condition given
Marking Roman numeral VI on the rating label artwork No efficiency marking on label

The Roman numeral point matters. The DoE standard requires the efficiency level to be marked on the nameplate of the external power supply itself as "VI", and retailer compliance teams know to look for it. A label proof with no numeral is an early warning that the design was never tested to the standard, or that the label was drawn by someone who did not know. Get the label artwork at the sample stage. For the wider set of marks a label may need to carry, CE, FCC and product certification for China-made goods covers what each mark does and does not prove.

Then ask for the test report behind the datasheet. A report from a recognised lab lists the sample serial number, input voltage and frequency, the four load points with measured input and output power, the calculated average, the 10% point if the EU text was applied, the no-load reading with its averaging window, and the instrument model. A factory's in-house report should show the same structure; if it only shows a pass/fail column, it is not evidence.

What changes between the wattage tiers

Manufacturers that cover a wide range of ratings do so across several platforms, and the efficiency behaviour changes with them. Lexiang Power, a Shenzhen-headquartered maker with its plant in Yichang, lists its wall-plug adapters in tiers of 1–6 W, 9–18 W, 12–24 W, 24–42 W and 45–65 W, with desktop bricks above that up to 240 W, according to its CMH page. Whatever supplier you use, treat each tier as a separate compliance question:

  • 1–6 W. No-load is the hard constraint; the fixed overheads (X-cap discharge, controller bias) are the same size as the load losses. Ask for the no-load figure first.
  • 9–24 W. The efficiency line climbs steeply through this band (from about 82% at 9 W to about 86% at 24 W). Synchronous rectification starts to matter for 5 V and 9 V outputs.
  • 24–65 W. The average line flattens toward 88%; the 49 W boundary raises the no-load allowance to 0.21 W. A "65 W" design that is really 60 W with margin is tested against the 60 W nameplate.
  • 90 W and above. Desktop supplies are held to 0.88 average and 0.21 W no-load under both regimes. Bricks with a constant-current charging mode behave differently at no-load and at the 25% point than a fixed-output brick, one reason the two modes are treated separately in power mode versus charging mode in desktop adapters.

Interchangeable-plug kits: the measurement is on the converter, not the plug head, so one body with U.S., EU, UK and AU heads needs one report at each mains voltage, not one per head — provided the report actually covers both 115 V and 230 V.

Writing it into the purchase specification

Put the following into the spec sheet or purchase order so the sample and the production lot are held to the same numbers. The structure in how to write a product spec sheet for a Chinese factory is a good frame; these are the adapter-specific lines to put inside it.

  1. Nameplate output voltage and current, and the resulting nameplate power. State it; do not let the factory infer it from a part number.
  2. "Average active efficiency ≥ [value for your P] measured per 10 CFR 430 Appendix Z at 115 V 60 Hz and per (EU) 2019/1782 at 230 V 50 Hz." Fill the number in yourself from the table above.
  3. "Efficiency at 10% of rated load ≥ [value]" if the goods will be sold in the EU or UK.
  4. "No-load input power ≤ [0.10 W or 0.21 W] at both 115 V and 230 V." For margin, write 0.075 W; many compliant designs achieve it.
  5. "Rating label to carry the efficiency mark VI; label artwork to be approved before pilot run."
  6. "Supplier to provide a test report (third-party or in-house, stated which) showing the four load points, the 10% point, no-load reading, input voltage and instrument, for the golden sample and for one unit per production lot on request."
  7. Incoming inspection: no-load measurement on a sample of units with a suitable power analyser; reject the lot if any unit exceeds the limit.

Line 6 is the one buyers skip. An adapter that meets the standard as a golden sample and drifts over it in production — a cheaper X-cap discharge part substituted, an optocoupler with higher bias current — is a real failure mode, and only a lot-level report catches it.

What to ask the supplier next

  • Which input voltage was the quoted no-load figure measured at, and what is the figure at the other one?
  • What is the efficiency at 10% of rated load for this model, and was it measured or calculated?
  • Is the X-capacitor discharged by a resistor or by a discharge IC? What is the measured no-load contribution of the feedback network?
  • Does the rating label artwork carry the VI mark, and can you send the label proof now?
  • Is there a third-party test report for this exact model and nameplate rating, and under which legal entity was it issued? If only an in-house report exists, can it be released before order?
  • For an interchangeable-plug kit: does the efficiency report cover both 115 V and 230 V, and is the same converter used under every plug head?
  • Will you supply a lot-level efficiency and no-load record with each shipment, and at what sampling rate?

Lexiang Power's CMH listing is a useful test case: it names an in-house lab with automated burn-in racks, in-process combined testers and a high/low-temperature chamber, and a wall-adapter range from 1 W to 65 W across five tiers, but publishes no measured no-load or efficiency figures per tier and does not say whether lab reports are released pre-order. Those are the gaps to close in the first email — read the factory profile and send the seven questions above with your nameplate rating attached.