Countdown to 1 November: GB 46519's 9-point weighted efficiency rule is hardest on transformers and inductors

On 5 October 2025, China's State Administration for Market Regulation published GB 46519-2025, 'Minimum allowable values of energy efficiency and energy efficiency grades for electric vehicle supplying equipment' - the country's first mandatory energy-efficiency standard for charging infrastructure, effective 1 November 2026. Only days remain. Most industry discussion has centred on power modules and topology selection, but reading the clauses closely reveals something else: what the 9-point weighted method really re-prices is the transformer and the inductor. This article moves from the test method to the loss structure, and back to the retrofit paths inside the grace period.

1. A mandatory red line enters its countdown

GB 46519-2025 covers all non-vehicle conductive supplying equipment - integrated DC chargers, split DC chargers and AC chargers. Its core content is three things: efficiency grades, minimum allowable values, and the test method. The timeline is fixed: published 5 October 2025, mandatory from 1 November 2026; products shipped before that date may be deferred to 1 November 2027, but must complete upgrading within that year.

The standard sets three efficiency grades for DC equipment: grade 3 is the market-admission floor, energy-saving certification is issued only for grades 1 and 2, and grade 3 products are excluded from the certification system. Official estimates put the national electricity saving at about 1 billion kWh per year once fully implemented. Scale is what gives this weight: China has 12.818 million charging installations in place delivering over 110 billion kWh a year - every 0.1 percentage point of efficiency is an enormous amount of power.

2. The 9-point weighted method: from a peak figure to a full-conditions report card

The most disruptive part of this standard is not the efficiency numbers but the assessment method. The traditional 'peak efficiency' tests one optimal operating point; the 9-point weighted method requires testing at 3 voltage platforms (400V / 600V / 800V) × 3 load points (20% / 50% / 100%) - nine operating points in total, each for at least 5 minutes, with the final efficiency computed as a weighted average (Figure 1).

Test conditions are strict too: ambient temperature 25°C ± 5°C; power analysers with relative error no more than 0.05% and resolution 0.1W for charging efficiency; maximum permissible error 0.05W and resolution 0.01W for standby measurement. Standby power is measured after the unit rests for 3 minutes in standby mode, for at least 30 minutes, by integrating energy. The implication in one sentence: a headline '97% peak efficiency' optimised at a single operating point loses its meaning against a 9-point report card - the unit must be stable across the whole voltage and load range, and any weak cell drags the total down.

Why design it this way? Because real operations never sit at rated voltage and rated power: daytime fast charging concentrates at high power, nights and hot summers bring derating, and different vehicle battery platforms keep switching. The 'ideal condition' behind a single-point figure covers only a fraction of the real operating curve; the 9-point matrix effectively compresses the statistical distribution of real duty into one test table, aligning laboratory numbers with operating bills for the first time.

3. How high is the bar: a gap of 0.5-1.5 percentage points

The numerical bar looks small; the engineering bar is large. For integrated DC chargers: grade 1 requires charging efficiency of at least 96.5% with standby power no more than 30W; grade 2, 95.5%; grade 3, the admission floor, 94.5% (Figure 2). Split DC chargers get a 48W standby allowance; AC chargers are not graded and only face a standby limit of 7.0W.

The problem is the installed base. A large share of legacy chargers in service run at 93-94% efficiency - still 0.5 to 1.5 points short of the admission line. This is precisely the segment that peak-efficiency figures hide: many units can claim 96%+ at their laboratory peak, yet fall below the line once weighted over 9 points. Procurement has moved first: tender documents across regions increasingly write efficiency grades into bid thresholds, and after Bull Group took the industry's first 'grade 1 energy-efficiency' certification (certificate no. 001, issued by CVC) in October, 'grade 1' is becoming the new language of competition.

The arithmetic is tangible for operators too. On a simplified assumption of one 120kW DC charger running at full power 10 hours a day, each percentage point of efficiency saves roughly 4,380 kWh a year - across dozens of chargers per site, two or three points is real money.

4. Why magnetics: the loss structure flips at light load

The hardest of the nine points is not 800V at full load but 400V at 20% load. The reason is an inversion of the loss structure under light load (Figure 3).

At full load, winding copper loss (I²R) is the largest single contributor, and the optimisation lever is conductor cross-section and DC resistance. At 20% load, current falls to one fifth and copper loss falls quadratically to about 4% of its full-load value - but core loss barely moves. Core loss is set by excitation frequency and flux density; at light load the output voltage is unchanged, the volt-second product per turn stays nearly the same, so flux density barely changes. Many topologies also raise switching frequency at light load, and by the Steinmetz-type empirical relation (core loss scaling roughly with frequency to the 1.3-1.7 and flux density to the 2-2.7), iron loss holds or even rises. Full load is a copper problem; light load is an iron problem - copper loss is a function of design variables, but core loss is dictated by the operating point. Magnetics optimised for full load routinely contribute the very percentage points that sink the light-load cells.

Two more reasons stack onto the 400V low-voltage cell: a lower bus voltage means higher current at the same power and duty cycles closer to their limits, so the converter drifts further from its optimum; and most modules run deeper into frequency modulation under low-voltage light load, where the raised switching frequency directly inflates core loss. Two layers on top of each other - this cell is where the nine points are most easily lost, and therefore where design capability is best told apart.

5. Fixing constraints at both ends: magnetics design under the 9-point matrix

Take the 9-point matrix as the design input, and the constraints can be fixed at both ends and checked in between:

• The full-load point sets the size. The 800V/100% point determines core cross-section, current density and the temperature-rise budget. This part matches conventional practice, but Bmax selection must leave margin for both high temperature and transients, not just the rated condition.
• The light-load point sets the material strategy. The 400V/20% point decides the core material and frequency approach: lower specific-loss core materials, clamping the light-load switching frequency in the control scheme, or burst mode on the system side. The ferrite-versus-metal-powder trade-off is re-shuffled here: distributed-gap powder cores bring high saturation and excellent DC bias, but their specific loss is usually higher than MnZn power ferrite; projects with a tight light-load iron-loss budget lean back toward gapped ferrite.
• The standby point sets the auxiliary budget. With the standby limit quantified to 30W (7W for AC chargers), the loss of the auxiliary supply transformer and standby circuitry faces a hard budget for the first time: no-load loss of the auxiliary transformer and the minimum operating point of the standby supply must both be back-calculated from 30W.

There is also a temperature dimension easily overlooked: core-loss data is usually given at room temperature, while magnetics inside a charger work long-term at 60-100°C, and the loss-temperature characteristic of the material changes the light-load result. MnZn ferrite loss falls as temperature rises in the low-temperature region, so in-cabinet measurements may actually look better than the datasheet; metal powder cores often behave the opposite way. The standard locks the test environment at 25°C ± 5°C, standardising exactly this variable: material selection must be compared on one temperature basis, or the light-load result cannot be reproduced.

A magnetic component that converges at all three points passes the report card. A transformer optimised only for full load may look great on the datasheet and still fail the matrix.

6. One year of grace: three retrofit paths

For products shipped before 1 November, the grace period runs to 1 November 2027. Three paths, in ascending order of cost:

Path one: replace the magnetics. Change only the main transformer, PFC boost inductor and resonant inductor - materials and winding structure - leaving topology and enclosure untouched. Lowest cost, shortest cycle, suitable for units within 1 percentage point of the line - which covers most legacy chargers.

Path two: replace the power module. Magnetics are upgraded together with the module, switching frequency and magnetics parameters re-designed jointly - for larger gaps, or when upgrading to SiC anyway. Path three: replace the unit. Redesign straight to grade 1, for product lines already near their replacement cycle.

The schedule constraint is tighter than the choice of path. Only 20 laboratories in China are accredited for charger CCC testing, and queuing for efficiency testing adds to retrofit and retest time - one year is not generous. For magnetics suppliers, the sampling window is now: every transformer and inductor in a retrofit plan must complete validation and production transition before the grace period ends. Wait until enforcement begins and the queue you join is at the test bench, not in the market.

A practical note for magnetics suppliers: before submitting samples, run the 9-point matrix yourself - at minimum provide measured efficiency and loss curves at the two corner points, 400V/20% and 800V/100%, with the test temperature stated. A datasheet reporting only peak efficiency will be filtered out under the new procurement process.

Conclusion

GB 46519 rewrites the charger efficiency narrative from 'a peak-efficiency marketing figure' to 'engineering capability across all operating conditions'. For charger OEMs, 1 November is the compliance line; for magnetic component makers, the real watershed comes earlier - whoever first proves a light-load iron-loss design method and ships retrofit samples captures the orders this one-year grace period releases.

What does your charging module score when the 9-point weighted method is applied?