EV Charger Current Control

Four limits, and the lowest one wins

This is why a 22 kW charger frequently delivers 7 kW and nothing is wrong. The vehicle's onboard charger is the binding limit, and no setting on the wall changes it.

How the charger actually commands a limit

In AC charging the charger does not push current. It advertises a maximum, and the vehicle draws up to that. The mechanism is the control pilot signal defined in IEC 61851: a square wave whose duty cycle encodes the current the vehicle is permitted to take.

• The charger sets the duty cycle to correspond to an available current
• The vehicle reads it and limits its own draw accordingly
• The charger can change the duty cycle mid-session, and a compliant vehicle follows within a defined response time
• The proximity circuit in the cable separately caps the whole arrangement at the cable's rating

Two consequences follow. First, current control is advisory in the sense that it depends on the vehicle honouring it, which is why protection still has to exist independently. Second, a limit can be lowered during a session without interrupting it, which is what makes dynamic management possible at all.

Static limits versus dynamic control

1. 1

   Static allocation

   Each point is fixed at a share of the supply. Simple, provably safe, and wasteful: on a site with eight bays and two cars charging, six shares sit idle while the two cars charge slowly.

2. 2

   Static with diversity

   Points are set assuming not all will be in use. Cheaper in capacity terms and it fails on the day everyone plugs in, which in apartments is most days.

3. 3

   Dynamic sharing

   The available capacity is divided among active sessions and reallocated as vehicles arrive and leave. Uses the supply properly and requires the points to be coordinated.

4. 4

   Dynamic against measured site load

   The charging allocation is what the site is not using, measured in real time from the incoming supply. This is the version that lets charging share a connection with a building rather than needing its own.

The last of these is where most constrained sites should end up, and it depends entirely on having a trustworthy measurement of what the rest of the building is doing.

Why control usually beats reinforcement

A site sized so every bay can charge simultaneously at full rate needs capacity, switchgear and civil work that a site with managed load does not.

In apartments, which show the highest utilisation of any AC site we operate and concentrate it overnight, the load is also almost entirely schedulable, because nobody is leaving before morning. Holding charging out of the evening peak and completing across the remaining hours removes a supply upgrade from many projects. Nothing about the charger changed. The assumption did.

Where control must not be trusted alone

Current control is a management function, not a protective one, and the distinction matters.

• A vehicle that ignores the advertised limit is not prevented by the pilot signal from drawing more, so overcurrent protection exists independently and must not be relieved by the presence of load management
• A load management system that loses its measurement input must fail to a safe fixed allocation rather than to whatever it last knew
• Communications loss between coordinated points must reduce allocation, not hold it, because two controllers each assuming they have the whole supply is the failure that trips the site

These are design requirements rather than optional robustness. A load manager whose failure mode is optimistic is a load manager that will eventually take a building off supply.