A session running below the nameplate has three possible authors, and the dashboard shows only the result. The vehicle asks for a current. The charger permits a current. The site can supply a current. The session runs at whichever of the three is lowest, and the other two are invisible from the outside.
This is why the same complaint — the charger is slow — can lead to three completely different investigations, and why so many of them end in an argument rather than a repair. Separating the three limits is the only reliable way to find out which one was binding, and it is the difference between replacing hardware and changing a setting.
Three Limits, One Result
Think of the session as a chain of permissions rather than as a supply of power. Each participant can reduce the figure, and none of them can raise it beyond the one above.
The vehicle sets a request based on what its battery can accept right now. The charger sets an allowed value based on what its own hardware can deliver, given temperature, module health and how many connectors are in use. The site sets an available value based on the incoming supply, the load management scheme and whatever else is drawing power at that moment.
The minimum of the three is what the driver sees. The other two continue to exist, unmeasured, which is the root of most misdiagnosis. A charger that is permitted to deliver a high current but blocked by the site looks identical, from the app, to a charger that is derating because its own cabinet is hot.
The Vehicle Request
The request comes from the battery management system and is a function of cell voltage, temperature, state of charge and the pack's internal limits. It is the only one of the three that the operator cannot influence, and the one most often assumed to be constant when it is not.
The request is visible to the charger when the two are talking over a high-level protocol such as ISO 15118 or DIN 70121. On older or simpler communication, the charger sees only the resulting current, not the demand behind it. That distinction decides whether a diagnostic is possible at all: without a readable request value, an operator can see that a session was slow but cannot establish who chose the figure.
Two features of the request catch people out. It falls as the pack fills, which is normal and is not a fault. And it falls faster in the cold, before any warming has happened, which is also normal. In both cases the request is doing its job, and any intervention at the charger achieves nothing.
The Charger Allowed Value
The charger's permitted output is not a fixed number stamped on the label. It is recalculated continuously from several inputs, and any of them can pull it down.
Thermal headroom is the first. A cabinet in direct sun, or one whose intake filter has loaded up over a season, reduces output before any fault appears. The reduction is gradual and shows up as a summer problem rather than as a failure, which is why it is often attributed to the grid.
Module availability is the second. A unit with several power modules can continue operating with one out of service, at a lower ceiling. The session works, the driver notices nothing except a lower figure, and the fault sits unreported in the cabinet.
Connector sharing is the third. Where two connectors draw from a shared pool, the second vehicle reduces what the first can receive. This is deliberate design, not degradation, and it is the reason a site can look underperforming at lunchtime and fine at midnight.
The permitted value also depends on how finely the unit can divide its capacity. A cabinet that can only allocate in large blocks leaves capacity unused when a vehicle asks for less than one block. Parameters such as a 300 to 1000 V constant-power window on a 120 or 180 kW unit, or the minimum switching step of a hub, describe this behaviour and belong in a comparison between two units of nominally identical rating. The DC charging station specification lists both the power window and the switching granularity.
The Site Available Value
The third limit is the one the charger does not control and often does not report clearly. It is whatever the site can spare at that moment, after the building, the other chargers and the demand ceiling have taken their share.
Where load management is active, the site value changes through the day according to a schedule or a measured total. The charger receives a cap and honours it. From the driver's side this looks like the charger refusing to deliver, and from the charger's side it looks like an instruction being followed correctly. Both are true.
This limit also moves for reasons unrelated to electric vehicles. A kitchen starting service, an air-conditioning plant reaching peak, or a second tenant drawing more than usual all reduce what is available to the chargers. If the fall in power correlates with the time of day rather than with the vehicles arriving, the site value is the likely author.
Reading Both Numbers in the Same Minute
A log that shows only delivered power cannot answer the question. A log that shows delivered power and the requested value at the same timestamp can, because the comparison identifies which of the two moved first.
If the request falls and the output follows it, the vehicle is the author. If the request stays high and the output falls, the charger or the site is the author, and the charger's own permitted value separates those two. If the request is low from the first minute, the session had no high demand to lose, and the investigation belongs at the vehicle or at the connector handshake rather than at the power stage.
Why This Distinction Matters in Warranty Conversations
A claim that the charger underperforms is only actionable if the delivered figure can be compared against what the charger was permitted to deliver at that moment. Without both numbers, the conversation becomes an exchange of impressions, and it usually ends with the supplier asking for a log that the operator does not have.
For the operator, the practical requirement is that the backend records the requested value alongside the delivered value. For the supplier, the corresponding requirement is that the permitted value is exposed rather than inferred. Where either is missing, no amount of investigation after the fact recovers the evidence.
There is a second reason to keep the three separate. Remedies belong to different parties. A vehicle limit is answered by the driver or by the fleet's route planning. A site limit is answered by the electrical design or the load management profile. Only a charger limit is answered by the equipment supplier, and it accounts for a smaller share of cases than the volume of complaints suggests.
Three limits act at once, and only the lowest is visible. Before asking who is at fault, establish which of the three numbers was the one in force.
The seasonal version of this problem — a fall in power that appears only in cold weather — follows the same logic but with the vehicle's thermal limits doing most of the work. That case is treated separately in the note on cold-weather charging behaviour, and the diagnostic order there starts from the same two readings.
Trying to explain a power figure on site?
Send the session log with the requested and delivered values side by side. Which limit was in force becomes clear from the timestamp where the two lines separate.
Ask a specification question