Power Modules in DC Chargers: Why They Are Modular

Two DC chargers carry the same rating on the label and behave differently on site. One delivers smoothly across the whole range a vehicle asks for. The other jumps between large blocks, leaves capacity idle and serves the second car worse than the first. The difference is usually not the total power but how that power is built and divided.

The unit of construction is the power module. Understanding what it does, where it is efficient and how many of them sit behind each connector explains most of the behaviour that a datasheet summarises in a single number.

What a Power Module Does

A module is a self-contained converter. It takes three-phase AC from the cabinet's distribution, corrects the input current so the supply sees a clean load, rectifies to a DC bus, and then converts that bus to the voltage and current the vehicle is asking for. It carries its own control loop, its own cooling path and its own protection.

Modules are wired in parallel on the DC side. Their outputs combine, so a cabinet's rating is roughly the number of modules multiplied by the rating of one. That arithmetic is simple, and it is also where most procurement comparisons stop, which is a mistake: the parallel arrangement is what creates both the flexibility and the constraints that follow.

Building a charger from repeated identical units brings three practical benefits. The design effort is repeated rather than reinvented, so qualification effort is shared across configurations. Manufacturing yield improves because a faulty module is replaced rather than a whole cabinet reworked. And service becomes a swap rather than a repair, which is the difference between a site returning to service the same day and waiting for a technician with a soldering iron.

The Efficiency Curve Is Not Flat

A converter has two kinds of loss. Some are proportional to the power being processed: conduction losses in semiconductors and copper. Others are roughly fixed whenever the unit is running: switching losses, control electronics, fans and auxiliary supplies. At light load the fixed part dominates, so efficiency is poor. As load rises the fixed part is spread over more output, and efficiency climbs. Near full load, conduction losses begin to dominate and the curve flattens or turns down.

The consequence is that the best operating point sits somewhere in the middle of the range, not at the top. A charger quoted at its peak efficiency figure is being quoted at a load point that a real session may pass through briefly or not at all.

Modular construction gives a way to manage this. Rather than running all modules at light load during a slow session, the controller can run fewer of them closer to their efficient band and leave the rest idle. This is one reason a modular cabinet can beat a single large converter on real-world energy use, despite having more components in total.

It also means the efficiency figure depends on how the cabinet is dispatched, not only on the hardware. Two units with identical modules and different dispatch logic show different consumption over a month of mixed sessions.

Why Module Count Sets the Smallest Power Step

Modules can be switched in and out, but they are indivisible. If each module contributes a fixed amount, then the output can only take values that are multiples of that amount, plus whatever partial capability the module itself offers at low output.

This is the smallest power step, and it is a specification in its own right. A cabinet with four large modules cannot serve a vehicle requesting a little more than one module's worth without switching in a whole second module and then throttling it. The extra capacity exists but cannot be sold to anyone else at that moment.

Where the step is small, capacity is allocated finely and less is stranded. Where the step is large, the cabinet behaves well at full output and poorly in the middle. A 120 or 180 kW unit that switches in steps of 30 kW, or a hub that allocates in steps of 40 kW, is describing exactly this property — not the total rating, but the resolution at which the total can be divided.

Central Dispatch vs Fixed Per-Connector

In a fixed arrangement each connector owns a set of modules. A car on one connector has a hard ceiling and cannot borrow from a stall standing empty. This is simple to reason about and simple to certify, and it wastes capacity whenever demand is uneven.

In a central arrangement all modules sit behind a common bus and the controller assigns power to connectors as sessions require. A lone vehicle can draw nearly the whole cabinet; two vehicles split it according to a policy. The charging hub host is built this way, with the host cabinet holding the conversion and terminals drawing from it.

The policy matters more than the topology. A fair split gives both cars half and can leave a nearly full car holding power it can no longer accept. A demand-following policy gives each car what it asks for until the pool runs out, then allocates the remainder. The first is easier to explain to a driver; the second delivers more energy per hour from the same hardware.

What N+1 Redundancy Really Buys

The term is often read as meaning the cabinet can lose a module and still deliver its rated power. That reading is wrong. Redundancy means the cabinet can lose a module and continue operating, at reduced output, rather than shutting down.

What it buys is availability, not capacity. A site with N+1 modules stays open through a single module failure; a site without it loses the cabinet until a replacement arrives. For a motorway location where a closed stall is a headline, that trade is worth paying for. For a depot where sessions can be rescheduled across the night, it often is not.

The value depends on a second factor as well: whether the module can be replaced live. Hot-swap capability converts a failure into a maintenance task with no service interruption. A module that requires the cabinet to be shut down turns redundancy into a scheduled outage instead, which is a much smaller benefit than the label implies.

Comparing Two Units of the Same Rated Power

Six questions separate two cabinets that share a headline number.

How many modules, and at what rating each? What is the smallest power step the controller can allocate? Is the pool shared across connectors or fixed per connector, and what policy governs the split? Does the efficiency figure come from the middle of the load range or from the top? Is a module replaceable without taking the cabinet out of service? And when one module is out, what output remains?

Those answers predict behaviour on site far better than the rated power, which is only an upper bound. The 120 and 180 kW units publish their module configuration and switching step for exactly this comparison, and the same figures are worth requesting from any supplier evaluated alongside them.

This is why the same top-line figure can perform so differently between a busy retail site and a quiet depot. The retail site cares about the step, because capacity has to be reassigned every few minutes. The depot cares about overnight energy, where a coarser step costs little. Neither is wrong; they are asking different questions of the same specification.

A practical way to test a proposed configuration is to model the busiest hour rather than the average day. Take the number of vehicles expected in that hour, the energy each needs and the time each stays. If the allocation policy and the step size can serve them without leaving modules idle while a queue forms, the configuration is sound. If not, adding total power does not fix it, because the constraint is resolution rather than capacity. The range of cabinet and terminal combinations is set out under fast charging stations.