How a Bidirectional DC Power Supply Works

Sourcing and sinking through one stage

Inside a bidirectional supply the output converter is built so its current can be positive or negative relative to the terminals. When the device under test demands power, the converter pushes current outward — it sources. When the device tries to push power back (a charged battery, a regenerating inverter), the converter accepts that current and processes it — it sinks.

The key point is that there is no separate "load mode" hardware to engage. The same transistors, the same magnetics and the same control loop handle both polarities of current. That is what makes the transition continuous rather than a switched event.

The CC and CV regulation loops

Two nested feedback loops keep the output disciplined. The constant-voltage (CV) loop holds the terminal voltage at the programmed value; the constant-current (CC) loop limits or holds the current at its programmed value. Whichever limit the load hits first takes control — this is the classic CV/CC crossover.

• In CV, the supply adjusts current as needed (positive or negative) to keep voltage on target. Charging a battery to a set voltage is a CV task.
• In CC, the supply holds a fixed current and lets voltage float. A controlled discharge or charge at a defined rate is a CC task.
• CP and CR modes layer power and resistance regulation on top of the same loops.

Because the current term in these loops is signed, the very same equations govern sourcing and sinking. The controller does not need to know which way energy is flowing; it only regulates to the setpoint.

Seamless crossover and no overshoot

The hard engineering problem is the moment current passes through zero as the supply transitions from sourcing to sinking. A naive design has a dead band there — a small region where neither path conducts — producing a glitch in voltage or current. A well-designed bidirectional supply eliminates that dead band so the output current crosses zero smoothly.

Equally important is dynamic response without overshoot. When a setpoint changes or the load steps, the loops must settle quickly but not ring past the target, which could stress a battery or trip an inverter. Platforms like the N35500 specify response of 5 ms or better with 0.02% full-scale accuracy, meaning fast settling held to a tight final value. Ultra Power Systems documents these loop characteristics for the N35500 platform.

Why the architecture matters in testing

For charge/discharge cycling of cells and packs, the supply must reverse current thousands of times across a test campaign. A clean crossover means each reversal is invisible to the device under test — no transient that would corrupt a capacity measurement or trigger a protection fault. For inverter and drive testing, the supply must absorb regenerated energy on demand while holding the DC-bus voltage steady. The same architecture serves both, which is why one bidirectional channel replaces a supply-plus-load pair.

Frequently asked questions