Why Does a Constant-Power Load Draw More Battery Current as Voltage Falls?

Identify what the equipment is trying to hold constant

Many electronic systems regulate an output rail or maintain a controlled mechanical or optical task. Over part of their operating range, their battery input resembles a constant-power demand more closely than a fixed resistor. This is an approximation with limits: startup, current limiting, standby and shutdown may behave differently. Ask the host designer which behavior applies in each mode rather than assigning one load model to the entire product.

Keysight distinguishes constant-current, constant-resistance and constant-power modes in its electronic-load documentation. That distinction explains why a convenient fixed-current discharge may not reproduce an actual regulated host. A constant-current test can still be a valid reference capacity test. Its result simply needs to be labeled for the question it answers.

Work through current at several battery voltages

Consider a hypothetical converter delivering 60 W at an assumed 90 percent efficiency. Battery-side power is about 66.7 W. At 25 V input, estimated current is 2.67 A; at 20 V it becomes 3.33 A. The lower-voltage point demands 25 percent more current for the same useful output. These values are an arithmetic illustration, not specifications or recommended operating points for a Jetray pack.

Real conversion efficiency changes with input voltage, load and temperature. Some converters also stop or limit current before the battery reaches its own approved discharge endpoint. Use the relevant efficiency curve and converter input limits, then confirm with measured input power. If the power was already measured at the battery connector, dividing it by efficiency again would count the loss twice.

Understand the feedback between current and voltage drop

A simple resistance model helps illustrate one part of the problem. For an unchanged resistive path, voltage drop scales with current and resistive heating scales with current squared. Increasing current from 2.67 A to 3.33 A would increase that path's resistive heating by about 56 percent if its resistance were unchanged. This is not a complete battery thermal model; it is a reason to inspect the current path at the low-voltage end of the mission.

The battery is not an ideal voltage source with one fixed resistor. Its response depends on operating condition and time, while the converter has its own control and limits. A higher current can deepen terminal sag, leading a regulating load to request still more current until a system limit intervenes. The actual endpoint could be host undervoltage shutdown, pack protection, current limiting or thermal restriction. Preserve the stopping reason rather than reporting every endpoint as an empty battery.

Qualification should use the approved wiring and connector arrangement. A host that runs from laboratory leads may behave differently with a longer production harness. Do not use this explanation to justify lowering protective thresholds or running below the approved voltage window.

Compare delivered energy and useful runtime at a defined endpoint

Delivered charge and delivered energy are different quantities. Charge is the time integral of current; energy is the time integral of voltage multiplied by current. For a time-varying profile, use synchronized measurements and the actual sampling interval. An amp-hour result without its voltage record cannot fully describe the energy supplied to a regulated load.

Keep two useful tests in the specification when appropriate. A controlled reference test supports consistent comparison between lots or suppliers. A mission-profile test answers whether the installed equipment performs its required task. They can have different load modes and stopping conditions without either being invalid. The mistake is presenting the result of one as proof of the other.

Record capacity and energy at the agreed boundary, along with temperature, starting condition and cutoff. A manufacturer's example curve belongs to its own model and conditions. It is supporting context, not a substitute for finished-pack evidence for your converter and mission.

Choose a corrective action from the observed limitation

If the limiting factor is insufficient mission energy under the approved conditions, a different capacity or a lower duty-cycle demand may be relevant. If the host stops because its input drops too far during a transient, voltage support and current-path behavior require investigation. If the converter reaches a current limit, adding nominal amp-hours may not address the constraint. Ask for an explanation tied to the recorded event.

Potential design changes can affect more than runtime. A different voltage architecture changes charger and host compatibility; a changed harness affects installation; a different cell or protection revision may require renewed qualification. Have the responsible supplier and equipment engineers review the complete change. Treat linked catalog products as inquiry references, and request the current approved operating window before choosing between them.

1. Classify each host mode as measured power, current or another controlled behavior.
2. Calculate expected current across the approved input-voltage window.
3. Capture actual battery and host input behavior under the relevant mission.
4. Identify the first limiting condition and assess a controlled design change.

Questions to include in a runtime inquiry