To get 5V at 3A (15W output) from a 12V DC source using a switching buck converter with 90% efficiency, the input draws exactly 1.39A. The governing formula for this electrical unit conversion is $I_{in} = (V_{out} \times I_{out}) / (V_{in} \times \eta)$. Substituting our exact query values: $I_{in} = (5V \times 3A) / (12V \times 0.90) = 15W / 10.8W = 1.388A$. This is the fundamental math behind how DC-DC converters work: they trade voltage for current (or vice versa) while losing a fixed percentage of power to switching losses, conduction losses, and inductor core heating.
The Core Assumption: Efficiency and Topology
The single assumption that fixes the answer above is the efficiency rating ($\eta$). If you do not know the efficiency of your converter topology, any input current calculation is a guess. How DC-DC converters work internally dictates this efficiency gap.
When designing power rails, always pull the efficiency curve from the IC’s datasheet. Efficiency is not static; it peaks at a specific load (often around 50-70% of max rated current) and drops off at light loads due to quiescent current ($I_q$) and at heavy loads due to $I^2R$ conduction losses in the internal MOSFETs and the inductor's DC resistance (DCR).
System Voltage Shifts: 12V, 48V, 120VDC, and 230VDC
While 120V and 230V are universally recognized as AC mains voltages, in high-voltage DC (HVDC) telecom, data center, and EV architectures, 120VDC and 230VDC distribution buses are increasingly standard. The DC-DC conversion math scales identically across these buses, though component selection shifts drastically.
Here is how the input current answer shifts for our 15W (5V @ 3A) load across different nominal DC input systems, assuming a 92% efficiency typical of modern synchronous buck controllers:
| Input System | Nominal $V_{in}$ | Input Power (at 92%) | Calculated $I_{in}$ | Design Consideration |
|---|---|---|---|---|
| Automotive / Marine | 12V | 16.30W | 1.36A | Standard 18 AWG wire, low $V_{ds}$ MOSFETs |
| Solar / Telecom | 24V | 16.30W | 0.68A | Lower input ripple current, smaller input caps |
| Industrial / PoE | 48V | 16.30W | 0.34A | Requires high-voltage rated ICs (e.g., 60V+ max $V_{in}$) |
| HVDC Telecom Bus | 120VDC | 16.30W | 0.136A | Requires isolated topologies or specialized HVDC-DC modules |
| HVDC Data Center | 230VDC | 16.30W | 0.071A | Strict creepage/clearance, arc-flash mitigation required |
Note on 3-Phase: 3-phase AC is entirely outside the scope of a DC-DC converter stage. If you are stepping down from a 3-phase 480V AC line, you must first use an AC-DC Active Front End (AFE) or passive rectifier with Power Factor Correction (PFC) to establish a high-voltage DC bus (typically ~650VDC to 800VDC) before a DC-DC converter can take over.
Neighboring Load Values (±20% Range)
Load currents in embedded systems rarely sit at a perfect integer. Microcontrollers, RF amplifiers, and motor drivers draw dynamic current. Below is the conversion table for a 5V output rail fed by a 12V source, covering a ±20% range around our 3A baseline, assuming 90% efficiency.
| Output Current ($I_{out}$) | Output Power | Input Power (90% Eff) | Input Current ($I_{in}$) |
|---|---|---|---|
| 2.40A (-20%) | 12.00W | 13.33W | 1.11A |
| 2.70A (-10%) | 13.50W | 15.00W | 1.25A |
| 3.00A (Baseline) | 15.00W | 16.67W | 1.39A |
| 3.30A (+10%) | 16.50W | 18.33W | 1.53A |
| 3.60A (+20%) | 18.00W | 20.00W | 1.67A |
When sizing your input traces or selecting a polyfuse for the 12V rail, always design for the +20% to +30% threshold. A 1.67A continuous draw requires at least a 2A fuse, but accounting for inrush current into the input ceramic capacitors, a 3A slow-blow fuse is the correct bench choice.
When This Power Conversion Becomes Meaningless
The formula $I_{in} = P_{out} / (V_{in} \times \eta)$ is mathematically sound, but practically meaningless in three specific scenarios:
- Dropout Voltage Violation: Every converter has a minimum headroom requirement. If you attempt to use a buck converter rated for a 4.5V minimum input to step 5.0V down to 4.8V, the math suggests a tiny input current. In reality, the internal high-side MOSFET cannot maintain a 100% duty cycle without dropping out of regulation. The output will sag, and the IC may enter an undefined linear state, frying the die.
- Thermal Shutdown Limits: If you calculate that a tiny SOT-23 packaged linear regulator drawing 0.5A from 12V to 5V will only pull 0.5A from the source, the math ignores thermals. The IC is dissipating 3.5W. Without a heatsink, the junction temperature will exceed 150°C in seconds, triggering thermal shutdown. The conversion stops entirely.
- Unknown Topology (Charge Pumps):strong> If the DC-DC converter is an inductorless charge pump (switched-capacitor) rather than an inductive switcher, the efficiency curve is highly non-linear and heavily dependent on the exact output voltage ratio. Assuming a flat 90% efficiency for a charge pump stepping 5V to 3.3V will yield incorrect input current estimates, as charge pumps suffer severe efficiency drops when the target voltage is not an exact integer fraction of the input.
For reliable calculations across complex topologies, engineers rely on tools like the Analog Devices LTpowerCAD simulator, which models inductor ripple, MOSFET switching losses, and PCB trace resistance simultaneously.
Frequently Asked Questions
How do DC-DC converters work to step up voltage (boost)?
A boost converter steps up voltage by storing energy in an inductor while a switch (MOSFET) is closed, connecting the inductor directly across the input source. When the switch opens, the inductor's collapsing magnetic field forces its voltage to reverse polarity. This induced voltage adds in series with the input voltage, pushing current through a diode (or synchronous rectifier) into the output capacitor at a higher potential. Because power must be conserved ($P_{in} \approx P_{out}$), stepping up the voltage results in a proportional decrease in available output current compared to the input current drawn.
How do DC-DC converters work with PWM duty cycles?
In a continuous conduction mode (CCM) buck converter, the output voltage is directly proportional to the Pulse Width Modulation (PWM) duty cycle ($D$) multiplied by the input voltage ($V_{out} = D \times V_{in}$). If you need 5V from a 12V source, the controller adjusts the PWM duty cycle to roughly 41.6% ($5 / 12$). The internal oscillator runs at a fixed switching frequency (e.g., 500 kHz), and the error amplifier continuously tweaks the on-time versus off-time of the PWM signal to maintain the 5V output despite load transients or input voltage sags.
How do DC-DC converters work in battery-powered systems regarding quiescent current?
In battery systems, the primary enemy is quiescent current ($I_q$)—the current the DC-DC IC consumes just to keep its internal logic and oscillators running, even when the load is drawing 0A. A standard buck converter might draw 3mA of $I_q$. If your IoT sensor sleeps for 23 hours a day, that 3mA continuous drain will deplete a small lithium coin cell rapidly. For these applications, you must select converters with "burst mode" or "pulse-skipping" architectures (like the TI TPS62840) that drop $I_q$ to the nano-amp range (e.g., 60 nA) during light loads, fundamentally altering the light-load efficiency math.
