To convert a 48V DC input to a 12V, 5A output using an isolated buck converter (specifically, the forward converter topology), you need a transformer turns ratio (Ns:Np) of 3:5 (0.6) and a primary switch duty cycle (D) of 43.4%. While many hobbyists loosely refer to a flyback as an isolated buck, the flyback is technically an isolated buck-boost; the forward converter is the true isolated buck, storing energy in the output inductor rather than the transformer core.
Vout + Vf = Vin × D × (Ns / Np)
Substituted Values: 12V + 0.5V (Schottky drop) = 48V × D × 0.6
Solved: 12.5 = 28.8 × D → D = 0.434 (43.4%)
What assumption fixes this answer? Unlike AC-DC conversions where Power Factor (PF) and phase angle dictate real power, in an isolated DC-DC buck, the fixed assumptions are Continuous Conduction Mode (CCM) and a standard transformer reset winding. A standard reset winding requires the duty cycle to remain strictly below 50% to allow the core flux to reset before the next switching cycle. Our 43.4% duty cycle safely satisfies this physical constraint.
The 48V to 12V Isolated Buck Conversion: Exact Math
In real-world 48V telecom or Power-over-Ethernet (PoE) applications, the input voltage is never exactly 48.0V. The bus can sag or spike. Below is the conversion table for a ±20% variance around the 48V nominal input, holding the 12.5V secondary target (12V output + 0.5V diode drop) and the fixed 3:5 turns ratio.
| Input Voltage (Vin) | Variance | Required Duty Cycle (D) | Core Reset Margin |
|---|---|---|---|
| 38.4V | -20% | 54.2% | Fails standard reset (needs active clamp) |
| 43.2V | -10% | 48.2% | 1.8% margin (Risky for standard reset) |
| 48.0V | Nominal | 43.4% | 6.6% margin (Safe) |
| 52.8V | +10% | 39.5% | 10.5% margin (Very safe) |
| 57.6V | +20% | 36.2% | 13.8% margin (Excellent) |
Note: If your 48V bus can drop to 38.4V, a standard reset winding will fail because D exceeds 50%. You must either change the turns ratio to 1:1 (requiring D=26% at 38.4V, but pushing D to 62% at 57.6V) or switch to an active-clamp forward topology, which safely allows duty cycles up to 70%.
Component Selection: Controllers and Magnetic Cores
Selecting the right silicon and magnetics is where theory meets the bench. The physical size of the transformer core is dictated by your output power (12V × 5A = 60W) and your switching frequency. Higher frequencies shrink the core but increase switching losses and EMI.
| Target Power | Recommended Core Size | Primary Turns (Typ) | Controller IC Example | Max Switching Freq |
|---|---|---|---|---|
| 15W | EFD15 / RM5 | 18 - 22 | TI LM5015 (Flyback/Buck) | 500 kHz |
| 30W | EFD20 / RM6 | 14 - 18 | TI UCC2813-3 | 250 kHz |
| 60W (Our Target) | EFD25 / ETD29 | 10 - 14 | TI LM5017 (Isolated config) | 300 kHz |
| 120W | EFD30 / ETD34 | 8 - 12 | TI UCC2897A (Active Clamp) | 220 kHz |
For our 60W 48V-to-12V design, an EFD25 ferrite core (like the Würth Elektronik 750313734 series) paired with a current-mode controller like the Texas Instruments UCC2813 provides excellent line regulation. Current-mode control inherently solves the right-half-plane zero issue and provides cycle-by-cycle overcurrent protection, which is critical when stepping down from a high-energy 48V bus.
How the Math Shifts Across 12V, 24V, and 48V DC Buses
Just as AC-DC designs must adapt to 120V vs 230V vs 3-phase mains, isolated DC-DC designs must adapt to the three dominant DC bus standards. The conversion math shifts dramatically based on the input source:
- 12V Nominal (Automotive/Marine): The input range is brutal (6V during cranking up to 16V during load dump). To maintain a 5V output, the turns ratio must be stepped up (e.g., 1:1 or 2:1), and the duty cycle will swing wildly. You often need a buck-boost or flyback here rather than a strict forward buck, because the input can drop below the output voltage.
- 24V Nominal (Industrial PLCs): A much tighter range (typically 18V to 32V). A 24V-to-12V isolated buck converter thrives here. A 1:1 turns ratio yields a nominal 50% duty cycle, making magnetic design incredibly straightforward and optimizing the RMS currents in both primary and secondary windings.
- 48V Nominal (Telecom/Data Center): As calculated above, the high input voltage demands a step-down turns ratio (like our 3:5) to keep the duty cycle under the 50% reset limit. The primary benefit here is lower primary-side RMS current, allowing the use of smaller, cheaper primary MOSFETs (e.g., 100V rated Si FETs instead of high-current 40V FETs).
When the Standard Conversion Formula Becomes Meaningless
The formula Vout = Vin × D × (Ns/Np) is a CCM idealization. It becomes entirely meaningless—and dangerous to rely on—under two specific conditions:
- Discontinuous Conduction Mode (DCM): If your load current drops below the critical inductance threshold (typically 10% to 20% of max load), the output inductor current hits zero before the next switching cycle. In DCM, the output voltage is no longer fixed by the duty cycle and turns ratio alone; it becomes a function of the load current, the inductor value, and the switching frequency. If you design your feedback loop assuming CCM math, the converter will exhibit severe overshoot and oscillation at light loads.
- Unaccounted Leakage Inductance: Real transformers have leakage inductance (typically 1% to 3% of the magnetizing inductance). This leakage inductance causes a voltage spike on the primary MOSFET drain and creates a 'duty cycle loss' on the secondary side. At 60W and 300kHz, a 500nH leakage inductance can easily eat up 4% of your effective secondary duty cycle. If you don't add this loss term to your math, your 12V output will sag to 10.8V under full load.
Frequently Asked Questions
Can I just use a flyback converter instead of an isolated buck (forward)?
You can, but you shouldn't for 60W. Flybacks store energy in the core gap, requiring larger, gapped ferrite cores and resulting in much higher output ripple current. For anything above 30W, the forward converter (isolated buck) is vastly superior because the output inductor filters the current continuously, yielding lower ripple and smaller output capacitors. See the Analog Devices guide on isolated topologies for a deep dive on the crossover point.
Why not just use a non-isolated buck converter?
If safety standards (like IEC 62368-1 for IT equipment) require galvanic isolation between the 48V input and the 12V output to protect the user or downstream sensitive logic from high-voltage faults, a non-isolated buck is illegal to use. The transformer in the isolated buck provides that physical dielectric barrier.
How do I measure the actual duty cycle on the bench?
Probe the primary MOSFET gate-to-source voltage with an oscilloscope. Do not probe the drain voltage, as the ringing from the transformer reset and leakage inductance will distort your high-time measurement. Measure the pulse width at 50% of the gate drive amplitude (usually 5V or 12V) for an accurate D calculation.
