To convert a rectified 230V AC mains bus (approximately 325V DC) to a 12V DC output using an offline single-switch forward converter with a 4:1 step-down transformer ($N_s/N_p = 0.25$), the required steady-state duty cycle ($D$) is 0.148 (14.8%). If the input shifts to a 120V AC line (170V DC bus), the duty cycle increases to 0.282 (28.2%). The governing DC transfer formula for Continuous Conduction Mode (CCM) is $V_{out} = V_{bus} \times D \times (N_s / N_p)$. Substituting our 230V baseline values: $12V = 325V \times D \times 0.25$, which yields $D = 12 / 81.25 = 0.1476$. This calculation assumes ideal components, CCM operation, and ignores the forward voltage drop of the secondary Schottky rectifier.
The Core Conversion Formula and Fixing Assumptions
The forward converter is the workhorse of offline isolated power supplies in the 50W to 300W range. Unlike a flyback, which stores energy in the transformer gap, a forward converter transfers energy directly to the secondary during the MOSFET's on-time. The fundamental voltage conversion ratio is fixed by three variables:
Where $V_{bus}$ is the rectified DC input, $D$ is the duty cycle, and $N_s/N_p$ is the secondary-to-primary turns ratio.
What fixes this answer in the real world? The assumption of Continuous Conduction Mode (CCM). In CCM, the output inductor current never falls to zero during the switching cycle, making the voltage conversion ratio independent of load current. If your design relies on this math, you must size the output inductor ($L_{out}$) to maintain CCM down to your minimum expected load (typically 10% to 20% of full load).
Input Voltage Shifts: 120V, 230V, and 3-Phase Systems
The duty cycle must dynamically shift as the AC mains fluctuates or when the same power supply is deployed across different global grids. Here is how the math scales across standard AC inputs, assuming our 4:1 transformer and 12V output target.
- 120V AC Nominal (170V DC Bus): $D = 12 / (170 \times 0.25) = 0.282$. This is the sweet spot for single-switch forward converters, leaving ample off-time for transformer core reset.
- 230V AC Nominal (325V DC Bus): $D = 12 / (325 \times 0.25) = 0.148. Duty cycle drops, reducing secondary RMS currents but increasing primary-side switching losses relative to conduction losses.
- 3-Phase 400V AC (565V DC Bus): $D = 12 / (565 \times 0.25) = 0.085. Warning: While the math works, a single-switch topology is practically forbidden here (see below).
Neighboring Values: 230V AC ±20% Tolerance Table
Grid voltage is never perfect. The IEC 60038 standard allows for significant variance. Here is how your duty cycle must shift across a ±20% range of a 230V nominal input to maintain exactly 12V out.
| AC Input Voltage | Variance | Rectified DC Bus ($V_{bus}$) | Required Duty Cycle ($D$) | MOSFET Off-Time (at 100kHz) |
|---|---|---|---|---|
| 184V AC | -20% | 260V | 0.185 (18.5%) | 8.15 µs |
| 207V AC | -10% | 293V | 0.164 (16.4%) | 8.36 µs |
| 230V AC | Nominal | 325V | 0.148 (14.8%) | 8.52 µs |
| 253V AC | +10% | 358V | 0.134 (13.4%) | 8.66 µs |
| 276V AC | +20% | 390V | 0.123 (12.3%) | 8.77 µs |
When the Conversion Math Becomes Meaningless
There are two specific scenarios where plugging numbers into $V_{out} = V_{bus} \times D \times (N_s / N_p)$ will result in a fried prototype on your bench.
1. Discontinuous Conduction Mode (DCM) at Light Loads
If your load current drops below the critical inductance threshold, the output inductor current reaches zero before the switching cycle ends. In DCM, the output voltage is no longer defined solely by the duty cycle and turns ratio; it becomes highly dependent on the load current and inductor value. If your controller lacks pulse-skipping or burst-mode capabilities, the output voltage will ring up and overshoot your 12V target at no-load.
2. The Single-Switch Reset Limit and 3-Phase Inputs
In a single-switch forward converter, the transformer core must be reset (demagnetized) during the MOSFET's off-time. This physically restricts the maximum duty cycle to $D < 0.5$ (practically $D_{max} \approx 0.45$ to allow for leakage inductance ringing). Furthermore, during the off-time, the MOSFET drain voltage spikes to $2 \times V_{bus}$ (assuming a 1:1 reset winding).
If you attempt to use the 3-phase 400V AC calculation ($V_{bus} = 565V$), your MOSFET must withstand $>1130V$ during reset. Standard 800V or 900V Superjunction MOSFETs will avalanche and fail. For 3-phase offline inputs, the single-switch forward converter math is meaningless; you must pivot to a Two-Switch Forward topology (which clamps drain voltage to $V_{bus}$) or an LLC Resonant converter.
Topology and Controller Decision Tree
Use this decision path to select the correct forward converter variant and controller IC based on your input voltage and power requirements.
| Input Condition | Power Level | Topology Choice | Concrete Controller Pick |
|---|---|---|---|
| Universal AC (85-265V) | < 150W | Single-Switch Forward (Reset Winding) | TI LM5025 (Integrated 80V gate drive, max D clamped to 0.5) |
| Universal AC (85-265V) | 150W - 300W | Active Clamp Forward | TI UCC2897A (Recovers leakage energy, achieves ZVS, allows D > 0.5) |
| 3-Phase 400V AC (565V Bus) | 300W - 1000W | Two-Switch Forward | Microchip MCP1012 or TI UCC28C42 (Configured for 2-switch, D clamped < 0.5) |
| 3-Phase 400V AC (565V Bus) | > 1000W | LLC Resonant (Abandon PWM Forward) | TI UCC25600 (High voltage resonant controller) |
Forward Converter Design FAQ
Why does my single-switch forward converter blow the primary MOSFET at 265V AC input?
At 265V AC, your DC bus is roughly 375V. With a standard reset winding, the MOSFET drain sees $2 \times V_{bus}$ (750V) plus leakage inductance spikes. If you are using a standard 600V MOSFET, it is avalanching. You must use a minimum 800V CoolMOS (like the Infineon IPB60R099CP) and implement a robust RCD snubber across the primary, or switch to an Active Clamp topology to recycle that leakage energy.
How do I calculate the transformer turns ratio if I don't know the exact duty cycle?
Work backward from your maximum duty cycle limit. For a single-switch forward, set $D_{max} = 0.42$ (leaving margin for the reset time and transient response). Use the minimum expected DC bus voltage ($V_{bus(min)}$). The formula becomes: $N_s/N_p = V_{out} / (V_{bus(min)} \times 0.42)$. For a 12V output and 260V minimum bus, $N_s/N_p = 12 / (260 \times 0.42) = 0.109$, which translates to roughly a 9:1 primary-to-secondary ratio.
Can I parallel two forward converters for higher current?
Yes, but you must use interleaved synchronization. If you run two LM5025 controllers at the same 100kHz frequency without syncing their clocks, the input ripple currents will beat against each other, causing massive RMS stress on your bulk input capacitors. Use a master/slave sync pin configuration to offset the clocks by exactly 180 degrees.
For deeper magnetic design guidelines, refer to the All About Circuits forward converter topology guide and the Texas Instruments isolated DC/DC topology overview to verify your core selection and gap calculations before sending your PCB to fab.
