A rectifier converts alternating current (AC) into direct current (DC) by utilizing semiconductor PN junctions that permit electron flow in only one direction. However, in practical power supply design, asking "how does a rectifier work" requires looking past the raw diode bridge. A functional AC/DC rectifier circuit encompasses stepping down the mains voltage, rectifying the waveform, filtering the ripple with bulk capacitance, and regulating the output to a stable DC level. The choices you make at the rectification stage dictate your thermal management, efficiency, and electromagnetic interference (EMI) profile for the entire system.
The Core Mechanism: Diode Bridges and Topology Trade-offs
At the component level, a standard silicon rectifier diode (like the 1N5408) relies on a depletion region that collapses under forward bias, allowing current to pass with a typical forward voltage drop ($V_f$) of 0.7V to 1.1V. Under reverse bias, the depletion region widens, blocking current until the Peak Inverse Voltage (PIV) rating is exceeded, resulting in avalanche breakdown and catastrophic failure.
In a standard full-wave bridge configuration, current passes through two diodes in series during every half-cycle. If your load draws 10A and your bridge uses silicon diodes with a 1.1V drop, you are burning $2.2V \times 10A = 22W$ of power purely as heat in the rectifier stage. This is why topology selection is critical before you even pick a regulator. Below is a comparison of common rectifier topologies used in AC/DC conversion.
| Topology | Conduction Path | Efficiency (Rectification Only) | Voltage Drop & Heat Profile | Ripple Frequency |
|---|---|---|---|---|
| Half-Wave | 1 Diode | ~40% (High transformer DC saturation) | $1 \times V_f$ (Low drop, but high RMS current heating) | $1 \times f_{line}$ (50/60Hz) |
| Full-Wave Bridge | 4 Diodes | ~85-90% | $2 \times V_f$ (High drop, requires heatsinking >5A) | $2 \times f_{line}$ (100/120Hz) |
| Center-Tapped | 2 Diodes | ~90-92% | $1 \times V_f$ (Lower drop, but requires expensive dual-secondary transformer) | $2 \times f_{line}$ (100/120Hz) |
| Active Synchronous | 4 MOSFETs | >98% | $I^2 \times R_{DS(on)}$ (Negligible drop, complex gate drive required) | $2 \times f_{line}$ (100/120Hz) |
For hobbyist and low-to-medium power bench supplies (under 200W), the full-wave bridge remains the undisputed standard due to its low component count and the availability of integrated bridge modules like the Vishay KBPC5010. For high-current server power supplies, designers move to active synchronous rectification, replacing diodes with MOSFETs to eliminate the $2 \times V_f$ penalty, as detailed in Texas Instruments' power supply topology guides.
Linear vs. Switching Regulation: Dropout Math and Ripple
Once the AC is rectified and smoothed by a bulk capacitor, you are left with unregulated DC. The voltage will sag under load and ride on a sawtooth ripple waveform. You must now choose between a linear regulator and a switching regulator. The decision hinges entirely on your load current, acceptable heat dissipation, and noise tolerance.
Linear vs. Switching for a 12V @ 3A Load:
Assume your transformer and bridge yield a nominal 18VDC unregulated rail under a 3A load.
- Linear Path (e.g., TI LT1083 7.5A LDO): The LT1083 requires a minimum dropout voltage (headroom) of 1.5V. With 18V in and 12V out, your headroom is 6V, which easily passes the dropout requirement. However, the power dissipated as heat is $(18V - 12V) \times 3A = 18W$. You will need a substantial extruded aluminum heatsink to keep the junction temperature below 125°C.
- Switching Path (e.g., TI LM2596 Buck): The LM2596 has an internal switch dropout of roughly 1.5V, but because it uses pulse-width modulation (PWM) and an inductor to transfer energy rather than burning excess voltage as heat, its efficiency sits around 85%. Total heat dissipated is roughly $(36W / 0.85) - 36W = 6.3W$, easily managed with a small PCB copper pour or clip-on heatsink.
Ripple and Noise Expectations:
If you are powering precision analog sensors, audio DACs, or RF transceivers, the regulator topology dictates your noise floor. According to Analog Devices' application notes on power management, a well-designed linear regulator will yield output ripple and noise in the microvolt range (<2mV RMS), effectively rejecting the 120Hz input ripple. A switching regulator, conversely, will inject high-frequency switching noise (typically 20mV to 50mV peak-to-peak) at its switching frequency (e.g., 150kHz for the LM2596), which requires secondary LC pi-filters to clean up for sensitive loads.
Design Example: 120VAC to 12VDC 5A Rectifier Supply
Let's specify exact part values for a robust, non-isolated bench supply stepping 120VAC down to a regulated 12VDC at 5A. This design uses a step-down transformer for safety, followed by a bridge, bulk filtering, and a switching buck regulator.
Bill of Materials & Specifications
- Input Protection (MOV): Littelfuse TMOV20RP151M (150V RMS, 20mm disc). Clamps transient mains spikes before they reach the transformer.
- Inrush Limiting (NTC): Ametherm SL32 2R015. 2 ohms cold resistance, limits the initial capacitor charging surge to ~80A, dropping to 0.04 ohms at steady state to minimize $I^2R$ losses.
- Transformer: 120VAC Primary to 15VAC Secondary, 100VA rating. (15VAC yields ~21.2V peak DC, providing ample headroom for the regulator after diode drops and ripple sag).
- Rectifier Bridge: Vishay KBPC5010 (50A, 1000V). Massively overrated for a 5A load to ensure low thermal resistance and long-term reliability without a fan.
- Bulk Filter Capacitor: Nichicon LNR2A223MSE (22,000µF, 100V). Sized using $C = \frac{I}{2 \times f \times V_{ripple}}$. For 5A at 120Hz with a target 2V ripple: $C = \frac{5}{240 \times 2} = 10,416\mu F$. We double it to 22,000µF to account for capacitor aging and ESR increase over time.
- Regulator: RECOM R-78E12-2.0 (Switching Buck) paralleled, or a custom TI LM2596HVS circuit configured for 12V out.
Working with 120VAC requires de-energizing the circuit, locking out the breaker, and verifying dead with a CAT III multimeter before touching any terminals. Furthermore, a 22,000µF capacitor bank acts as a dead short the millisecond power is applied. Never omit the NTC inrush thermistor; without it, the initial current spike will weld your switch contacts and trip a standard 15A magnetic breaker instantly.
Thermal Derating: Why a 50A Bridge Fails at 15A
A common mistake on the workbench is reading the "50A" printed on a KBPC5010 bridge rectifier and assuming it can handle 50A of continuous current in free air. Datasheet numbers are measured under ideal, infinite-heatsink conditions at 25°C ambient. In the real world, you must apply thermal derating.
The KBPC5010 has a maximum junction temperature ($T_J$) of 150°C and a thermal resistance from junction-to-case ($\theta_{JC}$) of roughly 0.6°C/W per diode pair. If you mount it to a small 10°C/W heatsink inside an enclosure where the ambient air is 50°C, your total thermal resistance ($\theta_{JA}$) is roughly 10.6°C/W.
If you pull 20A through the bridge, the forward drop is roughly 1.0V per diode. Two diodes conduct at a time, so total power dissipation is $20A \times 2.0V = 40W$.
Temperature rise = $40W \times 10.6°C/W = 424°C$.
Add the 50°C ambient, and your junction temperature is 474°C. The silicon will melt, the epoxy will crack, and the diode will short out long before you hit 20A.
The Fix: For continuous loads above 10A in enclosed spaces, you must either bolt the bridge to the metal chassis using thermal paste (like Arctic Silver) to drop the heatsink resistance below 2°C/W, or switch to active synchronous rectification using low-$R_{DS(on)}$ MOSFETs controlled by a dedicated synchronous rectifier driver IC. Always design your thermal margins assuming a 40°C to 50°C internal enclosure ambient, not the 25°C room temperature listed on the datasheet cover.
