At the most fundamental level, a rectifier is a semiconductor device—typically a diode or a bridge of diodes—that converts alternating current (AC) into direct current (DC) by enforcing unidirectional current flow. But if you are asking 'what are rectifiers' in the context of building a bench power supply or an embedded system's power rail, the answer goes deeper. A rectifier alone only produces pulsating DC. To get a stable, usable voltage, you must pair the rectifier stage with filtering and a regulation topology.

Designing an AC-to-DC converter requires answering three critical questions: Which rectifier topology minimizes loss? How much bulk capacitance is needed to survive line sags? And finally, should you burn off the excess voltage in a linear regulator or chop it down with a switching converter? Let us run the math on a concrete 120VAC to 12VDC, 2A design to find out.

Rectifier Topologies: Efficiency, Heat, Noise, and Cost

Before we regulate the voltage, we must rectify it. The topology you choose for the AC-to-DC conversion stage dictates your baseline efficiency and transformer requirements. While half-wave rectification exists in theory, it is practically obsolete for power supplies due to transformer core saturation and massive ripple. The real choice is between a standard full-wave bridge and an active (synchronous) rectifier.

Topology Efficiency (at 2A) Heat Dissipation Output Noise / Ripple Component Cost
Full-Wave Bridge (Silicon Diodes) ~94% (rectifier stage only) ~2.8W (1.4V drop @ 2A) 120Hz fundamental ripple $0.50 - $1.50
Full-Wave Bridge (Schottky Diodes) ~97% (rectifier stage only) ~1.2W (0.6V drop @ 2A) 120Hz fundamental ripple $1.50 - $3.00
Active / Synchronous (MOSFETs) >99% (rectifier stage only) <0.2W (Rds_on losses) 120Hz fundamental + switching noise $4.00 - $8.00 + driver IC
Bench Tip: For loads under 5A, the cost and board-space penalty of an active rectifier rarely justifies the fraction of a watt saved over a standard silicon bridge. Stick to a robust silicon bridge like the GBJ2510 and manage the 1.4V drop in your transformer selection.

Design Example: 120VAC to 12VDC at 2A (The Raw DC Stage)

Let us design the front end for a 12V, 2A continuous load. We need a transformer, a bridge rectifier, and a bulk filter capacitor.

1. Transformer Selection: We need enough peak voltage to maintain headroom for the regulator after accounting for diode drops and ripple. A standard 18VAC RMS secondary is ideal.
Peak Voltage (V_peak) = 18V * 1.414 = 25.45V.

2. Rectifier Pick: We will use a GBJ2510 (25A, 1000V silicon bridge). It is heavily over-specced for 2A, which is exactly what we want. Running a 25A bridge at 2A keeps the forward voltage drop closer to 1.0V per diode pair rather than 1.4V, and it eliminates the need for a heatsink on the bridge itself.
DC Bus Voltage (no load) = 25.45V - 1.0V = 24.45V.

3. Filter Capacitor Math: The capacitor must supply the 2A load during the valleys between the 120Hz AC peaks. We will target a maximum ripple of 4V to ensure we do not drop below the regulator's minimum input voltage.
C = I / (f * V_ripple)
C = 2A / (120Hz * 4V) = 0.00416 Farads.
We select a standard 4700µF, 35V electrolytic capacitor (e.g., Panasonic EEU-FR1V472).

With a 4700µF cap, our actual ripple is roughly 3.54V. Our minimum DC bus voltage under full load will be 24.45V - 3.54V = 20.91V.

Linear vs. Switching Regulation: Dropout Math and Thermal Derating

Now we have a raw DC bus swinging between 20.91V and 24.45V. We need a clean 12V. This is where the linear versus switching decision is made. Let us look at the headroom math and thermal reality for both.

The Linear Route: LT1084-12

The LT1084-12 is a 5A low-dropout (LDO) linear regulator. Its maximum dropout voltage is 1.5V. Since our minimum input is 20.91V, we have 8.91V of headroom—plenty to maintain regulation.

However, look at the thermal math. A linear regulator acts as a variable resistor, burning excess voltage as heat.
Power Dissipated = (V_in_avg - V_out) * I_load
P = (22.68V - 12V) * 2A = 21.36 Watts.

Thermal Derating Warning: The LT1084 in a TO-3P package has a junction-to-ambient thermal resistance of roughly 40°C/W without a heatsink. Dissipating 21.36W will raise the junction temperature by 854°C, instantly triggering thermal shutdown and likely destroying the silicon. You would need a massive extruded aluminum heatsink (approx. 2°C/W) and forced air to keep this alive. For a 2A load with high input-to-output differential, linear regulation is a thermal nightmare.

The Switching Route: TPS5430DDA

The TPS5430DDA is a 5.5V to 36V input, 3A step-down (buck) switching converter. Instead of burning the excess voltage, it chops the input using a MOSFET and an inductor, transferring only the energy required by the load.

At 12V/2A, the TPS5430 operates at roughly 88% efficiency.
Input Power = (12V * 2A) / 0.88 = 27.27W.
Power Dissipated (Heat) = 27.27W - 24W = 3.27 Watts.

While 3.27W still requires thermal management (the TPS5430 uses a PowerPAD exposed pad soldered to a copper pour on the PCB), it is entirely manageable without external extruded heatsinks. The PCB copper acts as the heatsink.

Protection, Input Range, and Ripple Expectations

A robust power supply design must account for fault conditions and output noise. According to standard power design practices outlined by All About Circuits, protection at the input and filtering at the output are non-negotiable.

Input Protection: Place a 1A slow-blow fuse on the 120VAC primary side of the transformer. On the DC bus, immediately after the bridge rectifier, place an SMBJ24A Transient Voltage Suppressor (TVS) diode. If a mains surge couples through the transformer, the TVS will clamp the DC bus to 24V, sacrificing itself to save the downstream regulator.

Ripple and Noise Expectations: If you had used the linear LT1084, its Power Supply Rejection Ratio (PSRR) would have attenuated the 120Hz input ripple by roughly 60dB, yielding an exceptionally clean, low-noise 12V output (microvolts of ripple). This is why linear regulators are still used in audio and precision ADC circuits. The switching TPS5430, however, introduces high-frequency switching noise (at its 500kHz switching frequency). You can expect 20mV to 40mV of peak-to-peak output ripple. To mitigate this for sensitive loads, add a small LC pi-filter (a 10µH inductor and a 22µF ceramic capacitor) at the output of the buck converter.

The Decision Path: Which Architecture Wins?

Do not default to 'it depends.' Use this decision matrix to lock in your topology based on your specific load parameters.

Condition / Constraint If True... Recommended Topology
Load current is < 100mA AND V_in - V_out < 3V Thermal dissipation is under 0.3W. Noise must be minimal. Linear LDO (e.g., AP2112K-3.3)
Load is audio, RF, or precision analog (requires <1mV ripple) Switching noise will ruin signal integrity. Heat is secondary. Linear Regulator with heavy heatsinking
V_in - V_out > 5V AND Load current > 500mA Linear dissipation will exceed 2.5W, requiring bulky heatsinks. Switching Buck Converter
Battery powered or strict energy efficiency compliance (e.g., DoE Level VI) Linear quiescent current and resistive losses will fail efficiency mandates. Synchronous Switching Converter

The Final Pick for Our 12V/2A Design

For our specific scenario—stepping a 20.9V-24.4V raw DC bus down to 12V at 2A—the differential voltage is too high and the current is too large for a linear regulator to handle without absurd thermal overhead.

The Concrete Pick: Use the TPS5430DDA buck converter. Pair it with the GBJ2510 bridge rectifier and a 4700µF bulk capacitor. This combination yields a highly reliable, 88% efficient power supply that keeps component temperatures well within safe operating areas, requires no external heatsinks, and easily handles the 24W total throughput. For further optimization on high-current synchronous designs, reference the Texas Instruments Power Management design guides to refine your PCB layout and minimize EMI radiation from the switching node.