An unregulated power supply consists of a step-down transformer, a bridge rectifier, and a bulk filter capacitor, delivering raw DC that fluctuates with line voltage and load current. You use it when your load can tolerate voltage variance—such as high-current inductive loads (relays, solenoids, DC motors, contactors) or basic lead-acid battery charging. By skipping the voltage regulator stage, you eliminate the massive heat dissipation and cost associated with dropping 30V+ down to a precise 24V or 12V rail at high currents.
The Case for Raw DC: When Unregulated Wins
In modern electronics, we default to switching regulators for everything. But if you are driving a bank of 24VDC industrial relays (like the Omron MY4N) or a heavy-duty solenoid valve, the coil does not care if the supply voltage sags from 31V at no-load to 26V under full pull-in current. In fact, many relay coils are rated for a 10% to 15% voltage tolerance.
When you attempt to force a clean, regulated 24.0V onto a 3A relay bank using a linear regulator, you are burning money and managing severe thermal issues. An unregulated linear supply runs dead-quiet at high frequencies, lacks the electromagnetic interference (EMI) headaches of a cheap buck converter, and survives short-circuit faults far better than a switching IC that will instantly latch off or blow its internal MOSFET.
Topology Showdown: Unregulated vs. Regulated vs. Switching
Before laying out the PCB or wiring the terminal blocks, you need to know exactly what you are sacrificing by dropping the regulator. Here is how the three primary topologies compare for a nominal 24V, 3A DC load.
| Topology | Efficiency | Heat Dissipation | Output Ripple & Noise | Relative Cost | Best Use Case |
|---|---|---|---|---|---|
| Unregulated Linear | ~65-75% | Low (Transformer only) | High 120Hz ripple (~2.5V p-p), zero HF noise | $ (Cheapest) | Relays, solenoids, heaters, motors |
| Regulated Linear (e.g., LM317/LM7824) | ~40-50% | Extreme (15W+ at 3A) | Very Low 120Hz ripple (<50mV), zero HF noise | $$ (Heatsink cost) | Precision analog, audio, sensors |
| Switching (e.g., Buck / Forward) | 85-92% | Very Low (<2W) | Low LF ripple, High HF noise (50-100mV at 100kHz+) | $$$ (Complex BOM) | Microcontrollers, digital logic, high-efficiency needs |
The data makes the decision clear: if your load is purely inductive or resistive and doesn't require tight voltage regulation, the unregulated topology wins on cost, simplicity, and high-frequency noise performance.
Bench Build: 24V Unregulated Supply for Inductive Loads
Let’s design a robust 24V nominal unregulated supply capable of delivering 3A continuous current. We will use this to drive a control panel of contactors and indicator lamps.
Input/Output Specifications
- Input: 120VAC nominal, 60Hz (Acceptable range: 114VAC to 126VAC)
- Target Output: ~26VDC under 3A full load
- No-Load Output: ~32VDC
- Maximum Ripple: < 3.0V peak-to-peak at full load
Bill of Materials & Part Values
- Transformer: Hammond 165L24 (24VAC RMS, 4A, 100VA). We oversize the VA rating by 25% to account for rectifier losses and capacitor charging spikes.
- Bridge Rectifier: KBPC5010 (50A, 1000V). Overkill for 3A, but at $3, it runs stone-cold without a heatsink and handles the repetitive peak surge currents of charging the bulk capacitor.
- Filter Capacitor: Nichicon LNR1H103MSE (10,000µF, 50V, 105°C rated, snap-in). The 50V rating provides a 35% safety margin over the 32V no-load peak.
- Bleeder Resistor: 10kΩ, 2W metal oxide (discharges the cap safely when unplugged).
The Math: Peak Voltage and Regulator Dropout
Why not just add an LM317 or LM7824 to clean up the output? Let's look at the headroom and thermal math. The peak DC voltage from a 24VAC RMS transformer is calculated as:
V_peak = (V_RMS × √2) - (2 × V_diode_drop)
V_peak = (24 × 1.414) - (2 × 1.1V) = 33.9V - 2.2V = 31.7V
If you use a linear regulator to drop 31.7V down to a clean 24.0V at 3A, the regulator must dissipate the difference as heat:
Power_Dissipated = (31.7V - 24.0V) × 3A = 23.1 Watts
Dissipating 23W requires a massive extruded aluminum heatsink with forced air cooling to keep the silicon junction under its 125°C thermal shutdown limit. Furthermore, the LM7824 requires a minimum 2V dropout voltage. If your AC mains sags to 110VAC, your transformer output drops, your capacitor ripple valley dips below 26V, and the regulator drops out of regulation, passing raw 120Hz ripple directly to your load. By staying unregulated, you eliminate the 23W heat problem entirely.
Never use a 35V rated capacitor on a 24VAC transformer. Mains voltage can legally fluctuate up to +10% (132VAC), which pushes the transformer output to 26.4VAC. The peak DC voltage would then be 35.1V, exceeding the capacitor's dielectric limit and causing catastrophic venting. Always use a 50V rated capacitor for 24VAC systems.
Input Protection, Inrush, and Thermal Derating
An unregulated supply lacks the active foldback current limiting found in switching ICs. If the output shorts, the transformer will dump maximum current until it overheats or the primary fuse blows. Protection must be designed into the passive components.
Fusing and Surge Protection
On the AC primary side, use a 1A time-delay (slow-blow) fuse. Transformers draw a massive inrush current (up to 10x nominal) for the first few AC cycles as the magnetic core magnetizes. A fast-blow fuse will nuisance-trip on every power-up. On the DC secondary side, place a 5A fast-blow fuse directly after the bridge rectifier to protect the wiring and capacitor from a dead short.
For transient voltage suppression, wire a Metal Oxide Varistor (MOV) like the Littelfuse TMOV20RP300E across the primary winding. This clamps lightning-induced or grid-switching spikes above 300V, protecting the transformer insulation and the bridge rectifier diodes.
Managing Inrush Current
When power is applied, the 10,000µF filter capacitor looks like a dead short until it charges to the peak voltage. This causes a massive current spike through the bridge rectifier diodes. While the KBPC5010 can handle short surges, repeated hot-switching can degrade the silicon. If your bulk capacitance exceeds 10,000µF, or if you are feeding this from a sensitive 15A branch circuit shared with other equipment, add an NTC inrush current limiter (like the Ametherm SL32 2R015) in series with the AC primary.
Thermal Derating of the Filter Capacitor
The bulk capacitor does more than just store energy; it must absorb the high-frequency RMS ripple current generated by the rectifier charging pulses. For a 3A DC load, the RMS ripple current through the capacitor is roughly 2A. Check the datasheet for the capacitor's ripple current rating. If the capacitor sits too close to the transformer (which runs hot), the ambient temperature rises. Electrolytic capacitor lifespan halves for every 10°C increase above its rated temperature. Keep the capacitor physically separated from the transformer core, or use a 105°C rated part instead of the cheaper 85°C variants.
Ripple Expectations and the Linear vs. Switching Verdict
The primary drawback of an unregulated supply is ripple. Because the capacitor only charges at the very peaks of the AC sine wave, it discharges into the load for the remainder of the cycle. We calculate the peak-to-peak ripple voltage using the standard approximation:
V_ripple(p-p) = I_load / (f × C)
For our 3A load, full-wave rectified 60Hz mains (f = 120Hz), and 10,000µF (0.01F) capacitance:
V_ripple = 3 / (120 × 0.01) = 2.5V peak-to-peak.
This means your DC output will sawtooth between roughly 29.2V and 26.7V under full load. For a relay coil, this 2.5V variance is entirely invisible; the magnetic field remains stable enough to hold the contacts closed without chatter.
Linear vs. Switching: Which is right for this load?
If you were to replace this unregulated linear build with a DIN-rail switching supply (like a Mean Well DR-60-24, which costs around $28 in 2026), you would get tight 24V regulation and 90% efficiency. So why build the unregulated linear version?
- High-Frequency Noise: Switching supplies operate at 60kHz to 150kHz. Even with good filtering, they inject high-frequency common-mode noise into the ground plane. If your relay bank is switching near sensitive analog sensors, thermocouples, or audio preamps, that 100kHz switching noise will couple into your signal lines. The unregulated linear supply has zero high-frequency noise; its only ripple is a benign 120Hz hum that is easily filtered by analog circuits.
- Fault Tolerance: If a solenoid valve jams and draws a 10A stall current, a switching supply will instantly detect the overcurrent and shut down, dropping your entire control panel offline. The unregulated supply's heavy iron transformer will simply sag the voltage, allowing the solenoid to pull hard for several seconds before the secondary fuse blows, potentially clearing the mechanical jam.
- Repairability: A blown KBPC5010 bridge or a vented capacitor can be diagnosed with a $15 multimeter and replaced with a soldering iron. A failed switching IC inside a potted Mean Well brick requires total unit replacement.
Choose the unregulated linear topology when your load is heavy, inductive, and tolerant of voltage sag, and when your environment demands low high-frequency EMI. Choose switching when you are powering microcontrollers, digital logic, or when your input source is a DC battery bank rather than an AC mains transformer.
