One trades heat for noise. That sentence is the whole story of linear versus switching power supplies, because every difference between the two topologies traces back to whether they convert the line power with a switching element or without one. A linear supply is simple, quiet, and inefficient — most of the wasted energy leaves as heat. A switch-mode power supply (SMPS) is efficient, compact, and electrically noisy — the switching element that saves the heat creates ripple and electromagnetic noise that a linear supply never made. This article frames that core trade, works the efficiency math, explains the ripple-and-noise vocabulary, shows where linear still wins, and closes with a decision table for the applications that actually care.
One Trades Heat for Noise: The Core Trade
Every power supply trades something for efficiency, and the two families trade differently. The linear supply wastes the difference between the line voltage and the load voltage as heat, continuously. The switching supply saves that waste by chopping the power into pulses and converting them efficiently, but the chopping itself generates electrical noise. Neither is free; each topology is a different price tag.
The trade is visible in the box. A linear supply that delivers 1 A at 5 V from a transformer-based rectifier dissipates a large fraction of its input power as heat — it runs warm, needs ventilation, and its volume follows the transformer and heatsink. An SMPS delivering the same 5 V from a switching converter dissipates far less heat, runs cool, and packs the same wattage into a much smaller box — but it radiates switching noise, ripple, and harmonics that the linear supply never created. The SMPS guide and the isolated power supply article define the two families; this article isolates the one trade that explains every other difference.
The reason the trade explains everything is that the switching element changes the physics of the whole box. The transformer and rectifier in a linear supply run at 50/60 Hz; the switching converter in an SMPS runs at tens to hundreds of kilohertz. That frequency difference explains the size (higher frequency means smaller magnetics), the loss (less conduction waste), and the noise (the switching edges and their harmonics) all at once. Once the switch is in the design, everything else follows.
Efficiency Math: 30% vs 90% and What the Box Feels Like
The efficiency difference is the headline number, and the math is worth doing honestly. A linear supply rectifies the line and drops the voltage through regulation, wasting the drop as heat; a switching supply converts in steps that keep the loss small.
The rough numbers frame it. A small linear supply might be 30-50% efficient at full load — for a 5 V, 1 A output that is 5 W delivered and 5-10 W wasted as heat. An SMPS of the same output is typically 85-93% efficient, wasting 1 W or less. The difference is not cosmetic: it decides whether the box runs at 40 C or 80 C, whether a 60 W lab supply needs a fan, and whether a 500 W server supply can fit in a rack. The SMPS efficiency article and the server power supply article quantify the efficiency race at the power levels where it decides.
The interesting part of the math is where the lost energy goes. In a linear supply it is a steady heat that the box must shed; in an SMPS it is split between conduction loss, switching loss, and the recovery loss of the output rectifier — each with its own behavior and its own fix. The rectifier selection for SMPS article and the loss calculation approach show where the SMPS loss lives and how the parts list attacks it. The consumer-feel difference — the SMPS box is cool, small, and light — is the efficiency math made physical.
The efficiency difference also changes the economic math at scale, which is why the trade is not academic. A server room or an industrial bench running a hundred supplies multiplies the difference: ten points of efficiency at the same wattage means less heat to remove from the room, smaller air-conditioning, and a lower electricity bill every month of the year. The industry’s push toward high-efficiency supplies is not an aesthetic preference; it is the efficiency math compounding across thousands of operating hours. The data center power article and the server power supply article show how the efficiency marginal gains are valued in the facilities where they compound.
The same math explains why the SMPS dominates charging, tools, and anything battery-powered. A charger that wastes 50% of its input as heat cannot deliver fast charging in a confined space; an efficient SMPS can. A power tool with a linear supply would carry most of its weight as transformer and heatsink; the SMPS’s compact, cool converter is what makes modern high-power portables possible. The efficiency number is not a spec-sheet score; it is the physical constraint that decides what a product can be, and the trade with noise is the price of that capability.
Ripple and Noise: The Two Words on Every Switching Supply
The cost of the SMPS’s efficiency is the noise the switching creates, and two words carry most of the discussion: ripple and noise. Both are different things, and both matter differently to different loads.
Ripple is the periodic voltage or current variation on the SMPS output — the residual of the switching action that the output capacitors and the control loop could not fully smooth. It shows up as a ripple frequency and amplitude, and it is the property that a noise-sensitive load like an audio amplifier or a sensor rail reacts to. Noise is the wider category: the broadband and harmonic EMI the switching edges radiate, which can couple into nearby electronics and must be filtered or shielded. The ripple diagnosis guide treats both as a symptom chain, and the EMI filter design article covers the filter that tames them.
The design reality is that ripple and noise are dealt with by design, not by luck. The output rectifier’s speed, the output capacitor’s size, the filter’s cutoff, and the layout’s loop areas all set how much ripple and noise reach the load. A designer who understands the two words can budget for them; a buyer who only sees the efficiency number may connect a 90% supply to an audio amp and hear the switching. The trade is the point: the SMPS buys efficiency with noise, and the application decides whether the payment is acceptable.
Where Linear Still Wins: Audio, Sensors, and Lab Benches
The linear supply has not disappeared, because some loads care more about noise than about heat. In those applications the linear supply’s simplicity and silence are the deciding advantage, not a relic.
Audio is the classic case. An audio amplifier’s signal chain is sensitive to the ripple and switching noise an SMPS injects; a linear supply’s clean, quiet output is worth its heat and size in a high-fidelity system. Sensors and metrology are the second case: a measurement rail that needs a noise floor below what a switching supply produces will choose the linear answer for the bench even at the efficiency cost. Lab benches are the third: a lab supply is expected to be predictable and quiet, and its low noise floor matters more than the watt-hours. The isolated power supply article and the low noise design notes document where the quiet output wins.
The honest framing is that linear survives where the application’s primary need is electrical silence at modest power. The trade is not “old versus new”; it is heat versus noise, and the application decides which direction the balance should tip. A designer choosing for a lab does not need an SMPS’s watt-per-dollar; a designer choosing for a charger wants exactly that.
Choosing by Application: A Simple Decision Table
The decision closes with the application, and a table is the least ambiguous way to present it. The columns are the properties that matter — efficiency, noise, size, cost — and the rows are the application types.
| Application | Efficiency matters? | Noise matters? | Typical winner |
|---|---|---|---|
| Phone/PC charger | Yes | No | SMPS |
| Audio amplifier (hi-fi) | No | Yes | Linear |
| Sensor / metrology rail | No | Yes | Linear |
| Power tool / high-wattage | Yes | No | SMPS |
| Lab bench supply | Somewhat | Yes | Linear (or quiet SMPS) |
| Server / data center | Yes | Somewhat | SMPS |
The table is the trade made visible. Where efficiency and size dominate — chargers, tools, servers — the SMPS wins and its noise is managed with filtering. Where electrical silence dominates — audio, sensors, labs — the linear answer wins and its heat is accepted. The middle rows — a lab bench that wants both — are where the designer must pick a quiet SMPS and verify the ripple, not where one topology answers everything.
The trade you actually pay for is the core of the choice: one topology trades heat, the other traders noise, and the right pick is the one whose payment the application can afford. Understand the efficiency math, respect the ripple and noise vocabulary, keep the linear option for the quiet loads, and the decision table keeps the choice honest. The general rectifier category supplies the rectifier parts both topologies use, and the SMPS components article shows where the rectifier sits in the switching family.