Linear regulators when switching feels like overkill

The switching regulator reflex

We've been trained to reach for switching regulators. They're efficient, they step down or up, they handle wide input ranges. But for a home-lab project pulling 150 mA at 5V from a 9V wall adapter, a buck converter is honest overkill.

A linear regulator burns the voltage difference as heat. That's the tradeoff. In return you get zero switching noise, no inductor whine, and a parts count you can verify by looking at the board. For low-current weekend builds, that's a fair deal.

When linear makes sense

If your load is under 500 mA and your input-output delta is reasonable—say, 12V to 5V or 9V to 3.3V—a linear regulator will work without drama. The key number is power dissipation: (Vin - Vout) × Iload. Keep that under 1W without a heatsink, under 3W with a modest clip-on tab.

I use linear regulators for analog front-ends that hate ripple, for prototypes where I want to scope signals without fighting 500 kHz hash, and for anything where the enclosure is already tight and I don't want to EMI-test a switching layout. If you've fought grounding loops on a mixed-signal board, you know what clean rails buy you.

Thermal design you'll actually do

The 7805 in a TO-220 package can dissipate about 1W into free air if you're not aggressive. Add a small heatsink and you're at 2-3W comfortably. For anything above that, consider an LDO in a bigger package or a pre-mounted tab regulator on a PCB with a copper pour acting as a spreader.

I lay out a solid ground plane under the regulator and stitch it with vias. The thermal pad connects to that plane, which doubles as electrical ground and a heat sink. It's the same attention to thermal relief you'd give a high-current trace, just inverted: you want the thermal coupling here.

Run the regulator warm to the touch in final testing. Not too hot to hold, but warm enough that you know it's working. That's your margin check without a thermal camera.

Capacitor placement that matters

Every linear regulator datasheet shows input and output caps. Use them. I keep 10 µF ceramics and 100 nF ceramics on hand—one of each on input and output, placed tight to the regulator pins. The 100 nF handles high-frequency transients; the 10 µF gives you bulk stability.

If your input comes from a wall adapter through a couple feet of wire, add another 100 µF electrolytic at the input. Long wires look inductive at the frequencies where your regulator's feedback loop operates. The extra capacitance keeps the loop stable when your load steps.

When to switch anyway

If you're pulling more than 500 mA, or your input is more than 6V above your output, or you're running off a battery and efficiency matters, use a switcher. Don't fight the physics. But for the three-sensor rig on your desk that runs off a 12V brick and needs clean 5V, a 7805 and four capacitors will outlive the project.

Linear regulators are not a compromise. They're a tool that fits certain problems exactly. Build the power supply the weekend project actually needs, not the one that sounds impressive in the commit message.