Power hardware
Choosing a small DC supply by measured load, not label current
I needed one 12 V supply for a small monitoring enclosure containing two 120 mm fans, an ESP32 board, a relay board, and several sensors. Adding the printed maximum currents suggested nearly 3 A, which made a 5 A supply look like the obvious choice. Measurement showed that the real continuous load was much smaller.
The fans were the only meaningful 12 V loads. The controller and sensors used a small downstream converter. I measured current at the supply output with three operating states: idle, normal ventilation, and both fans starting at full duty. The start value was captured with a current logger rather than a slow handheld display.
| State | Current | Power | Duration |
|---|---|---|---|
| Controller active, fans stopped | 0.11 A | 1.3 W | continuous |
| Fans at normal duty | 0.74 A | 8.9 W | continuous |
| Both fans at 100% | 1.16 A | 13.9 W | continuous |
| Startup peak | 1.83 A | 22.0 W | < 0.3 s |
A good target therefore was not 12 V × 5 A. I wanted a supply that could handle the short 1.8 A peak without a voltage dip, run below roughly 60–70% of its rated output during the worst continuous state, and stay reasonably efficient around 5–15 W where the enclosure spends most of its time.
I compared a compact 24 W unit and a 36 W unit. Both survived the startup event, but the smaller one reached 57 °C on its case after two hours at 1.16 A in the closed enclosure. The 36 W unit stabilized at 44 °C under the same conditions. Its no-load draw was also slightly lower in my sample, about 0.19 W versus 0.27 W.
The larger model therefore won even though its nameplate capacity looked excessive. The reason was thermal margin inside the enclosure, not the desire for a round “50% spare capacity” rule. If airflow or ambient temperature changed, I would repeat the temperature test rather than assume the same margin still applies.
Wiring detail
The 12 V output goes first to a fused distribution block. Each fan branch has its own removable connector. The downstream 5 V converter has a separate 0.75 A fuse. Grounds remain common at the distribution block, which makes current measurements and fault tracing much easier than chaining device grounds from board to board.