The pattern is familiar: a camera works all day, then drops offline shortly after the infrared lights come on, recovers by morning, and passes every bench test. The camera is rarely at fault. What changed at night is the load — IR emitters and sometimes a heating element switch on and draw several times the daytime current — and on a long, thin DC cable that extra current pulls the voltage at the camera down past its brownout threshold. The device reboots, reboots again on reconnect, and the log fills with entries that look like a network problem.
The arithmetic nobody runs
Voltage drop is current times resistance, and cable resistance is set by the cross-section and the length of both conductors — out and back. A 12-volt camera drawing 300 mA by day and 900 mA with IR on, fed through thirty metres of thin cable, can lose one to two volts at night. Many cameras specify a 12-volt input with very little tolerance below it, so a supply set exactly at 12.0 volts leaves no headroom at all: the installation that measured fine at the adapter is starving at the camera. Doubling the conductor cross-section roughly halves the drop; halving the distance does the same. The numbers are simple, which is why it is strange that they are almost never checked.
Why PoE sidesteps the whole class of problem
Power over Ethernet delivers roughly 48 volts, so the same power arrives with a quarter of the current, and the voltage budget that a 12-volt run spends on cable resistance is spent on nothing at all. That is the real reason PoE installs are "more reliable" at distance — it is not the Ethernet, it is the voltage. Where PoE is not an option, the fixes in order of cost are: a thicker cable, a shorter run with the power supply moved closer to the camera, or a local supply at the camera fed by data-only cable. Setting the supply a little above nominal only helps if the camera tolerates the overvoltage, which is itself a spec question.
Diagnose it in ten minutes
Measure at the camera, not at the adapter, and measure under load: with IR forced on at midday via the app, read the voltage at the camera's terminal. If it sags below the camera's stated minimum, the case is closed. A second signature confirms it without a meter — the drop coincides exactly with night mode rather than with weather, traffic or time of day. Note that marginal cases hide: a camera may survive the first brownout and fail only when a motorised zoom parks, so test with every load switched on at once.
What to check before you buy
Three lines on a datasheet prevent this class of failure. The accepted input voltage range — a camera that runs from 9 to 15 volts has real headroom, one that says 12 volts only does not. The current draw with all loads on, not just the typical figure. And whether the vendor publishes a maximum cable length for the supplied adapter, which is the manufacturer admitting the drop exists. When a project runs power across a building, size the cable for the night load, not the day load.
XingLiao's camera range states input ranges and full-load current where they apply, and the camera manufacturer page covers power budget options for multi-camera builds. The factory and the contact form take questions about long-run power design.
Top comments (0)