Sources<\/a><\/li>\n<\/ul>\n<\/nav>\n<\/div>\nDirect answer: what matters when powering an industrial sensor?<\/h2>\n Four decisions, in this order: voltage at the sensor, power headroom, output noise, and environmental rating.<\/strong> Most sensor power problems trace back to one of them, and they are cheaper to get right on paper than to debug on a machine that is already built.<\/p>\n\nVoltage:<\/strong> 24V is the industrial default, but what matters is the voltage that arrives at the sensor after cable drop, not the voltage at the adapter terminals.<\/li>\nPower:<\/strong> size for inrush and peak, then add 30–50% headroom. A supply running continuously at its rating in a warm cabinet is a supply with a short life.<\/li>\nNoise:<\/strong> switching ripple couples straight into analogue front ends. If your sensor resolves millivolts, ripple is a measurement problem, not a power problem.<\/li>\nEnvironment:<\/strong> match the IP rating to the cleaning method and the ambient temperature, and demand a derating curve rather than a single wattage.<\/li>\n<\/ul>\nPart 1: voltage — what actually reaches the sensor<\/h2>\n Industrial sensors overwhelmingly run on 24V DC, with 12V and 5V common for smaller devices and for internal logic. The specification error is not choosing the wrong nominal voltage; it is assuming the nominal voltage arrives intact.<\/p>\n
Copper resistance over a long cable run quietly converts voltage into heat. On a 24V rail with a modest sensor tolerance band, a couple of volts of drop is the difference between a machine that runs and one that intermittently faults at the far end of a conveyor.<\/p>\n
\n\n\nCable run<\/th>\n 0.5 mm² at 0.5A<\/th>\n 0.75 mm² at 1A<\/th>\n 1.5 mm² at 2A<\/th>\n<\/tr>\n<\/thead>\n \n\n10 m<\/td>\n 0.4V drop<\/td>\n 0.5V drop<\/td>\n 0.5V drop<\/td>\n<\/tr>\n \n25 m<\/td>\n 0.9V drop<\/td>\n 1.2V drop<\/td>\n 1.2V drop<\/td>\n<\/tr>\n \n50 m<\/td>\n 1.8V drop<\/td>\n 2.4V drop<\/td>\n 2.4V drop<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\nFigures are approximate, for copper conductors at ambient temperature, and count the round trip through both conductors. Three practical responses:<\/p>\n
\nIncrease conductor size.<\/strong> Cheapest fix at design time, expensive after installation.<\/li>\nRaise the distribution voltage and regulate locally,<\/strong> converting 24V to 5V or 12V at the sensor cluster rather than at the cabinet.<\/li>\nUse a supply with remote sense,<\/strong> so the adapter regulates at the load rather than at its own terminals. This works well but needs the sense leads routed correctly, or it will make regulation worse rather than better.<\/li>\n<\/ol>\nCheck the sensor’s real input range, not its nominal rating<\/h3>\n A device sold as “24V” often accepts something like 10–30V DC, while another accepts 24V ±10%. Those are very different tolerances once cable drop and supply tolerance stack up. Add the worst-case low line, the worst-case load, and the worst-case drop before you conclude the design has margin.<\/p>\n
Part 2: power — sizing for peaks, not averages<\/h2>\n Sum the steady-state draw, then find the worst transient, then add headroom. The transient is where designs fail: a smart camera or a fieldbus coupler can draw two to three times its running current for tens of milliseconds at power-up, and several devices energising together multiply that.<\/p>\n
\n\n\nDevice<\/th>\n Typical draw at 24V<\/th>\n Sizing note<\/th>\n<\/tr>\n<\/thead>\n \n\nInductive \/ capacitive proximity<\/td>\n 10–200 mA<\/td>\n Small individually, adds up across a machine<\/td>\n<\/tr>\n \nPhotoelectric sensor<\/td>\n 20–100 mA<\/td>\n Watch emitter switching pulses<\/td>\n<\/tr>\n \nUltrasonic sensor<\/td>\n 30–80 mA<\/td>\n Sensitive to ripple on the rail<\/td>\n<\/tr>\n \nLaser distance \/ LiDAR module<\/td>\n 100–500 mA<\/td>\n Heater option can double the draw in cold stores<\/td>\n<\/tr>\n \nSmart camera \/ vision sensor<\/td>\n 5–15W<\/td>\n 2–3x inrush; often the dominant load<\/td>\n<\/tr>\n \n4–20 mA loop transmitter<\/td>\n Loop current plus ~0.5–1W<\/td>\n Loop supply stability affects accuracy<\/td>\n<\/tr>\n \nEncoder<\/td>\n 50–150 mA<\/td>\n Noise-sensitive on long cable runs<\/td>\n<\/tr>\n \nFieldbus coupler \/ IO block<\/td>\n 2–10W<\/td>\n Energises everything downstream at once<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\nWorking rule:<\/strong> take the sum of steady draws, add the largest single peak, and then add 30–50% on top. It is cheaper to buy a 96W supply instead of a 60W one than to chase intermittent resets in a commissioned machine.<\/p>\nAutomation cabinets concentrate sensors, vision payloads and fieldbus modules on one rail \u2014 which is why peak, not average, sets the rating.<\/figcaption><\/figure>\nPart 3: noise — the part that gets ignored until measurements drift<\/h2>\n A switching adapter is a noise source by construction. That is fine for a proximity switch and quite possibly not fine for a strain-gauge bridge, a thermocouple front end, or a precision 4–20 mA loop.<\/p>\n