PLC I/O Power Budget Calculator

Two supplies, two budgets

A modular PLC rack has two entirely separate power questions, and panel designs routinely answer only the second.

The backplane. Every module draws logic current from the rack, supplied by the CPU or a rack power supply — typically a few amps at 5 V in total. Fill a rack with analog and communication modules and you can exhaust it well before you run out of slots. The symptom is not a clean failure: it is modules dropping offline intermittently, usually under load, usually blamed on the modules.

The 24 V field supply. Separate, external, and feeding the sensors, solenoids, relays and HMI. This is the one everybody sizes — and usually the one they size by adding up nameplates and rounding up.

Diversity applies here too

The default figures assume every output is on at once. That is the right assumption for a supply that must not brown out, and the wrong one for estimating heat, because the panel does not sit in that state all day.

Size the supply for the worst case — everything energised, plus headroom. Estimate the heat from the realistic average. On a machine where most outputs are momentary, those two numbers can differ by a factor of three.

Sixteen digital inputs at 7 mA each is 112 mA per card and genuinely continuous — inputs are usually on. Sixteen solenoid outputs at 200 mA each is 3.2 A and almost never all at once.

Headroom, and what it is for

25 % spare is the conventional allowance, and it buys three different things:

• Inrush. Capacitive loads, and particularly a second DIN-rail supply or a DC-DC converter downstream, draw a large surge at power-up. A supply sized exactly to the steady load can go into current limit and never start.
• Temperature. Most DIN-rail supplies are rated at a reference ambient and derate above it — often from 50 °C or 55 °C. A supply in a hot panel is a smaller supply.
• The modification. Somebody will add two sensors and a relay next year without recalculating anything.

Beware stacking allowances, though. If your module figures were already worst-case, adding 25 % on top of a worst case that never occurs buys an oversized supply that runs at 20 % load — where its efficiency is poorest.

Heat is the output nobody asks for

Everything in the panel that consumes power and does not send it somewhere else turns it into heat. The two contributions:

• The supply's own loss. At 88 % efficiency, a 100 W load costs about 14 W of heat in the enclosure.
• Whatever is dissipated inside. A solenoid on the machine dissipates its heat on the machine. A relay, an indicator lamp or a barrier in the panel dissipates it in the panel. The share matters, which is why it is an input here.

That total feeds directly into the enclosure sizing — and into the ratings of everything else inside it. A sealed IP65 enclosure cannot shed heat, so the busbars and cables inside it derate; the numbers are in IP ratings for panels and the derating is applied by the busbar calculator.

Practical points that are not in the arithmetic

• Separate the supplies. One 24 V supply for I/O and a separate one for solenoids and contactor coils. Inductive switching noise on a shared rail causes analog readings to jump and inputs to false-trigger, and it is a miserable fault to chase.
• Fuse or protect each branch. A shorted field cable should drop one circuit, not the whole rack.
• Watch the voltage at the far end. 24 V down 60 m of thin cable to a remote sensor is not 24 V when it arrives. The drop arithmetic is the same as anywhere else — voltage drop calculator.
• Decide what a power failure should do. If the PLC must ride through a brief dip, that is a buffer module or a small DC UPS, sized from the load and the hold-up time you need — the battery calculator does that arithmetic for low-power DC loads as well as for inverter banks.
• Earthing and screens. 0 V reference, screen termination and the panel earth bar are a design decision, not something to leave to the wireman.

Inrush, and the supply that browns out at switch-on

A power budget built from steady-state consumption describes the panel a second after it energises. The first hundred milliseconds are a different circuit entirely.

Almost everything on a 24 V rail presents a capacitive input. Every module, every distributed I/O node and every field device has bulk capacitance across its supply, and at switch-on all of it charges at once. A supply feeding thirty such loads can see an instantaneous demand many times its rating — brief, but long enough for a switch-mode supply to detect an overload and either fold back or enter hiccup mode, retrying every few hundred milliseconds and never getting the rail up.

The failure looks nothing like an overload. The panel simply refuses to start, or starts inconsistently, and every measurement taken afterwards shows plenty of headroom. A supply chosen with 25 per cent steady-state margin can still fail to start the load it was sized for. Supplies that state a peak or startup current capability, or a slow soft-start on the larger loads, are what solve it — and staging the field supplies behind a timed contactor solves it on panels that are already built.

Volt drop on the 24 V rail

A 5 per cent drop on 415 V is 20 V and nobody notices. The same 5 per cent on a 24 V rail is 1.2 V, and the equipment at the far end has a much narrower window to lose it from — most 24 V devices specify 20.4 to 28.8 V, so the budget from nominal down to the lower limit is only 3.6 V in total.

Distributed I/O is where this appears. A remote node 40 metres away drawing 2 A through 1.0 mm² conductors loses roughly 1.4 V in the outward leg and the same again returning — comfortably outside the window before any of the devices on that node have been considered. Sensors misread, digital inputs become intermittent, and a communication module drops off the bus under load and recovers when the load stops.

Two habits fix it. Size 24 V distribution conductors for volt drop rather than for current, which usually means 1.5 or 2.5 mm² for anything leaving the panel. And where a run is genuinely long, put a local supply at the far end fed from 230 V rather than pushing 24 V down the cable — it is cheaper than the copper and it removes the problem instead of managing it.