Picking a loop power supply is not guesswork once you know the two budgets that govern it. A voltage budget decides whether the supply is high enough to feed every transmitter through all the drops in its loop, and a current budget decides whether one supply feeding many loops can deliver enough milliamps at once. This guide is a practical walkthrough: how to add up the numbers on both sides, where the common mistakes hide, and how to leave the margin that keeps a supply working when the site grows or the weather turns.
Loop Power Supply Sizing in one line: Sizing a loop power supply means checking two budgets. The voltage budget confirms the supply exceeds the sum of each transmitter's compliance voltage plus every series drop - load resistor, barrier, isolator, and cable - at 20 mA, with margin. The current budget confirms that a shared supply can deliver the combined draw of every loop it feeds at full scale simultaneously. Get both right, add margin, and 24 VDC covers the great majority of loops.
Start every sizing exercise from the transmitter's compliance voltage - the minimum it needs across its own terminals - because that is the floor the whole loop is built to protect. Then walk the loop and add every series voltage drop that occurs at 20 mA, since 20 mA is the worst case where drops are largest. Include the load or sense resistor (a 250 ohm HART resistor is 5 volts on its own at 20 mA), the internal resistance of any intrinsic-safety barrier, the input resistance of a signal isolator or a series panel indicator, and the resistance of the cable, counted for both the out and the return conductor.
Sum those drops, add the transmitter's compliance voltage, and you have the minimum supply voltage that loop can accept. For a typical loop the arithmetic is quick: a transmitter needing perhaps 10 to 12 volts, a 250 ohm resistor at 5 volts, a couple of volts for a barrier and a long cable, and you are already near 18 to 20 volts - which is exactly why 24 VDC is the default, since it clears that total with room to spare. If several different loops share one supply, the voltage budget is set by the most demanding loop, not the average, because the supply must satisfy the hungriest one.
Where the voltage budget bites is on the loops that grew after commissioning. A barrier added for a hazardous-area upgrade, a local indicator wired in series, or a cable extended to a relocated instrument each raises the total drop and can push a once-comfortable loop over its supply. The discipline is to redo this sum whenever a loop changes, and to design with margin - a few volts above the calculated minimum - so that the next addition does not silently starve the transmitter at the top of its range.
A single supply that feeds one loop only ever has to source up to 20 mA, so its current rating is trivial. The current budget matters when one supply feeds many loops, which is the common case in a marshalling cabinet or RTU panel. Here you total the worst-case current draw of every loop the supply serves and confirm the supply can deliver that sum at once. Each 4-20 mA transmitter can draw up to about 20 mA, and separately powered devices on the same rail draw whatever their supply spec lists, so the total can climb quickly on a busy panel.
The trap is assuming loops draw their average rather than their peak. Every loop can, in principle, sit at 20 mA simultaneously - an upset that drives many process values high at once is exactly when you least want the supply to sag - so size for the case where all loops are at full scale together, not for a typical mid-range draw. Add the quiescent draw of any signal conditioners, isolators, or indicators powered from the same rail, since those consume current whether or not the loops are active. The result is the minimum current the supply must deliver.
As with voltage, leave headroom. A supply loaded to its exact rated maximum has nothing left for a future loop, a warm-day derating, or the inrush of powering everything up at once, and it runs hot, which shortens its life. A common rule of thumb is to load a shared supply to a comfortable fraction of its rating and keep the balance as margin. That margin is what lets a panel absorb the next instrument without a supply swap, and it is far cheaper to buy at design time than to retrofit after a supply trips on a cold morning when everything drew inrush together.
Real sites push both budgets at once, and the two interact. Raising the supply voltage to satisfy a demanding loop also raises the power the supply dissipates when many loops draw current, which feeds back into thermal derating and thus into the current budget. The practical method is to fix the supply voltage first from the worst-case voltage budget - almost always landing on 24 VDC - then size the current rating from the total peak draw with margin, then confirm the chosen supply's derating at the site's ambient temperature still leaves that margin intact. Enclosed panels in the sun run hot, and a supply rated at benign lab temperature may deliver less in a field cabinet.
Shared supplies also raise a reliability question that sizing alone does not answer: one supply feeding a whole panel is a single point of failure for every loop on it. Where a site cannot tolerate losing all its instruments at once, the answer is a redundant supply pair with diode-OR coupling, or splitting the loops across two supplies so a single failure takes out only half the points. That decision is separate from the voltage and current arithmetic but belongs in the same sizing conversation, because it changes how much current each supply must carry and how much margin each needs to cover for the other.
For a SCADA or remote-telemetry site the power design has to survive on its own between visits, so margin is not a luxury. A cloud SCADA platform such as Merobix reads the field values through the site's RTU or edge device, and those field loops are only as reliable as the supply feeding them - a starved or overloaded supply shows up not as a power alarm but as instrument readings that clip, drift, or drop out. Trending every channel makes that visible: a cluster of points on one panel misbehaving together, especially under high process load, points back to a shared supply running out of budget rather than to a coincidence of failing sensors. Sizing the supply with real voltage and current margin at the outset is what keeps those remote sites reporting clean data without a maintenance trip.
For most loops, 24 VDC, because it clears the typical voltage budget with margin. Add the transmitter's compliance voltage to every series drop at 20 mA - load resistor, barrier, isolator, and cable - and 24 volts almost always exceeds that total. Only unusually demanding loops, with several barriers or very long cable, might need a check to confirm 24 volts still leaves headroom; short simple loops could run on less but rarely benefit from doing so.
Size it for the worst case: every loop it feeds sitting at 20 mA simultaneously, plus the quiescent draw of any isolators, conditioners, or indicators on the same rail. Do not size for average draw, because an upset can drive many values high at once. Add margin on top so the supply is not loaded to its exact rating, which leaves room for future loops, thermal derating in a hot panel, and startup inrush.
One shared supply is efficient but becomes a single point of failure for every loop on it. If a site cannot tolerate losing all its instruments at once, use a redundant supply pair with diode-OR coupling, or split the loops across two supplies so one failure takes out only half. This choice affects sizing, because each supply must then carry more current and hold more margin to cover for the other.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.