How to Verify DC Power Supply Load Sharing
A pair of 24 VDC supplies wired through a redundancy module looks fault-tolerant on the drawing, but redundancy that has never been measured or exercised is a hope, not a design. Whether the pair actually shares load - and whether the survivor can carry everything when its partner dies - depends on voltage setpoints and wiring details that drift out of truth over time. This guide shows how to measure the sharing, correct it, and prove the failover for real.
DC Load-Sharing Check in one line: To verify DC power supply load sharing, measure each supply's individual output current under normal load, match the output voltage setpoints per the manufacturer's procedure so neither unit idles, then prove redundancy by removing input power from one supply at a time and confirming the bus holds and the failure alarm asserts. A supply that has been idling unproven for years is not redundancy.
What You Need
You need a DC clamp meter or access to the supplies' own current displays or monitoring outputs, the manufacturer's manual for the supplies and the redundancy module, and a maintenance window in which a deliberate single-supply failure is acceptable. The manual matters because sharing behavior is a design property: some pairs share actively through a control connection, others share passively through voltage droop, and plain diode-ORed pairs may legitimately let one unit carry everything.
Also confirm how the pair signals failure - DC-OK relay contacts, a module alarm output, or both - and where those signals land. Knowing what the concept of load sharing and redundancy promises for your specific hardware is the baseline the measurements get compared against.
Measure Each Supply's Share of the Load
With the panel running normally, measure the output current of each supply individually - clamp each supply's positive output lead, or read each unit's display if fitted. In an actively sharing pair, the two currents should be roughly equal. In a droop-sharing or diode-ORed pair, expect the supply with the slightly higher voltage setpoint to carry most or all of the load; whether that is acceptable depends on what the design intends.
Record both numbers and compare their sum against the total load and against each supply's rating. The critical question is not elegance of sharing but survivability: can one unit alone carry the entire load with margin? If the total load has crept above what a single supply can deliver - a common result of years of added circuits - the redundancy is already gone, no matter how healthy both units look today. That arithmetic belongs in the panel's load documentation, not in someone's memory.
Match the Output Voltages per the Manufacturer's Procedure
Where sharing is worse than the design intends, the usual cause is mismatched output voltage setpoints: a small difference decides which supply sources current through the ORing stage, and the lower-set unit sits idle. Adjust the setpoints following the manufacturer's procedure - typically measuring at defined points with the loads in a defined state - rather than twiddling both trimmers until the clamps agree. The manual's procedure exists because ORing diodes and mosfet modules drop voltage differently under load.
An idle standby is not merely inelegant. It means the panel runs on one supply full-time while the other ages unexercised and unproven, and its first real test is the day its partner fails. Balanced sharing spreads thermal stress and, more importantly, makes both units continuously demonstrate they work.
Prove Redundancy by Failing One Supply on Purpose
In the maintenance window, remove input power from one supply - at its input protection, so the test exercises the real failure path - and watch the bus: the surviving unit must pick up the full load without the PLC, RTU, or comms so much as blinking. Confirm the failed unit's DC-OK or module alarm asserts, then restore it, let it settle, and repeat for the other supply. Both directions matter; asymmetric wiring faults hide in the direction never tested.
Verify the alarm actually reached a person, not just a terminal strip. The whole economics of a redundant supply in a SCADA cabinet rests on the first failure being visible: redundancy without a working failure alarm just converts a visible outage into an invisible countdown to one. Wiring the DC-OK contacts into the RTU and configuring a notification closes that loop.
Verifying the Result
After adjustment and testing, the steady state should look like this: both supplies warm and delivering current consistent with the design's sharing scheme, total load within a single unit's rating with margin, failure alarms proven in both directions, and the measured values recorded with the panel documentation for comparison at the next check.
From then on, the sharing measurement becomes a drift indicator: a pair that measured balanced last year and reads lopsided today has a shifting setpoint, a degrading ORing path, or a load change worth investigating - each cheap to find now and expensive to discover during a real failure.
Common Mistakes
The classic mistake is trusting green LEDs. Both units lit proves both have output voltage; it says nothing about whether either could carry the load alone, or whether anyone would find out when one fails. The second is testing failover by pulling the DC output connector rather than removing input power - that exercises a different failure path than the one a real supply failure follows, and it can disturb the shared bus in ways a genuine failure would not.
The third is load creep: every added circuit quietly erodes the margin that made one-supply operation survivable, and nobody re-runs the arithmetic. Tie the sharing check to the panel's load records so the redundancy claim is re-verified whenever the load changes.
Frequently Asked Questions
Why is one power supply carrying all the load?
Usually because the pair shares passively and the output voltage setpoints differ slightly: the higher-set unit sources everything through the ORing diode or module, and the lower-set unit idles. In plain diode-ORed designs this can be normal and intended. If the design calls for balanced sharing, match the setpoints per the manufacturer's adjustment procedure and re-measure each supply's output current to confirm the shares equalized.
How do I test redundant supplies without risking the panel?
Do it in a planned window, and confirm beforehand that the surviving supply's rating exceeds the measured total load with margin. Then remove input power from one unit at its protective device, observe that the bus and loads ride through, verify the failure alarm asserts and reaches a person, restore, and repeat in the other direction. If the load measurement shows one unit cannot carry everything, fix that first - the test would just cause the outage it predicts.
Do redundant supplies need to share exactly equally?
No. What matters is that the arrangement matches the manufacturer's design: active-share pairs should measure roughly equal, droop-share pairs approximately equal, and simple diode-ORed pairs may intentionally run primary-and-standby. The non-negotiables are that either unit alone can carry the full load with margin, and that the failure of either unit raises an alarm someone receives. Balanced sharing is desirable because it exercises and thermally relieves both units.
Automation services
Need help turning this into a working system?
Merobix integrates SCADA, programs Allen-Bradley and Siemens PLCs, and designs and fabricates industrial control panels.
Meeting requests are reviewed before confirmation.