Two Rack Feeds, One Equipment Inlet: What Does an ATS Change?

Locate the source selector before counting redundancy

A rack automatic transfer switch, or ATS, sits upstream of the connected equipment. It selects a supply from alternative inputs for the load. For a device with one inlet and one internal power-supply unit, both source choices still lead through that same inlet and PSU. The selection takes place before the device's existing power path.

Schneider Electric's December 2022 Rack ATS with Network Management Card 3 guide, document 990-91718A-001 for AP44XXA, describes two input power cords serving single-corded equipment. That is the named manufacturer's example here. It establishes the source-selection function; it does not establish the behavior of an arbitrary attached device.

The architectural consequence follows from where the selector sits. Adding it can introduce an alternative upstream source without replacing the device. It leaves the device's internal PSU population unchanged and adds the ATS itself to the equipment's supply path.

References: [S1]

Identify what a second internal PSU changes

In a device designed for separately supplied PSUs, the two external paths reach separate power-supply modules inside the equipment. The paths therefore remain separate farther into the device than in the single-inlet arrangement. They ultimately serve shared equipment, so the internal design determines where they converge and what remains common.

Two PSU positions are not evidence that two modules were delivered. Cisco's Catalyst 9300 hardware guide provides a useful, separate product example: its Product Overview, updated May 15, 2026, describes one or two internal power-supply modules and says the second module can be purchased separately. Supported population and supplied population are different purchasing facts.

A second module also needs an applicable redundancy function. Two installed PSUs could be required together for the offered load rather than providing the intended alternative path. Establish the exact configuration's supported behavior before describing it as PSU redundancy. The fictional comparison below expressly supplies its own architecture assumptions; it is neither an APC-to-Cisco compatibility example nor a specification for a particular switch.

References: [S2]

Compare two offers with the same two source labels

Consider a fictional buyer choosing a network device for the same stipulated network function. Both offers name source interfaces A and B. Their upstream independence is deliberately unspecified.

Offer A retains device N1 and adds one two-input ATS. N1 has one equipment inlet, one internal PSU and one main board. In this conceptual arrangement, either ATS input can be selected to supply the same downstream equipment connection. The count is two source inputs at the ATS, one inlet at N1 and one PSU inside N1. The ATS, the downstream cord and inlet, the PSU and the main board remain common to both source choices.

Offer B replaces N1 with device N2, which has two equipment inlets and two installed PSUs. The example stipulates a separate source-to-inlet-to-PSU path for each module and assumes either module can support N2's stipulated load on its own. N2 has one chassis and one shared main board. Its count is two source interfaces, two equipment inlets and two internal PSUs, with no external ATS in this sketch. These are invented architecture assumptions, not product ratings or observed operating results.

An upstream source-path issue is the problem Offer A is intended to address. A problem in N1's sole internal PSU remains downstream of the selector whichever input is chosen. Offer B changes that PSU population from one to two, while retaining shared equipment after the paths converge. Both offers may address a source-path requirement; only the stipulated second offer also provides an alternative internal PSU path.

That is enough to distinguish the purchases, but not to rank their availability. The counts assign no failure probabilities, transfer result or single-fault guarantee. Offer B's shared main board is still one board; Offer A's two inputs still serve one PSU. Calling both offers redundant without naming those boundaries hides the difference the buyer is paying for.

Keep source selection separate from energy and capacity

An ATS selects a source; its transfer function is not an energy store. Eaton's UPS component explanation identifies the battery as the element that stores energy for backup supply. A UPS may be part of an ATS input's upstream arrangement, but that stored-energy function belongs to the UPS, not to the presence of a selector. If neither input has an available source, selection alone cannot sustain the load.

Two alternative input ratings also do not become one additive load allowance. The ATS comparison concerns a selected supply path, not permission to combine the feeds into a larger pool. Actual supported load and interface limits require the applicable product and facility basis.

Likewise, A and B are identifiers. They do not reveal whether the feeds share upstream equipment or another dependency. In the fictional example, neither offer establishes independent sources because that information was not stipulated. A dual-PSU device can preserve two paths inside its chassis while still depending on shared infrastructure outside it. The existing PDU and facility-redundancy guides address those wider boundaries.

References: [S3]

Read transfer and service claims at their actual boundaries

Source selection answers where the load can obtain supply. Transfer behavior describes what happens during a change between sources. Load ride-through means the particular connected equipment continues operating through the resulting supply disturbance. Knowing that an ATS has two inputs does not establish that outcome for every device.

The exact ATS documentation and the equipment's applicable power-input evidence must support the proposed combination and conditions. A transfer-time headline alone cannot establish the load's response. The same discipline applies to a dual-PSU offer: the supplied modules, supported configuration and remaining-path capability must match the intended requirement.

Service access presents a separate buying question. Two ATS inputs do not establish a way to service the ATS, its downstream cord or the device's sole PSU while retaining supply to that device. Two internal PSUs do not, by their count alone, establish permission or capability for online replacement. The relevant product service boundary and operating conditions must support the actual maintenance objective.

Compare that objective explicitly. If the requirement concerns an unavailable upstream source, ask whether the proposed source-selection or separate-PSU arrangement addresses it. If it concerns service of a particular PSU or the ATS itself, identify whether that component remains common. Evidence for one boundary should not be presented as coverage of another.

Choose the purchase around the component you need to change

Retaining a single-inlet device and adding source selection can be a materially different purchase from replacing it with a device that has two applicable internal PSUs. The first preserves the existing device and its PSU dependency. The second changes the equipment architecture, with any associated mounting, network-interface and service implications. Neither is a universal recommendation: choose against the actual source-path, PSU and maintenance requirement.

Actual electrical design, connection, switching and acceptance belong to the appointed qualified facility and equipment parties. This architectural comparison provides no implementation procedure or work release.

Leviathan's rack-and-stack service can be discussed for agreed equipment mounting, IT cabling, labeling and physical handoff documentation around the customer-selected arrangement. Use the related service and project inquiry to describe the selected devices, released rack positions and requested physical work. That discussion does not establish equipment supply or electrical commissioning. The useful starting point is a clear choice of architecture and its remaining shared components.