Knowledge Center

CRPS2400D2: Why Input Voltage Changes Usable Output

  • 11 Nov 2026
  • Powernexu Team

CRPS2400D2 is Great Wall’s 2400 W class, Platinum, 12 V CRPS module. For a replacement or new server design, its decisive specification is the available main-output current at the actual input: the manufacturer lists 91 A, 181 A, and 197 A, while a model-specific sales datasheet maps those figures to different AC voltage bands and 240 Vdc. A 120 V rack feed therefore cannot be treated as a 2400 W feed simply because the module accepts low-line AC. This article explains how to use that input-dependent rating before committing a server load, redundant pair, or spare-parts order.

The 2400 W label has an input-voltage boundary

Great Wall’s public CRPS table gives CRPS2400D2 an input range of 90–264 Vac or 180–320 Vdc and reports +12 V main-output current at 91 A, 181 A, and 197 A. The range says the converter accepts those input families. It does not assign full output to every voltage inside them. The CRPS2400D2 specification and selection guide published by an authorized Great Wall sales channel provides the missing AC mapping: 91 A at 100–127 Vac, 181 A at 180–200 Vac, and 197 A at 200–264 Vac. It also lists 197 A at 240 Vdc. Great Wall’s page alone does not establish how to interpolate between those bands or what output is allowed at every point of its broader input range.

The guide labels the output +12.2 V, while Great Wall’s summary rounds it to +12 V. Using the guide’s nominal value, 91 A is about 1.11 kW, 181 A about 2.21 kW, and 197 A about 2.40 kW. Those calculations explain why a 2400 W product can have a substantially smaller low-line current limit. They do not replace the document’s combined-output rules, voltage tolerance, or revision-specific derating conditions. A rack planner should budget by the specified current at the feed actually delivered to the module, then check the allowed standby load and any applicable thermal limit.

One apparent discrepancy deserves attention. The guide’s input table names 200–240 Vac as an input row but its output-current text extends the top band to 264 Vac; the main Great Wall listing names 90–264 Vac. Treat that as a document-format limitation rather than filling gaps with a guessed curve. For an installation outside the explicitly paired conditions, obtain the latest approved technical specification for the ordered revision. The documented 240 Vdc point likewise cannot be generalized to every voltage in the public 180–320 Vdc acceptance interval.

Technician checks the protected AC feed to a rack power module while documenting the installed input condition.

Translate the site feed into a usable server budget

Nominal building voltage is not the same as voltage at the PSU inlet during a loaded event. A 208 V circuit may experience drop across branch wiring, PDU contacts, cords, and inlet hardware. If the actual inlet can fall below 200 Vac, the 197 A tier cannot simply be used for the whole operating envelope. The relevant question is the minimum voltage during the period when the server can draw its peak sustained current. Record that number at the module terminals or conservatively account for the upstream drop. Then pair it with the manufacturer’s current tier and permitted duration.

Consider a hypothetical server with a steady 12 V distribution demand of 170 A. At 12.2 V, that is approximately 2.07 kW at the PSU output, before considering its separate standby allocation. The example lies below the published 181 A and 197 A points but above 91 A. Thus it may be supportable on an adequately maintained high-line feed and plainly exceeds the listed low-line main-current point. The conclusion is about this example’s current comparison, not proof that the server’s connectors, control pins, transient behavior, or cooling are compatible with CRPS2400D2.

A 190 A example exposes a narrower boundary. It fits beneath 197 A but exceeds 181 A. If the deployment expects a 180–200 Vac inlet condition, the guide does not support a 190 A sustained main load on one module. At a stable inlet in the top band, only 7 A separates the example from the published 197 A figure. That small arithmetic margin may disappear when the actual load includes converter losses, fan speed changes, storage spin-up, or a rack-level power policy that permits higher demand. None of those loads should be guessed; measure or obtain the host’s bus-current budget at its defined operating states.

Surviving a module loss changes which current matters

A dual-module server can operate normally while each CRPS2400D2 carries part of the total bus current. That condition says little about the moment one module is removed or loses its source. In a 1+1 design, the surviving module must supply the protected load by itself, at its own inlet voltage and within the host’s allowed transfer behavior. If a 170 A server normally shares roughly 85 A per module, a single unit sees approximately 170 A after the other is unavailable. The example is below the guide’s 181 A high-line middle tier but exceeds the 91 A low-line tier. A healthy redundant pair fed from two 120 V circuits would not make that 170 A load one-module tolerant under the published low-line current limit.

The source topology can create a subtler case. Suppose one module receives a stable 230 V feed and the other receives a lower-voltage backup path. During ordinary sharing, both may carry modest current. After loss of the high-line feed, the lower-voltage survivor inherits the full bus load. Its relevant rating is the tier available on the backup path, not the rating of the unit that failed. The A/B distribution diagram should therefore identify the voltage envelope at each PSU independently and which source remains after each credible event.

Load transfer also has a time dimension. A processor or accelerator power cap may eventually lower the server’s draw, but the electrical path must carry the load until that control action takes effect. The public Great Wall table does not state a transfer waveform or an approved excursion duration. The sales guide mentions active current sharing and redundancy as product features; it does not certify any particular server’s 1+1 continuity. Chassis isolation, ORing, connector sequencing, firmware behavior, and load response belong to the host design and approved integration record.

Illustrative complete redundant server and lab instrumentation checking the surviving power module during a low-line AC event

Keep the D2 identity separate from adjacent CRPS models

The model suffix is not a cosmetic label. Great Wall places CRPS2400D2 near CRPS2200D2 and CRPS2700D2 in its CRPS range. The 2200 D2 table lists 91 A, 164 A, and 181 A; the 2400 D2 lists 91 A, 181 A, and 197 A; the 2700 D2 lists 98.4 A, 205 A, and 221.4 A. Their headline wattage and upper current entries differ even though the public table gives all three the same 185 × 73.5 × 40 mm body dimensions. CRPS2700D2 is also listed with FR or RF airflow, whereas CRPS2400D2 is listed as FR. A replacement record that says only “D2, 185 mm” loses distinctions that affect both power capacity and airflow choice.

CRPS2000N3 is another visually plausible neighbor with the same published external dimensions and 90–264 Vac or 180–320 Vdc input families. Its public main-current entries are 65.6 A, 148 A, and 164 A, below those of CRPS2400D2. Moving from N3 to D2 is therefore not a mere 400 W nameplate change. A platform built around a lower-current part may have PDB contacts, conductors, cooling, and system firmware sized for its original approved assembly. Conversely, installing a smaller N3 as an emergency spare in a host that truly requires the D2’s higher current could put the host beyond the substitute’s documented capability.

The related CRPS2700T2 article addresses a Titanium 2700 W model and an independent efficiency test at 230 V. That evidence should not be carried over to the Platinum CRPS2400D2. Similar CRPS dimensions do not transfer efficiency measurements, temperature derating, certification, or host qualification. A spare-parts list should include exact model and revision, the host manufacturer’s approved part number when one exists, and the inlet conditions under which the system was specified. That record prevents a technician from choosing a module by faceplate similarity or a broad “12 V CRPS” description.

The physical interface is a separate source of evidence

The published 185 × 73.5 × 40 mm dimensions establish the outer body envelope. They do not guarantee that the PSU seats correctly in a specific bay or mates with its PDB. The front handle, retention point, connector depth, contact plating, guide rails, fan direction, extraction clearance, and service-side inlet geometry all affect interchangeability. Even a unit that slides fully into the opening can fail host identification or standby sequencing. A chassis maker’s approved spare list is the strongest practical starting point for an existing server; a new PDB design needs the revision-specific mechanical and electrical drawings.

The authorized-channel guide contains a connector-definition table for a CRPS2400/2700/3200 family. It names power ground, +12 V, standby, presence, control, sharing, and management contacts. That table is useful evidence that the interface is more than two power terminals. It is not permission to assume every similarly shaped unit or host implements the family interface identically. Before a board layout or field substitution, check the exact connector revision, pin assignment, mating contact rating, and host logic thresholds against the latest manufacturer-approved drawing. Do not copy a PMBus register map or firmware expectation from a different product simply because both claim CRPS compliance.

At 197 A, contact and PDB layout deserve particular care. Total main-bus current is distributed across multiple contact pairs and board paths; the aggregate rating does not state the allowable current through one contact, trace, or cable. Small contact resistance can make a localized hot spot that is not visible in an average PSU outlet temperature. The system designer should confirm the contact-current allocation, copper and busbar cross sections, protection path, remote-load voltage, and thermal readings at the highest intended current. These are host design questions, not hidden promises of the CRPS2400D2 product label.

Great Wall specifies FR airflow and an operating-temperature range of 0–55 °C. The inlet air to a dense PSU bay may be warmer than room air because of recirculation, and the server’s flow resistance can differ from a bench fixture. The public summary does not publish a full-output derating curve versus inlet temperature or a required pressure-flow curve. A deployment expecting high current near the temperature boundary needs those manufacturer conditions and a host airflow assessment. The outer dimensions alone provide no thermal approval.

What the CRPS2400D2 evidence supports

For a 2400 W class 12 V server module, CRPS2400D2 is a plausible candidate when the host accepts the exact Great Wall revision and its feed remains in the band that supports the needed main current. The strongest public distinction is the input-dependent current ladder: 91 A at 100–127 Vac, 181 A at 180–200 Vac, 197 A at 200–264 Vac, and 197 A at the guide’s 240 Vdc point. Those are not additive rails. They are alternative capacity conditions, so a system’s protected one-module load must be compared with the tier available on the feed that survives its defined failure scenario.

Where the planned inlet lies in an unpaired part of the public input range, or the server requires a precise surge duration or temperature derating, request the current revision’s detailed specification. Confirm the host-approved interface, its standby and management behavior, and the PDB path before treating physical fit as compatibility. That evidence may establish CRPS2400D2 as the correct spare or design choice; without it, the 2400 W name alone cannot resolve a 120 V rack, a mixed A/B feed, or a high-current server bay.

Share:

Leave a Reply

Your email address will not be published. Required fields are marked *