By manufacturer
Dell PowerEdge HPE ProLiant Lenovo ThinkSystem Fujitsu Primergy Supermicro IBM System x / Power Acer AltosServer PSUs: hot-swap replacement with no system downtime.
On servers with redundant PSUs (2N or 1+1) the faulty module is replaced with the system running. Diagnosis via the BMC SEL and status LEDs, verification of adequate capacity in high-draw scenarios (GPU compute, dense NVMe), replacement with an official part or a certified equivalent.
Amber LED, SEL, then electrical confirmation.
An enterprise PSU reports its status via the front LED and via the BMC SEL. Sequence:
- Solid amber LED on the module = PSU in fault, output not guaranteed.
- Blinking amber LED = warning (e.g. loss of redundancy, AC lost but DC still supplied by the twin module).
- SEL events: Power Supply Failure, Power Supply Configuration Mismatch, AC lost.
- Electrical verification, optional: AC input reading, output sampling with a load tester, stand-by current check.
On systems without redundancy (single PSU) a PSU fault means immediate downtime — emergency call-out.
Dell, HPE Common Slot, Lenovo CRPS, Cisco, Supermicro.
- Dell PowerEdge: hot-swap redundant PSUs 350W / 495W / 750W / 1100W / 1400W / 2000W / 2400W / 2800W. Platinum, Titanium.
- HPE ProLiant: HPE Common Slot 500W / 800W / 1000W / 1600W / 1800W. Common Slot is cross-compatible across many generations.
- Lenovo ThinkSystem: CRPS (Common Redundant Power Supply) 550W / 750W / 1100W / 1600W / 2000W / 2400W.
- Cisco UCS: PSUs 1050W / 1600W / 2300W depending on B / C series.
- Supermicro: PSUs from 600W up to 3000W Titanium, with variants for 1U / 2U / 4U chassis and GPU servers.
Not just "the PSU is dead" — sometimes "the PSU is no longer enough".
Adding GPU compute, dense NVMe drives or high-TDP CPUs to a server designed for a less demanding configuration can push the system beyond the margin of its rated PSU. Symptoms: sudden shutdowns under load, spontaneous resets, "phantom" loss of redundancy because both PSUs are running at 95% instead of 50%. The correct diagnosis leads to replacing the PSUs with a higher-rated version — where the chassis supports it.
The questions we get asked most.
Can I replace a hot-swap PSU myself?
Yes, on systems with active redundancy (PSU 2 operational): the operation itself is physically simple. What we add is the correct diagnosis (is it really the PSU? which rating is needed?), sourcing the compatible spare and verifying system status via the BMC after replacement. For many customers the benefit is not the physical swap but avoiding the mistake of buying incompatible PSUs.
Are Dell, HPE and Lenovo PSUs interchangeable across brands?
No, never. Each brand has its own proprietary backplane connectivity. Within the same brand they often are (HPE Common Slot in particular is cross-generational).
Are refurbished PSUs an option, or only new original parts?
On this component certified refurbished units have an excellent track record: they come from controlled decommissioning, are factory tested and typically carry a 12-month warranty. The cost difference versus new is significant, and on systems no longer under active vendor warranty it is the usual choice. On mission critical systems we recommend new original parts.
How long does a PSU replacement typically take?
Hot-swap on a running system: 5-15 minutes with remote hands. On an on-site call-out most of the time goes on travel and paperwork, not on the physical work itself.