IBM ServeRAID 7K SCSI U320 Compatible Battery 3.7V 1000mAh

IBM ServeRAID 7K SCSI U320 RAID Controller — 3.7V Li-Polymer Replacement Battery (90P5245)

This 3.7V, 1000mAh Li-Polymer cell replaces the backup battery on the IBM ServeRAID 7K SCSI U320 RAID Controller. It powers the controller's cache memory during a system power loss, holding write-pending data until the server recovers. Also cross-references OEM part numbers 71P8642 and 39R8804.

• ServeRAID 7K cache backup circuit: The 7K controller uses a dedicated battery-backed write cache. When this cell degrades below the controller's minimum voltage threshold, the card drops to write-through mode — removing the performance benefit of write-back caching until a healthy cell is installed and recalibrated.
• Bench tested on actual hardware: We ran charge and discharge cycles on this cell and confirmed the BMS accepts a full charge without error. The cell voltage held within spec under sustained low-current draw, consistent with cache backup duty.
• Learn cycle after installation: After fitting this battery, trigger a learn cycle from the controller management interface. On systems using StorCLI, run StorCLI /cx bbu show to monitor status. The controller stays in write-through mode until the learn cycle completes and the backup window estimate is recalibrated against the new cell's actual capacity.

RAID controller staying in write-through mode after battery swap

Fitting a new cell does not automatically restore write-back mode. The ServeRAID 7K controller must complete a learn cycle to measure the new cell's actual backup window before it trusts the battery again. Until that cycle finishes — which can take 24 to 72 hours under normal server load — the controller keeps caching disabled as a precaution. Manually initiate the learn cycle via StorCLI or the controller's management interface; do not wait for it to trigger on its own schedule.

Management console showing a battery fault days after the swap

The ServeRAID 7K runs its battery self-assessment on a timed interval, not at the moment a new cell is detected. A fault flag showing up two or three days post-swap usually means the scheduled assessment ran before the learn cycle had fully recalibrated the new cell. Check battery status with StorCLI /cx bbu show — if the learn cycle is still in progress, the fault clears automatically once it completes. If the fault persists after a full cycle, verify the cell voltage is reading above 3.5V at rest.