I manage 72 Dell Precision 5820 workstations purchased together and now about four and a half years old. They recently began failing to POST in large numbers: 15 failed within a short period, and after testing and reseating components, 10 appeared to have motherboard-related faults. Several showed the 3-1 flashing power-button code, which points toward the motherboard, memory, or CPU.
Since then, more systems have failed, including several immediately after a building power shutdown for electrical testing. In total, 26 of the 72 machines now appear dead. The fans do not spin, BIOS recovery has not helped, replacement CMOS batteries made no difference, and moving the CPU, RAM, GPU, and power supply into a known-good motherboard brought one system back to life. The machines were purchased in the same order, and the failures are happening well beyond what I would expect from normal component wear.
Power quality has been checked with a logger, there were no obvious BIOS-update failures, and the systems were generally running the same software and firmware. Dell has repaired some systems for a fee but will not investigate the common cause because they are out of warranty. Could this be a defective motherboard batch, a power-related issue, or another known failure mode? I need a likely root cause before the remaining systems fail too.
5 Answers
The failure pattern is unusual enough that I would treat this as a fleet-level hardware or power investigation rather than a software problem. Since the machines were bought together and many fail after power removal, preserve several failed boards and document their service tags, manufacturing dates, failure dates, POST codes, and the exact room or power circuit. Ask Dell for an engineering or product-quality escalation, not just another repair ticket. A common board-component or power-event failure is possible, but the evidence needs to be correlated before calling it a manufacturing defect.
Because several units died after an electrical shutdown, have the electrician verify the complete power path: voltage, neutral integrity, earthing, surge protection, UPS or PDU behavior, and whether any circuits were switched or re-energized abnormally. A logger that shows normal voltage during ordinary operation may not capture a brief transient or switching event. Compare the failed machines with systems on different circuits. If the failures are concentrated in one room or circuit, power becomes much more likely than a random motherboard defect.
Even if the shutdown only exposed marginal boards rather than directly damaging them, the timing is important evidence. I would not dismiss the electrical work until the circuit and switching equipment have been checked.
A CMOS battery is easy to replace and can cause POST problems on some Dell systems, but if brand-new batteries and full power drains changed nothing, it is probably not the common cause here. Warranty extension might reduce repair costs, but it will not answer the root-cause question. Preserve failed hardware and escalate through Dell’s business account or product-quality channels while collecting evidence from the building electrician and the affected machines.
The fact that transplanting the CPU, RAM, GPU, and power supply into a known-good motherboard restored a system strongly points to the original motherboard, but it does not identify which part of the board failed. Do not assume visibly bulging capacitors are required; regulators, MOSFETs, standby power rails, or solder joints can fail without obvious physical damage. Keep at least one failed board untouched and have an electronics repair specialist test its standby voltages and power sequencing.
A thermal camera or freeze-test may help identify a component that overheats or becomes stable when cold, but those tests should be done carefully and are only diagnostic clues, not proof of the original cause.
BIOS recovery, firmware updates, Windows updates, and BitLocker are worth ruling out on a few systems, but they are unlikely to explain machines with no fan movement or other signs of standby power. A failed BIOS can prevent POST, yet the simultaneous failures and successful component transplant make a board-level electrical fault more plausible. Check whether the 3-1 code is consistent across the affected units and compare it with a known-good board using the same memory and CPU.
That is exactly what I am trying to get from Dell. Replacing boards individually does not explain why more than a third of the fleet is failing in such a short period.