This thread documented a full recap project on an Asus A8N-E motherboard running an AMD Athlon 64 X2 4800+ processor. The board had developed symptoms consistent with capacitor failure during the 2007–2009 period.
The Platform and Symptoms
The Athlon 64 X2 era saw widespread motherboard cap failures due to the capacitor plague. Many boards from 2003–2006 shipped with budget capacitors from brands known to have used a defective electrolyte formula.
Symptoms on this particular board included random reboots under CPU load, memory training failures at startup, and occasional display corruption when running 3D applications. The CPU and memory were verified working on a different board before the capacitors were investigated.
Diagnosis Process
ESR measurements were taken on all 16 capacitors in the CPU VRM section. Eleven of the sixteen showed ESR values above 200 mΩ, compared to a datasheet maximum of 25 mΩ for that series. Three caps measured open circuit, indicating complete electrolyte dry-out.
The remaining caps in the secondary voltage rail and DRAM power section showed borderline ESR readings of 60–90 mΩ. The decision was made to replace all caps in the project rather than only the confirmed failures.
Replacement Selection
The original 1000μF 6.3V Tayeh caps were replaced with Panasonic FM 1000μF 6.3V as like-for-like electrolytic replacements. The original 3300μF 6.3V bulk caps near the ATX12V connector were replaced with Chemi-Con KZE 3300μF 10V for additional voltage headroom.
The DRAM power section received polymer caps where the physical size allowed. Four Nichicon PW 270μF 2.5V polymer caps replaced the original 330μF 2.5V electrolytics in the memory VRM output positions.
Results
After the recap, the board completed POST on the first attempt. Memory training succeeded without errors. Prime95 ran for 12 hours without a single restart or error. Oscilloscope measurement of the CPU VRM output showed 18 mV p-p ripple, compared to an estimated 120 mV before the work.
The board continued in service for four additional years according to the follow-up posts later in the thread. Total cost of the recap project including caps and flux was less than the cost of a used replacement board, making it worthwhile even accounting for the labor time.
Lessons Documented in This Thread
This thread became a reference for A64 platform recapping because of the detailed before-and-after measurements. The main lesson was to replace all caps in a section rather than only the obviously failed ones. Borderline caps often fail within weeks of a partial recap, requiring a second desoldering session.
A second lesson was the value of polymer upgrades in the DRAM power section. Members who performed only like-for-like electrolytic replacements saw continued DRAM instability in some cases, while polymer upgrades eliminated it.