Tips for chasing random memory errors that I wish I'd known sooner

Spent two weekends chasing one of these on my dad's old office machine. It would freeze maybe twice a month, always mid-afternoon, never while idling. He was convinced it was a virus. I pulled all the sticks and ran MemTest86 on each one overnight in a diferent slot. Three sticks came back clean, the fourth had errors clustered on the same addresses every time. New stick of the same size and speed, and the machine has run without a hiccup for about eight months now. The slot testing is the part most people skip. When I first tried it I had two sticks in and got errors, so I assumed both were bad. Turned out one stick was fine and the third slot was the problem. Once I moved everything around the picture got clear fast. On the cosmic ray thing, my brother-in-law works at a lab and says they see single bit flips in thier machines often enough that they log them. Sounds like sci-fi but it's just particles coming through the atmosphere. For a home PC though, I'd bet on heat or a dusty contact long before random particle hits. A can of compressed air and an hour of patience has fixed more of these for friends than anything else.

I wonder if anyone has considered the voltage rail stability beyond just the main power supply, particularly given that many modern systems integrate the memory controller directly into the CPU itself. A marginally underspecified power rail on the motherboard can certainly introduce transient electrical noise that mimics intermittent memory failure, even if all the sticks themselves are perfectly sound and clean. It is such a vexing issue beacuse the resulting error reporting is frequently non-existent until the entire architecture finally gives out under heavy load.

The random corruption is often linked to firmware issues before it is acutally the RAM itself. I remember one build where updating the BIOS completly cleared a series of phantom memory errors. It is funny because the error codes pointed directly at the chips, but the problem was just the memory controller trying to talk to the chips at an outdated clock speed. On one ocasion, a simple 1.05V adjustment to the integrated memory controller voltage fixed a crash loop entirely. This is far more common than most people realize.

Update: reseating the sticks and cleaning the slots fixed the weekly freezes, it's been two weeks quiet. But it still threw one error overnight, so I tried the BIOS update like that poster suggested and the phantom errors dropped but didn't stop. Did yours stay clean after the firmware fix, or did they creep back after a while?