My desktop has developed severe memory instability after a history of moisture, oxidation, transport, cleaning, and a minor physical accident involving the RAM slot. The system previously had cold-boot problems that improved after repairs to the power supply and work on the motherboard and graphics card. More recently, the PC was transported in a bus luggage compartment, and shortly afterward a cat urinated above the case. I cleaned the case and components, but the machine later began crashing within seconds or minutes, even in the BIOS or bootloader. It also produced a kernel error.
The system is currently stable with one 16 GB DDR4 module installed in slot B1. Any of several known-good RAM modules work reliably in B1, but placing a module in slot A1 eventually causes crashes, failed boots, or the system to stop recognizing the memory. Sometimes it works for many hours before becoming unstable, and after cooling down it may work again temporarily. I have tested multiple matching modules, cleaned the sticks and slots, reset the CMOS, changed XMP settings, and tried lower memory speeds. The problem consistently follows slot A1 rather than a particular RAM stick.
I also accidentally tried inserting a RAM module in the wrong orientation and broke part of the slot's plastic retention guide, although the instability seemed to begin before that happened. The motherboard is a Gigabyte A520M K V2 with a Ryzen 5 5500, 2x16 GB DDR4-2133 memory, an RX 6600, and an XPG Core Reactor 650W power supply. Could the A1 slot or motherboard be damaged, or could this instead be caused by the CPU's memory controller, socket contact, cooler pressure, moisture, or board flex? Is replacing the motherboard the most sensible next step, and are there safe tests I should perform first?
2 Answers
A memory slot that works for hours and then starts failing is still a hardware fault; the delayed failure does not make it a software or operating-system problem. Because the crashes also happen in the BIOS and bootloader, the OS is not the cause. A damaged slot, cracked solder joint, contaminated contact, or a failing memory channel can behave intermittently as temperature and board flex change.
The broken plastic guide is not necessarily the electrical cause, but if the module was forced in backward it could have damaged contacts or the slot’s solder connections. Do not keep using a module that has damaged contacts. Test with one confirmed-good stick at JEDEC defaults, disable XMP, and avoid mixing unmatched kits. If the fault remains exclusive to A1 after inspection and an out-of-case test, the motherboard is the most practical part to replace; the CPU is possible but less common.
Since several known-good modules work in B1 but fail or become unstable whenever A1 is populated, the slot or the memory channel is the main suspect—not the RAM itself. The CPU’s integrated memory controller and its socket contacts are also possible, because one memory channel can be lost through a bent or contaminated contact. Moisture, oxidation, transport stress, or board flex could all contribute.
Before buying parts, remove the motherboard from the case and test it on a nonconductive surface with only the CPU, cooler, one module, and graphics output. Inspect A1 closely with bright light and check for corrosion, bent contacts in the CPU socket, debris, or cooler pressure that could be flexing the board. Make sure the cooler is evenly tightened rather than excessively tight. If A1 still fails with every module outside the case, replacing the motherboard is reasonable. Keep the system powered off until any urine residue or corrosion has been completely cleaned and dried, since liquid damage can worsen over time.
I tested multiple 16 GB and 8 GB modules individually, and the behavior always follows A1. B1 remains stable with the same sticks. I’m going to inspect the CPU socket and cooler mounting, then test the board outside the case before replacing anything.
That matches what I’m seeing: A1 sometimes survives for many hours, then becomes progressively less reliable, while waiting and letting the system cool can make it work again temporarily. I’ll stop treating it as a memory-speed problem and focus on the board, socket, and physical damage.