Voltage dip incident 4090
Voltage dip incident 4090
Initially, my Aero OC 4090 had a damaged connector that Gigabyte replaced in August 2023. The voltage data from HWInfo showed stable readings—around 12.230V at idle and 12.120V–12.160V under load—consistent from 2023 through 2025. But a few months ago, the voltage began to fall below 12V, settling near 11.8V–11.9V. While ±5% is generally acceptable, even at 0.900V at 2325MHz the values dropped below 12V. My usual readings had stayed between 12.120V and 12.160V for years, so this change is quite alarming. At first, I thought the Seasonic Prime GX-1000 was malfunctioning, so I upgraded to a brand new NZXT C1000 (ATX 3.1). Unfortunately, it performed worse than my older model, with slightly lower voltages. After ruling out the power supply as the problem, I sent it back for repair. Gigabyte confirmed the card was operational with NFF (No Fault Found) and verified it using Nvidia’s diagnostic tool. The challenge remains: why did the voltage suddenly drop a couple of months ago? Although the readings technically met specs, my typical values were always well above 12V—undervolted or not. Anyone know what might be causing this? I’ve also checked the motherboard, especially the VRM section, and found no signs of damage. I changed the PCIe generation from Gen 4 to Gen 3, but the voltage readings stayed unchanged. Regarding the screenshots, I stopped testing before capturing them. This explains why the "Current" shows normal on some images but the "Min" value reflects earlier issues. My specs include an AORUS Z690 Pro with an i7-12700K, RTX 4090, and Seasonic Prime GX-1000. All cables are original PSU-provided. For the 16-pin connector, I’ve used both 12VHPWR to 2×8 and 12V-2×6 to 2×8 setups without difference. The diagnostic images are available, but I didn’t capture them fully—possibly due to timing or system state.
Could the "drop" be caused by an update from HWinfo, Bios, NVidia...? One of the headers on the PSU side has become a bit looser or corroded, and resistance has increased slightly. Shouldn't we be fine today? It would be a good day. (If you're concerned about the connector melting again, you can check its temperature with an infrared gun or a thermal sensor—both are affordable.) Edited August 8, 2025 by leclod
Sadly, the fix hasn't been applied yet. Since the connector broke, I've become very careful with my GPU's performance. The voltage measurements don't look low enough to trigger the melt, yet they've unexpectedly dropped to levels I haven't seen in years. (The earlier melt was caused by the first-gen CableMod angled adapter, which I no longer use.) I've reinserted everything and double-checked all cables to ensure they're securely connected, but the problem continues. I've attempted everything you suggested—DDU, reinstalling Windows, updating BIOS (both motherboard and GPU), rolling back BIOS, and even disassembling the case to rebuild it. That's why I'm feeling really confused. One thing that stands out is the GPU FBVDD reading would fall to 11.900v even when idle, but only for a short time, which hasn't happened before.