[SOLVED] Red CPU LED, GPU/PSU fans don't turn off after shutdown. PC works fine.

[bakigabi55]

Hi

Few days ago my PC shut down instantly in game menu (MW), but it was able to turn on normally, no bios resets, no burnt smell etc. I did a stress test with Furmark (cpu+gpu) for 1 hour, temps were fine (max 71°C on CPU avg 68, GPU 67°C) and nothing happened.

Yesterday, when tried to wake from sleep it stuck in a self check loop. It turns out it couldn't start at 3200Mhz XMP (RAM) only on it's stock 2666Mhz. After restart, appcenter update the firmware on its own (yes it was only said firmware) and "update" the bios to the stock F4 from F11 on both main and backup bios. I didn't want to diagnose at night so I turned it off.

Today before turn on, I heard a quiet moaring from the PC, the GPU and PSU fans were spinning at low speed, but the PC was off. CPU led was red after I stopped the blades by hand it started blinking and tried to spin the fans up again even if the power cables pulled off, so it get power from the PCI slot. Error code D4 (or 04, anyway it is the "PCI resource allocation error. Out of Resources. ") but it wasn't so bright as it used to. (Btw i realised the PSU fans were spinning too in eco mode, after start it turns off under low load)
After start, everything works fine, no CPU led after check, runs with XMP profile on 3200MHz. Did some driver reinstalls, latest bios update. Nothing wrong when it's running.

After shutdown all fan stop but the GPU fans keep spinning and the CPU led turns on, POST code: D5 ("No Space for Legacy Option ROM initialization" ) or! D0 ("CPU initialization error "). It changes after every shutdown.
If I turn the switch on the PSU off, then on, GPU tries to spin up, CPU led is on and a very dim 00 (probably D0) POST code appears. No buttons have pressed except he PSU switch.

What I have tried:

• - Check if both bios works fine ✔
• - Check all power cables in the system ✔
• - Pull out every power cables from the system, except he 24pin
• --> CPU led on, PSU and GPU fans are spinning or at least the gpu tries to spin.
• - Get out the GPU and start the sys with cables connected (before start, CPU led)
• --> GPU led (obviously), no CPU led
• - Reseat the CPU
• --> same thing. CPU led and D0
• - Start without CPU (I know it's not a good thing)
• --> CPU led (obviously), constant rapid on-off loop
• - Bios reset, new bios ✔
• - Check the PSU on it's own. Short the green-ground wires
• --> Click sound, like when it's turn on but no fan spinning
• - Pull out the case pwr switch from mobo
• --> No changes
• - Every cables plugged in, except 24pin
• --> No LED, no fan (GPU, PSU)
• --> Plugged in the 24pin --> CPU led, fan spin (GPU, PSU) and dim POST code

What I forget to / don't try:

• Reseat the memory sticks, but it every time pass except the first time from sleep.
• New PSU, currently I don't have in spare
• New CPU, same
• New Mobo, same

My toughts:

• PSU failure (voltages are fine even under stress test)
• Motherboard failure
• CPU failure
• I tought it might be corrupt bios at first time, but after update i don't think so.

Specs:

CPU: Ryzen 7 3700x
Cooler: Be Quiet Dark Rock Pro 3
Mobo: Gigabyte x570 Aorus Master
PSU: EVGA Supernova 850 P2
GPU: Gainward Phoenix GTX 1080
RAM: 2x8 Gigabyte 3200MHz
Case: Fractal Design Meshify S2 Dark TG

I know it is a bit messy how I describe it, but i hope someone can help. Pc works fine just it can't turn off completely, just only with the PSU switch.
No overclock, voltages were set to auto (yes it hits 1,48V but as you can see it tops at 71°C)

Update 05.28:
Next morning PSU switch on, GPU/PSU fans spin up, little bit faster for GPU. CPU led flashing, no visible POST code. Probably not enough power for it due to the GPU fan. Might be this is why the led is flashing?...

Could be a shortage in PSU or a bad Mobo? It is affected at the 24 pin.

After start every bedug light went out, PC works as is should be. Nothing happened at night during game

Utdate 05.29:
No CPU led after shutdown, just fan spinning. Probably out of juice. Next start same thing.

 

[bakigabi55]

I've read that someone had the same issue on an intel platform.
His solution was motherboard repleacement. New PSU did no change for him.

For me currently not aviable but first i want to try with my old PSU when i have chance

 

[bakigabi55]

Update:

Not PSU failure. Test with 3 other one, all of them turned on. Bigger the PSU faster fan speed.
Currently only the PSU fan turns on and the CPU debug led flashing

 

[Karadjgne]

Try replacing the cmos battery entirely. There's been reports/rumors/suggestions that Gigabyte got ahold of a crappy batch of batteries and that's the root cause of many issues with bios.

Ryzen can use upto @ 1.5v at idle or very low current levels. For middling loads it shouldn't be beyond @ 1.325v, for extreme loads 1.2v. Of course lower is better.

Balanced power plan.

Check windows event viewer for the initial shutdown, it'll be a red flagged critical error.

With that cooler you should have no issues running Ryzen master with PBO enabled which will enable the cpu to have better control of its current and voltage settings.

Also turn off/disable Fast Boot in bios. With an ssd/NVMe Fastboot will save only 1-2 seconds at best at boot, but can/does/will cause a bunch of errors at Windows Shutdown with regards to cmos loading.

Debug D0 is a cpu initialization error. Check thoroughly for bent pins.

 

[bakigabi55]

Try replacing the cmos battery entirely. There's been reports/rumors/suggestions that Gigabyte got ahold of a crappy batch of batteries and that's the root cause of many issues with bios.

Ryzen can use upto @ 1.5v at idle or very low current levels. For middling loads it shouldn't be beyond @ 1.325v, for extreme loads 1.2v. Of course lower is better.

Balanced power plan.

Check windows event viewer for the initial shutdown, it'll be a red flagged critical error.

With that cooler you should have no issues running Ryzen master with PBO enabled which will enable the cpu to have better control of its current and voltage settings.

Also turn off/disable Fast Boot in bios. With an ssd/NVMe Fastboot will save only 1-2 seconds at best at boot, but can/does/will cause a bunch of errors at Windows Shutdown with regards to cmos loading.

Debug D0 is a cpu initialization error. Check thoroughly for bent pins.

Thank You for your answer.

I repleaced the battery but it didn't help. But I noticed, under the backplate there is some resodue run down from the thermal pads on the VRMs. (Silicon maybe?)

I never used fast boot exactly because what you mentioned.

The power plan does not do much because this happening too when the PSU switched off then on. I've trien a buch of power saving options but non of them help.
24pin connected = electricity flow. --> the problem starts somewhere here

I've removed the ram sticks, the cpu and every component from the PC, but when I pluged the 24pin in, PSU starts to spin and a dim POST code flashing D0 (yes the cpu initialization error) and CPU debug light flashing too.
If a GPU is connected, it tries to spin the blades with exactly the same pulse as the post code flashing. Still no cpu connected.
CPU pins are straight AF, it was only get off the socket when I tried some methods.

RMA is around 2 months so if it is possible, i would try to solve it at home... Because I have no spare parts.

 

[Karadjgne]

I'm assuming you are using the 4+4 (+4)pin to the EPS plug too right? Just checking.

Honestly, you've gone through everything a consumer should ever have to go through. Have you unplugged all the front panel connectors, the usb headers, everything except the cpu/1 stick of ram in A2, cooler and EPS/mains? Use a flathead screwdriver to short the poweron +/- pins.

The only thing that really bothers me is that residue from under the heatsink. VRM's usually use a thermal pad because of mm differences in height, not normally paste or such. But there are caps under there too and if a cap has failed you absolutely have no option but rma. Of the breadboard fails to get a clean boot, same thing, rma. The timing sucks for sure, but you paid for the warranty and you may just have to use it.

 

[bakigabi55]

I'm assuming you are using the 4+4 (+4)pin to the EPS plug too right? Just checking.

Honestly, you've gone through everything a consumer should ever have to go through. Have you unplugged all the front panel connectors, the usb headers, everything except the cpu/1 stick of ram in A2, cooler and EPS/mains? Use a flathead screwdriver to short the poweron +/- pins.

The only thing that really bothers me is that residue from under the heatsink. VRM's usually use a thermal pad because of mm differences in height, not normally paste or such. But there are caps under there too and if a cap has failed you absolutely have no option but rma. Of the breadboard fails to get a clean boot, same thing, rma. The timing sucks for sure, but you paid for the warranty and you may just have to use it.

Update at the end

I've connected the 2x(4+4) plug to the CPU. I know a single 4+4 would be far more enough, but i tought why not, maybe it ballance the load on each rails.

The board has pw/reset switch on it, so I used those for power on while testing. Not a single cable was plugged in except the 24pin.
I've tried, with the minimals as you said, cpu & cooler, 1 ram stick. Later +gpu that suck most of the power from the LED so they were dim.

The residue is on the back of the board and the thermal pad is about 10mm thick that transfer the heat from the back of the VRMs to the backplate. The front side is dry, the caps are not swallowed.

https://pokde.net/review/gigabyte-x570-aorus-master-review/

In this article you can see it under the spec table.

"Taking a Closer Look
The VRMs"

There is a pic that shows the same residue but mine has a bit more run down to the first PCI slot.

Update:

I have a PSU from a PC in use. I gave it a try and of course it is working. No debug light, no fan spins, starts just fine. It is a Thermaltake Paris 650W gold PSU that operate my old FX based PC for 6 years now.
The EVGA works normally with it that is annoying.

The warranty is good till end of '22 so I think I'm going to use it with the Thermaltake PSU and later get a good kind of Seasonic that has plenty of warranty on it. Maybe there is some incompatibility between the EVGA and the motherboard or a shortage somewhere. If the problem still occurs with that, the board go to rma. The reason why I don't send it, because If they can't reproduce it with their PSU like my Thermaltake, they just send it back with the service bill.

 

[bakigabi55]

Short update:

About 2 months ago the mouse's reciever was burned out in the front USB port when I plugged in a pendrive next to it.

Today when I connected my phone to charge it (also front USB and the same port) the xbox controller that was connected via USB cable to the other port, start smoking. As soon as I saw that instantly disconnect it, but the controller's charging port was damaged. Data is still going through usb, but it can't get any power from it.

Note: On the phone's port, the fast charge was turned on through the motherboard official "UsbTurboCharge" software both time. It could be software issue but my opinion something absolutely go wrong in the mobo... This is another psu in it so it is not from a bad psu.

 

[bakigabi55]

Update:

Mobo has sent to RMA. Probably it was chipset problem, but i dont know for sure because the shop where a bought it paid back the price instead of give a repleacement. The reason was "too much extra cost for them"...

In conclusion, when i spoke with gigabyte esupport, he said it is probably chipset problem.