Question No POST, no LAN, infinite power cycle - but lights work...?

[MaxMillrose]

Hello everybody.

My PC won't start. Specific components down below, but here is the general setup of what I have:
- A test bench / open air case
- A MoBo with a CPU (and CPU-Cooler) and RAM installed
- PSU connected to the MoBo in the following ways:
20 + 4 Pin connector for MoBo power
4 + 4 Pin connector for CPU power
- The CPU-Fan is connected to the MoBo via the designated connector.
- Front-Panel was also connected (PW- and RESET-Switch and LEDs for HDD and Power)

Nothing more. No PCI-Graphicscard or even some RGB-fan or such. Quite minimal in my eyes.

Actual Behaviour:
When I connect the PSU to the power-outlet, switch it on and then press the power-button, it starts.
I determine this by the fact, that the PSU audibly starts and I can see it's and the CPU-fan spin.
But after a while the fans just stop spinning (technically there is still inertia, but the motor turns off) and the power-LED turns off.
Shortly after (a second?) the fans spin up again and the power-LED is lit again.
That power-cycle repeats infinitely.

Expected Behaviour:
I expect the PC to boot at least to POST and even BIOS.
Since the SSD in not installed, no OS or even bootloader can be found, I understand that.
But all things considered, should there not at least be the option to enter BIOS or get some sort of feedback via POST?

Hardware:
Please list the specs to your build like so:
CPU: Intel I7 i770T
CPU cooler: be quiet! PURE ROCK 2 150W
Motherboard: Gigabyte Z390 I AORUS PRO WIFI (rev. 1.0)
Ram: two sticks, both from Cruical (MoBo only has two slots)
First: 8GB / DDR4-2666 / came with system; brought ~ 2.5 years ago
Second: 16GB / DDR4-2666 / from different system; should roughly be same age.
SSD/HDD: None ATM / Samsung 850 EVO 500GB / connected to SATA Port 0 on MoBo
GPU: none / Internal GPU of CPU and MoBo output (HDMI / DVI; both dead)
PSU: be quiet! System Power 9 Cable Management 400W / brand new
Chassis: Open air Frame Computer Case for ITX Motherboard (see Amazon)
OS: Proxmox VE 7 (so Debian stable)
Monitor: none, since server -> tested with old SEG TV and PiKVM -> no output from neither output

Backstory on the previous build:
The system was originally a passive cooled build from Tuxedo Computers.
It came with a power brick with 84W (see here).
The problem was, that the build did not allow for easy swapping of RAM or using the Front_Panel-Connector on the MoBo for PiKVM.
Hence I opted for a new build. Since is tight ATM, I went with the aforementioned open case. It was cheap and it allowed easy access to the board and so usage with the PiKVM.


Thoughts and tests so far:
- The MoBo, the CPU and the RAM sticks have worked previously over two years
- One of the sticks in the stick that came with the system; second one is later addition
- The RAM might have some bad sectors, but I would doubt that one or both are broken beyond boot.
- I have re-seated and experimented with the RAM-Sticks (single and dual channel)
- I do not have any other RAM sticks ATM; only these two
- re-seated the CPU-fan and the 20+4 connector
- NOT re-seated the CPU-fan-connection to the MoBo, not the 4+4 connection between CPU and PSU
- The PSU-cables and the cable from the CPU-fan can only be connected one way, so I don't think I put them in upside down or something like that
- re-seated the Front-panel connection -> no problem with the Reset-SW; even when removed -> no avail

The PSU delivers 400W of power. My CPU has a TDP of 35W; the fan 150W. Altogether, the CPU draws 185W. means the meagre rest of HW has (on paper) 215W of power at their disposal. So the PSU is not too weak.

I have also experimented with the RAM (no RAM, single stick, double stick reverse, etc.) -> no avail, constant PC; no video; no nothing. Currently could not find any spare RAM, these two sticks is all I got.

Could it be, that this is a issue on the CPU-Socket? I have messed up pretty bad and at one point got thermal paste on the CPU-socket, as well as the (underside of) the CPU. I removed all of it with a tooth brush and isopropyl alc. Looked pretty clean to me. But do you think, that this is perhaps an issue with the CPU and perhaps the sockets or the cooling? I have installed the CPU-fan myself twice now, within the timespan of 24hours. So dried up thermal paste can not be the problem.

On a related note, I would like to rule out overheating, right? Do you agree?
The cooler is brand new and I installed it twice. The previous system was passively cooled so I *think* there is even a chance, that the CPU was undervolted.
Plus if the CPU got to hot, would it not just run slower and cap out at 100 deg. Celsius? Plus It should cool down over night to at least get me to one POST at least, right?

On the front my MB has a row of LEDs - eight in total. During boot up, after a while (maybe five seconds) these LEDs all start to light constantly in some sort of yellowish colour. Never seen them in another colour. Or before TBH. They stay in that state constantly, no matter the state of the machine. Once the are lit and the contraption has power, the LEDs are lighting up the room. If I hard kill the machine, the LEDs stay lit. (Probably trivial, but I thought I mention that).

There is no debug interface installed on the MoBo . No speaker and no Hex-LCD-Display.
A connector for a speaker does exist, but the manual does not have a section about error-codes. Please see link below for PDF manual (in english).

https://download.gigabyte.com/FileList/Manual/mb_manual_z390-i-aorus-pro-wifi_v3_e.pdf?v=6607af19a8a9d88e9bb0c8e6575035ce

Should I even bother to spent money on a speaker? I know it's not much, but wasted money is wasted money.


Apologies for the disjointed text and any important stuff I forgot and left behind the wall of despair and rage, that has build inside my mind out of frustration over the last four days.

Kind regards
Max

EDITS:
- Updated the form of the HW-List and added Amazon-link to chassis. I have seen two videos on YT about the same chassis and both used different names for the kit / board / case and named different vendors. Hence I believe providing the link is better and faster.
- Also added backstory about prev. build

 

[Lutfij]

CPU-Cooler = be quiet! Pure Rock (brand new)
If this is a new cooler, what were you on prior to the Be Quiet! CPU cooler? Try and relieve the stress on the mount and see if the issue is averted.

Should I even bother to spent money on a speaker?
You call the above rig as a test bench, you should think that anything you have in terms of troubleshooting even parts are something worth having, not wasted money. + A POST Speaker is invaluable to help you diagnose when you're flying blind with motherboard. Doesn't help that your motherboard doesn't have a Flash BIOS option whereby you could revive your platform if the issue was indeed caused due to a corrupt BIOS.

Try and clear the CMOS on your motherboard. You should also source at least one stick of ram to help identify the issue to be your rams in itself. Curious though what were your specs prior to the one listed above?

Please list the specs to your build like so:
CPU:
CPU cooler:
Motherboard:
Ram:
SSD/HDD:
GPU:
PSU:
Chassis:
OS:
Monitor:
include the age of the PSU apart from it's make and model.
^ We should be seeing two sets of specs, one for before and one for after.

 

[MaxMillrose]

Hey and thanks for the reply.
CMOS-Reset and Speaker, I will try and get tomorrow.

1.) Prior to the QB-Cooler, there simply was none. The original build was passively cooled.
(see below or first post)
2.) Old PSU was an external power brick with 84W .

I updated the spec-list in my first post.
Age of PSU is brand new. Make is "be quiet!" and Model is "System Power 9 Cable Management 400W"

The system was originally a passive cooled build from Tuxedo Computers.
It came with a power brick with 84W (see here).
The problem was, that the build did not allow for easy swapping of RAM or using the Front_Panel-Connector on the MoBo for PiKVM.
When I last tried to remove the MoBo to connect my new PiKVM, I had to disassemble the case completely and still got thermal paste all around the CPU.
After I took it out of the socket, I even got TP on the CPU. That was the breaking point and the last straw. I was done with this passive build. It's always a hassle to do the most mondane things.
But one you actually clean the CPU and the socket, you have a good CPU lying around.
Hence I opted for a new build. Since money is tight at the moment, I went with the aforementioned open case. It was cheap and it allowed easy access to the board and so usage with the PiKVM.


When you say "relieve the stress" -> Should I adjust the backplate or the front plate or both? Were should I best start?