Feedback / Suggestions on a C$3k Build

[Strato]

No luck on anything I tried today, including one test each with each of the following:
- Removing the GPU.
- Removing the extra m.2 drive.
- Splitting the CPU power cable to separate the extra 4-pin header.
- Removing case fans connector.
- Removing RGB connector.
- Removing media connector.
- Removing USB C connector.
- Removing the front USB connector.
- Enabling / disabling XMP/Expo.
- Sacrificing a chicken and praying to Jobu (and still striking out - a major league headache!)

@JD – Your comment about the CPU fan header initially had me scratching my head since I knew the CPU cooler fans were running. However, I had plugged the cable into the fan controller at the top-rear of the case (which is meant for case fans). I switched it to where it belongs – thanks for catching that! Also, I *think* the cooler is installed correctly without it touching anything it shouldn't – It was a very straightforward installation. I agree that the problem seems like a short, though.

Next, I guess I'll fully disassemble everything and start over by only connecting just what is essential to get me into the BIOS, probably doing this outside the case. Then I'll power down and try to power back up. For that, am I correct that all I need are the power cables, CPU, and one stick of RAM (no m.2 drives at all, no fans / RBB / Audio / front case connector / USB C)?

If it works with that minimal setup, I can then add one connection back at a time until I find the point of failure.

Is it safe to assume that for this basic power-up test (no Windows and only for a short period of time) it is safe to not have the CPU cooler installed (even for chips like the 7800X3D)???

Not sure if I'll have time to tackle this on Monday, so it might be Tue or Wed before I have anything new to report. Thanks again for everyone's help.

 

[Bond007]

I would not power up the cpu without a cooler.

The fact that it always runs after a cmos reset really pushes me to thinking it is bios settings (or issue)/RAM. If it was anything to do with wiring, I would think it would always be a problem, or randomly intermittent (not consistently on the second boot)

Before you take anything further apart, I would go through the bios settings after a cmos reset and turn off (if it's on) anything along the lines of fast boot.

If that doesn’t work I would try single stick of ram as (IMO) the next most likely cause.

 

[Strato]

@Bond007 – Ok. Thanks. I'll do that. I don't recall seeing anything like fast boot enabled but will look more closely.

If that doesn't work, I'll experiment with the RAM – for that, though, I'll need to remove the CPU cooler first since the cooler sits above both sticks of RAM. Thanks for the warning about having the cooler installed even for a test power-up without any load; I wasn't sure about that. I'm glad I still have a bit of thermal paste left or I'd be stuck waiting on another delivery!

Hopefully I'll have some quiet time later today to experiment some more. Thanks again for your help and patience with my questions.

 

[JD]

Just removing that fan on the heatsink though isn't enough to get to the RAM? For these tests, I wouldn't worry about re-applying the paste, whatever is on the heatsink + CPU should be adequate to not cause it to immediately overheat. Cleanup the paste and re-apply once everything is in working order.

And when it doesn't boot, have you tried removing the power cable from the PSU and waiting a few mins/pressing the case power button repeatedly (to drain any leftover power). Does it then boot up or is CMOS clear still needed? There should be a setting named "ErP" in the BIOS too, could try setting that to the opposite of what it is currently.

 

[Strato]

Hey everyone – I made some progress (of sorts) today.

After removing all non-essential components (both m.2 drives and one of the memory sticks) and disconnecting all cables other than power and the F-panel, I still had the same power-up problems. However, after removing and then re-seating the F-panel cable (something I had done multiple times in the past…) I found I could reliably power-up into the BIOS repeatedly without resetting the CMOS in-between. This behaviour was not happening previously.

Victory, right?

Alas, not quite. Though one thing that we do know now is that the system is in fact *capable* of powering-up repeatedly without needing CMOS resets. That’s definitely progress.

Regrettably, it is very easy to trigger the power-up issue again (and it can be done in reliably reproducible ways).

After reinstalling all of the components and hooking up the cables again, I spent a lot of time experimenting. I’ve isolated the following situations which I think might (hopefully!) be helpful in identifying the root problem. Each of these situations appears to be fully reproducible (individually; they don't need to be done in sequence).

1. Do a CMOS reset. Power-up then hit Del to go into the BIOS. Don’t make any changes. Just power-down via the PSU’s switch. Wait 10 seconds. Power back on – works! No CMOS reset needed. The power-ups can be done as many times as desired without any CMOS reset.

2. Do the above, except this time while in the BIOS select the ‘Save & Exit’ option (with or without making any changes – same result). Let the BIOS re-boot, hit ‘Del’ again when the option appears to go back into the BIOS then power-down same as before using the PSU’s switch. Power back on – failure. CMOS reset necessary. Conclusion: even though no changes are actually made, the ‘save ‘ action in the BIOS still does something that triggers the power-up issue.

3. Do a CMOS reset, power-up, and let the system boot into Windows. At the Windows log-in screen, use the ‘shut down’ command to properly power-down the PC (i.e. without using the PSU’s switch). Power-up again – failure. CMOS reset necessary. Conclusion: something in the boot process into Windows is enough to trigger the power-up issue. Note, using the PSU’s switch at the log-in screen to power down has the same result, meaning the trigger isn’t something that Windows does during its shut-down procedure; the problem is triggered during the boot process.

4. Do a CMOS reset and power-up. When the BIOS start-up screen appears (with the ‘Press Del to enter BIOS’ option) power-down via the PSU’s switch. Power-up again – works. So, going into the BIOS itself isn’t necessary to maintain the CMOS’s integrity. Whatever Windows is doing that triggers the power-up issue is happening sometime between when the ‘press Del’ screen appears and when the log-in screen appears

5. Do a CMOS reset and power-up. This time wait until slightly after the ‘press Del’ screen clears and Windows begins its boot process then power-down using the PSU’s switch. Power-up again – works. So, the trigger is further along.

6. Do the same as above again except this time wait until roughly halfway between the ‘press Del’ screen and the log-in screen and power-down. Power-up again – works. So, the boot process for Windows doesn’t trigger the problem until at least halfway through its boot process.

Very strange behaviour! The same trigger is somehow caused by a ‘save’ action in the BIOS (even with no BIOS changes) or by something Windows does during its boot process. But what is the common underlying aspect there??? And why did all of the above become possible only after re-seating the F-panel cable despite my having done that multiple times previously – if there was something wrong with how the cable was originally seated, one would think that correctly re-seating it would simply fix the problem entirely not give the strange results described above.

I don’t have enough expertise to know what to make of the above but it definitely gives some interesting data.

Lastly, an update on things already suggested and other things I’ve tried:
- I tried toggling various BIOS settings (‘ErP’ and XMP/Expo for ex, making only a single change per test) without any affect.
- I’ve tried clearing residual power from the PSU without any change.
- @JD: You were correct that removing the fan gave enough room (barely) to remove the memory without removing the heatsink. Thanks for noting that!
- I tried various things with RAM including using only a single stick (and trying both sticks separately as the single stick), trying other memory slots, etc.
- Checked carefully to see if the cooler or its fans were coming into contact with anything and it all looked fine.
- I disassembled the front of the case and checked the manufacturer-installed front panel connectors. They all were seated firmly and looked fine.
- I checked all cables looking for any fraying or other defects. All looked good.

So that's where I'm at. Any ideas or suggestions on what I should try or look at next?

(It might sound odd to say this, but as much as I’d like my new computer to ‘just work’, trying to figure out this problem is weirdly fun – the hunt is compelling! LOL)

 

[djbrad]

Before going further, I would use the q-flash function and update to F6b bios, in case it fixes such bugs.

 

[Bond007]

I am running short on ideas other than it is one of the settings in the bios on "Auto" that is incorrectly making a different choice after "learning" your hardware on the first boot.

 

[Sagath]

Is the RAM on the motherboards QVL list?

DDR5 and AMD motherboards can be very problematic, and this seems like classic symptoms of incompatible ram.

Edit: https://www.gigabyte.com/Motherboard/B650-AORUS-ELITE-AX-rev-1x/support#support-memsup

Edit Edit: I looked up the wrong ram. It is supported on QVL
That being said, if JD's point about the front panel doesnt help, I would suspect an issue with RAM. Did you try a single stick of EACH stick when you did your troubleshooting?

 

[JD]

Have you tried without the front panel connected at all? If you look at the pin layout in the manual, you can short the 2 pins for the power button with your screwdriver to turn it on.

 

[Strato]

@djbrad – Hi. It's already been flashed to F6b.

@Bond007 – That's a possibility but nothing I've experimented with so far has made any difference. I'll play with options a bit more though.

@Sagath – Yes, RAM is on the QVL list (checked before purchasing as I know it can be quirky).

@JD – I haven't tried that yet. Will do so today. Thanks.