This morning I was waking up my 7910 from sleep to start work, and it immediatly sounded like a plane about to take off and threw blinking amber lights.
I went to see what this amber light meant, and 2-2 means PSU or Motherboard problems (a bit vaigue but ok).
I started by unplugging and pulling the PSU out, and after 15 minutes i plugged it back in, no change.
I started to unplug my video card (3080ti), no change
then i removed all my ram and left 2 dims
and it booted
plugged in video card, still booted
installed ram 2 dims at a time and it booted everytime
I ended up with a complete working system
I ran a DELL system checup, and it came out 100% good
3 hours later it went out cold, same issue no blue screen just, power out.
I tried to unplug for some minutes and then restart and
sometimes it would Post, and while starting windows, no blue screen just, power out.
Then I waited 30 solid minutes, with the PSU out and system booted up again
Tomorrow I am going to further test with RAM testing every dim individually(paired since its dual cpu)
Could this be a worser scenario ? could i be facing motherboard or PSU problmes ?
Sounds more like a PSU than an MB. If the MB were bad, I don't think you would be getting any starts whatsoever.
But before we rule that it might be the PSU, replace the size 2032 CMOS battery. Bad batteries cause seemingly unrelated crazy problems, and they affect BIOS.
@bradthetechnut I agree, bad motherboard would not post, and a faulty RAM stick would either throw me BSOD or just hang on POST.
I forgot to mention that sometimes, the power button will have 0 effect as in it wont even turn the workstation on.
there is that little test button on the psu, when i press it, the workstation goes into plane mode, for some moments and then stops (so i assume the internal components are all beeing correctly connected and have live electrical connectivity, however, as i read in other posts, this does not necessarily mean good PSU)
is there a way to bench test these PSU like the regular ones ? so I could get a readout of the voltage is outputting.
I love my workstation, and i want to use it for many years, but i dont want to go parts cannon on it.
I will do as you recommend and switch out the 2032 CMOS battery (that i had removed initially aswell when i started to debug this situation, also reset the CMOS and did the old power out, psu out, switch on a couple of times to clear the condensors)
If the RAM were bad, I don't think it would throw you BSOD, but would throw you an error code.
As for the battery, believe it. Besides boot problems, bad batteries have also made PC's throw erroneous error codes.
When the battery was bad in my 3620, I kept getting freezes that would result in BSOD. When I thought I needed to reload the SSD, I couldn't even do that. So I changed the battery and did a BIOS reset.
That's why with your scenario, I thought we better try the battery first. Decision was also based on the age of the machine. Good 2032 batteries, like Energizer and Duracell, will last around 5 years.
I didn't have you do a BIOS reset just in case you had anything you set yourself in BIOS. Decided to see if the battery would do the trick first.
Sometimes, a battery change would even give a more accurate error code.
Amazon sells PSU testers, but I don't know if they have them for server PSU's.
When the fans tend to spin up that way at boot, it's very easily the battery ( experienced on my own skin more than once on a few brands ) , and then is down to luck.
Some systems operate as expected even when the battery is a goner as long as the wall plug is kept ... plugged , and the current to the wall outlet is not interrupted ( power outage ).
Some other misbehave as you experienced instead ( and some don't care aside reinserting the correct time in the bios )
I have also realized my cord had been twisted and the ground wire was, not only exposed, the ligaments of copper were half way gone. That explained why i was getting zapped! i always thought i had bigger issues on the outlets.
Currently I am using my 7910, for the past 12h, and its pretty solid. I am a programmer so thro the day i put thos 56 cores thro some pacing. All is well.
I had my eye on a replacement PSU, its not expensive $60, not sure i would buy it and have it around...
PSU tend to deteriorate if used , then put away ( electrolytes deteriorate if not kept under load ), I would more or less trust keeping a new never used spare as ... spare, while i would not trust keeping an used psu disconnected and kept unpowered for an extended period of time.
It may work, it may work but the output could be dirty, it may decide to go belly up after a short while with a kaboom sound.
Buy when you need ( and so the seller limited warranty covers your purchase if the refurbished psu has issues during the first 15 or whatever days )
Honestly an old psu can be reconditioned (without pre-emptively changing the capacitors), but it requires a way to provide current in a controlled way starting from near 0 and slowly, very slowly, going upward to specs ( 220v - 240v - 110v ). The time and the slow scaling is a way to try to revive/regenerate a bit the properties of the electrolytes. It's not something done unless having the equipment.
bradthetechnut
9 Technologist
•
9649 Posts
•
40713 Points
0
1
Posted November 2nd, 2023 00:46
Sounds more like a PSU than an MB. If the MB were bad, I don't think you would be getting any starts whatsoever.
But before we rule that it might be the PSU, replace the size 2032 CMOS battery. Bad batteries cause seemingly unrelated crazy problems, and they affect BIOS.
(edited)