Old PSU has died, need replacement suggestions quickly.

Associate
Joined
23 Sep 2012
Posts
124
Location
South Yorkshire, UK
Hi folks Need some help here.

Ok,....story time.

A few days ago, wednesday maybe, my computer (Built 3 months ago) suddenly locked up and went into a BSOD/restart cycle, showing a WHEA_UNCORRECTABLE_ERROR stop code, so, after trying and failing to re-install windows, thinking that was the issue, i ran memtest, that through a couple of errors at me, so i replaced the ram yesterday with some QVL listed 3600Mhz memory (the old memory was 2666MHz and wasn’t on my board’s QVL list, as it was recycled from my old intel system, as was the 850W PSU that has now died)

After installing the new ram, i tried to shut down/restart the system by pressing the power and reset buttons on the case, neither worked, so had to do it through the bios. Now that i think about it, that should have been my first warning there was something wrong with the PSU, but i didn’t consider that.

ALL Attempts to reinstall windows from a usb failed, it would get to stage 2 of the install process, get to about 5%, then go back into a BSOD/Reboot cycle.

Then i went into My bios’s flashback utility, and i noticed that my boot drive was reading as having data on it, a full windows install, which it should not have had, because I’d previously wiped it, so i wiped it again, ran a smart selftest, that came back fine as well.

Next i ran memtest again with the new memory installed just to be sure, that came back clean.

I tried running Parted Magic from a usb, but got an error while trying to load it saying my ssd was missing/had invalid SUBNQN field, which is apparently a firmware issue.

SO, i tried loading up another Linux distro on another usb to load up a disk manager app sebrant released to update the nvme ssd firmware, the second distro wouldn’t boot (Tiny Core Linux), so i tried Parted Magic again, i somehow got that to load, but it would not let me run that disk manager app, It’s own disk health app wouldn’t recognise the nvme ssd, but the partition editor app did see it.

So having no joy with the sebrant rocket ssd firmware update attempt, i thought “Ok, time to roll back the bios, as that was the last thing i updated before it crashed”, but i didn’t get the chance.

After shutting down the system from parted magic’s desktop, i tried to power on the system again to roll back the bios, but it Would. Not. Power. On.

So after a quick talk on discord, with some people who are more minded that me, they concluded my psu was dead.

Now I’m worried it’s damaged the ssd or motherboard, and that board was expensive.

Sooo....I need a replacement PSU for a system with the following components.

3900X cpu,
H100i AIO cooler,
x570 ITX ASUS motherboard,
2070 Super,
32 gigs of RAM (brand new),
1TB Sebrant nvme ssd as OS and application drive (hopefully not bricked).

Will 850W be enough or should i go higher? Someone i spoke to on discord seemed to think 850 watts was a bit tight, considering what I’ve got in the system. (The old PSU was 850W)

And what are the chances the rest of my system has survived intact? Since now that i think about it, the failing psu could be the reason the ram i also reused from my old system was showing faults in memtest (plus it’s also 5 yo, and wasn’t on my new board’s QVL).
 
Associate
OP
Joined
23 Sep 2012
Posts
124
Location
South Yorkshire, UK
I haven’t tried the paper clip method, not even sure what that is.

If the system doesn’t power on when the power button is pressed (and it’s switched on at the back and on the main outlet), then logically, the psu should be dead, yes?.

I’m not planning to get a 3000 series card, couldn’t afford one even if i wanted one, and i don’t.

So you guys are saying 850W should be enough for my current build? I don’t need to go 1K watts? I just like having a little bit of extra ‘room’ in terms of avialable power, i don’t want to starve the system with a psu that only just meets requirements.

The old 850 lasted 5 years or so (just under 5 years actually), so I’m thinking it might have failed due to age rather than strain, but still, I’d rather go overkill on the psu than not have enough, and really run the risk of frying something i potentially can’t afford to replace rn.
 
Soldato
Joined
17 Aug 2003
Posts
20,158
Location
Woburn Sand Dunes
I haven’t tried the paper clip method, not even sure what that is.

If the system doesn’t power on when the power button is pressed (and it’s switched on at the back and on the main outlet), then logically, the psu should be dead, yes?

no. The motherboard tells the psu to turn on. If the motherboard is dead, it wont be doing anything and buying a new PSU wont solve it.

get the psu out of the case and test it.

https://www.evga.com/support/faq/FAQdetails.aspx?faqid=59582
 
Associate
OP
Joined
23 Sep 2012
Posts
124
Location
South Yorkshire, UK
Well, i just tried turning it on again, and it’s completely dead, even the case fans and board/fan LEDs won’t turn on. So it’s either the psu, or the motherboard is dead. Can’t do the paper clip test at the moment, as I’ll need help dismantling this thing to do it.

RMAing the board will not be fun. That was expensive.

Though remember, this all started after that uncorrectable error stop code i got on Wednesday, after the latest ASUS bios (2606) update, plus a windows update some time before that, maybe a few hours, i don’t know. But the memory was definitely showing up with errors in memtest, but the new ram i installed yesterday did not show any errors.
 
Associate
Joined
11 Feb 2019
Posts
100
So you guys are saying 850W should be enough for my current build? I don’t need to go 1K watts? I just like having a little bit of extra ‘room’ in terms of avialable power, i don’t want to starve the system with a psu that only just meets requirements..

Definitely don't need 1k watts, watch the video! For the PSU to be efficient it needs to be operated within a certain range (50-75% of max I think?), so if you overbuy your PSU it won't be very efficient. 850w should be plenty of safety margin if you want to overclock
 
Associate
OP
Joined
23 Sep 2012
Posts
124
Location
South Yorkshire, UK
Ok, slight update, i just flipped the switch on the back of the psu, and the motherboard’s LEDs are ON, but the system still won’t power up, nor do the case fans or their leds work. Someone I’m talking to on discord still thinks its the psu, but it’s just not sending enough power to the board, but if the psu was dead, the leds shouldn’t work at all.

I mentioned the paper clip method to my dad, but he didn’t seem convinced it was safe to stick a piece of metal into something that’s plugged into a mains socket
 
Soldato
Joined
17 Aug 2003
Posts
20,158
Location
Woburn Sand Dunes
Ok, slight update, i just flipped the switch on the back of the psu, and the motherboard’s LEDs are ON, but the system still won’t power up, nor do the case fans or their leds work. Someone I’m talking to on discord still thinks its the psu, but it’s just not sending enough power to the board, but if the psu was dead, the leds shouldn’t work at all.

I mentioned the paper clip method to my dad, but he didn’t seem convinced it was safe to stick a piece of metal into something that’s plugged into a mains socket

It's perfectly safe. look at the website i posted the like to - EVGA. they manufacture, amongst other things, PSUs. They're telling you how to test their own PSUs. It's dead easy, not rocket science. Just follow the instructions and you'll be fine :)


anyway yes, light's on the motherboard = psu not totally dead. until you know for sure the psu wont turn on, do not buy another. Especially given that the pc died after a bios flash, i have no reason to believe it's not the motherboard at this stage. What you can do is pull the battery right out and leave it for an hour as sometimes just resetting the CMOS isnt enough to clear the bios entirely, if fact i had a board that wouldnt boot at all until i pulled the battery.

Paperclip test is fine and you don't need to disassemble everything. Just remove the 24pin plug from the MB, you don't need to take everything apart

No, he needs to unplug everything. IF the psu is faulty (i don't believe it is, but IF) then the last thing he wants to do is force it to start up whilst it's plugged in to everything else.
 
Last edited:
Soldato
Joined
7 Jul 2004
Posts
7,784
Location
Gloucester
No, he needs to unplug everything. IF the psu is faulty (i don't believe it is, but IF) then the last thing he wants to do is force it to start up whilst it's plugged in to everything else.

When I tested one I was using for mining a few years back I left everything plugged in when doing it.
Looking back on it and taking onboard what you've said I probably should have unplugged everything. Oh well :D
 
Associate
OP
Joined
23 Sep 2012
Posts
124
Location
South Yorkshire, UK
Ok, update.

The PSU is working fine, the issue of not being able to power up the system seems to have been as simple as plugging a few lose front panel plugs back into where they belonged, though the case’s front fans still won’t power up, and they’re plugged into a hub on the back of the case, so there’s probably another wire knocked lose somewhere. Bought a PSU Tester and plugged the ATX and CPU cables into that, and it worked fine, all the read outs were normal, according to a youtube video, so the PSU is definitely not the cause of my issues, as i previously thought.

The Motherboard’s POST state LEDs, which on my board are just above the ram slots (and are difficult to see when the 240mm Radiator is installed at the top of the case), seem to be indicating there’s a problem with the boot device, so we’ve removed the hard-drive and we’re mailing that back to amazon tomorrow for a refund, hopefully replacing the drive will be all I’ll need to do. (Will also need a second internal drive in case this happens again)
 
Associate
OP
Joined
23 Sep 2012
Posts
124
Location
South Yorkshire, UK
Well, not really, the refund will only cover the cost of the new nvme ssd, I will still need to buy a second ssd/3.5 mech drive as backup in case the replacement ssd borks like this one did, hopefully the TLC ram/Phison controller on the drive isn’t the cause of the failure, according to one of the reviews on amazon, there’s an issue with the phison controller on all the Gen4 M.2 drives, as they all apparently use the same controller. And apparently, the more layers the cache memory has, the less reliable it is. Problem is, all drives seem to be going MLC/TLC/QLC these days.

Plus I’m thinking of getting an internal usb hub, most likely the NZXT one, i got a Hubby7 internal usb hub, but haven’t figured out yet where to plug it into, and the instructions are well, not good.
 
Back
Top Bottom