The tower system shuts down accompanied by a click noise.
The tower system shuts down accompanied by a click noise.
Hi,
I set up a system using detailed guidance from Aeacus about a year ago, found at the provided link.
System Specifications
MSI ACE motherboard
RYZEN 7950X3D processor
RTX 4060 graphics card
Corsair DDR5 memory
Enthoo Pro enclosure
Prime TX-1600 power supply
Dark Rock Pro 5 cooling solution
Noctua A14 fans
Everything has worked well so far, but a persistent issue has appeared.
I frequently back up mainly 2TB magnetic drives and double-check the data integrity over several days.
Occasionally, backups take longer than expected—often exceeding 12 hours. This means the machine has been operating under increased load for many days.
I’m attempting to switch storage from USB2 SATA drives to USB3 drives, mixing magnetic and some SSDs (such as 5TB magnetic USB3 and 4TB SSDs, plus a few 2TB and 1TB SSDs).
During one verification session, I left the PC unattended. When I returned, the machine had powered down despite lights still being on. I couldn’t restart it from the power button at the top of the Enthoo, so I unplugged it entirely.
Now, the situation varies: sometimes the system boots normally but then emits a click before shutting down, while other times the power cuts out before it even reaches the boot menu. This is quite annoying.
I’ve tried turning off all USB hubs—there are three USB3 hubs with 10 ports each (30 sockets), and more than 20 are active at once—but I usually only use a few. The remaining ones remain disconnected. I’m using Sabrent 10 hubs, which switch off when mains power is available or when powered via the hub socket. The goal was to reduce wear from frequent connections.
I also detached the SATA cables from two magnetic drives in the tower to rule out that device as the source. However, the problem continues. I can sometimes restart and the machine powers up again, but then a click occurs inside, and the Corsair rainbow lights stay on. The power button at the top doesn’t work for a full shutdown.
I’m still trying to locate the origin of the clicking sound.
I don't know the exact location of the click. It might help to focus on the PSU, as it seems likely where the noise originates. For relevance, check the BIOS version you were using for your motherboard. You could also try refreshing the BIOS on your board. Additionally, consider removing the discrete GPU and all connected peripherals to see if POST occurs. I suggested removing them to eliminate the chance that USB hubs are causing an electrical problem.
Fancy seeing you again here.
Probably the "click" is linked to the PSU.
Maybe one of the PSU's safeguards activates and forces a full shutdown to protect it, which is why the PC loses power.
It’s possible your USB hubs are overloading the PSU's 5V rail, causing the protection to kick in.
Lutfij mentioned something useful, so that seems like a solid suggestion.
Another possibility is an issue with the main electricity grid.
I’m not sure if you had an UPS to back up your system either.
🤔
But it’s clear now that the build worked well at first, and your problem started later?
BIOS Version/Date American Megatrends International, LLC. 1.E0, 28/03/2024
I give further hardware info below at the end of the message
I'll give the info first and we consider options before I try that in case the problem can be fixed at a more outer level. I would like to try the most lightweight remedies first!
this machine is dual BIOS, so I might try switching BIOSes before reflashing, but I'll see what replies on the info so far first, eg to maybe remove the GPU first.
I do have a 1000 Watt Corsair PSU in my earlier PC, RM1000X, where as another option to try substituting the Seasonic with that temporarily to see if the problem is the Seasonic. The Seasonic has a 12 year warrantee.
not sure what you mean by POST
the problem persists even if I switch off all 3 hubs at the mains. where they are still connected to the PC via the USB cables. these are all powered hubs, ie they have their own mains supply and dont draw power from the PC.
I give a link later for the specific hub I use.
could the powered off hubs connected by USB cables cause a problem?
ordinarily I just have 1 hub powered on, the 1st hub has the most frequently accessed items, the 3rd hub has the least frequently accessed items, so I only switch on the other hubs when I need items connected to them.
this also makes it easier to disconnect a swathe of different items with one switch.
the arrangement is so I can connect and disconnect items entirely with power switches, not by physically attaching or detaching cables, as the latter wears out the contacts. I use extenders also to reduce wear and tear ie hub -- extender --- extender --- item, where I only attach and detach extenders.
the USB hubs are the sabrent 10 hubs, which dont draw power from the PC. namely the following which you recommended last year:
Sabrent USB3 10 hub
where each socket has its own power switch, and the entire unit has its own mains cable.
3rd June there was a major power cut, the first in almost 20 years, where the lights wavered and then there was a flash of light and a sound, the power supply firm said it was caused at a nearby road, not by anything I was doing! I saw their van there in the afternoon fixing it, there was no power at all for several hours. The PC was in the middle of a lengthy disk copying op when that happened, so I had to restart that all over again.
but the problem emerged yesterday 8th June, where I had meanwhile done a lot of lengthy backups on Linux, many overnight, with no power loss to the machine as the shell commands were still running in the morning.
the first power cutting out yesterday I was going for a walk so dont know what happened, but the subsequent ones, I didnt notice any wavering of the lights, where usually I have the lights on even in the day as the room is shielded a bit from daylight.
I didnt because long ago, approx 2005 I got a UPS, but when I tried it, I could feel some buzz in the air like there was an ocean of invisible stuff being held back, and I got scared of that and never used it again. I was going to sell it on ebay but when I tested it, found it was defunct, so I had to junk it! I noticed then that the documentation said you had to use it from time to time!
the build has worked since I built it, I think over a year without problem. A few times when I thought it wasnt booting, it was always that the monitor had powered off completely and I had to power it with its switch and then eg the Windows desktop would appear! ordinarily, I only need to switch the mains on for the monitor. but if I leave the computer and monitor running overnight, often I have to also press the integral power switch of the monitor. I did try that with the initial problem, but nothing happened, I tried the mouse, the keyboard, nothing. the mouse light was off also. I detached and reattached it to the USB extender and the light reappeared but wouldnt start the machine.
the power switch on top of the enthoo pro wouldnt wake up the computer with single presses, and where ordinarily a prolonged press powers off the PC fully, nothing happened! I had to use the mains switch to power off. same problem with the subsequent loss of power, but unusually the corsair modules rainbow lights remain on during these problems. that's why I thought it was hibernating. so it doesnt cut all power, power continues to the corsair memory modules.
the system gives the info as:
OS Name Microsoft Windows 11 Pro
Version 10.0.22631 Build 22631
Other OS Description Not Available
OS Manufacturer Microsoft Corporation
System Name ************
System Manufacturer Micro-Star International Co., Ltd.
System Model MS-7D69
System Type x64-based PC
System SKU To be filled by O.E.M.
Processor AMD Ryzen 9 7950X3D 16-Core Processor, 4201 Mhz, 16 Core(s), 32 Logical Processor(s)
BIOS Version/Date American Megatrends International, LLC. 1.E0, 28/03/2024
SMBIOS Version 3.6
Embedded Controller Version 255.255
BIOS Mode UEFI
BaseBoard Manufacturer Micro-Star International Co., Ltd.
BaseBoard Product MEG X670E ACE (MS-7D69)
BaseBoard Version 1.0
Platform Role Desktop
Secure Boot State Off
PCR7 Configuration Binding Not Possible
Windows Directory C:\Windows
System Directory C:\Windows\system32
Boot Device \Device\HarddiskVolume16
Locale United States
Hardware Abstraction Layer Version = "10.0.22621.2506"
I possess a set of multiplug adaptors connected to a cable from the wall mains socket. The PC is plugged into one of these adaptors, followed by two more multiplug cables, and then additional cables from further adaptors. In total, there are six multiplug adaptors with 27 items connected at all extended sockets, including the adaptor plugs, though most remain inactive most of the time.
Currently, everything is just the multiplug adaptors and a monitor, a PC, one 10 hub, a laptop, and a USB2 enclosed SATA magnetic drive that I’m transferring to a newer USB3 5T drive (Seagate One Touch, drawing power from the laptop’s unpowered external hub). The transfer will take about 17 hours because the drive is slower technology. All other 27 items are off.
The laptop allows me to keep backing up non-M.2 drives. Could using the first multiplug with the PC connected cause any issues, especially since it’s quite old? Should I replace that multiplug with a new one? Also, might the downstream adapters affect the upstream connections at the PC’s plug?
I believe all the multiplugs have fuses in place. The setup is as follows: mains → multiplug (with PC, monitor, 1st hub, etc.) → two more multiplugs → additional cables → further adaptors.
I suspect your PSU may have suffered damage when the AC mains supply failed in the street outside your home. Uncontrolled high voltage mains transients caused by arcing (as indicated by a bright light) might have zapped the MOV in your PSU or some other component (semiconductor).
Apparently you're losing at least one of the main DC power rails (12V, 5V or 3.3V) so the system stops working. Perhaps your Corsair rainbow lights are fed from the auxiliary +5VSB rail which is present all the time the PSU is receiving mains power with the rear panel on/off switch set to on.
The click you hear from some ATX PSUs is the "bypass relay" associated with a Thermistor (Varistor) in series with the Line input. After the initial high current switch on surge, this relay kicks in and bypasses the Varistor, to increase PSU efficiency (the volt drop across the Varistor wastes heat unnecessarily).
If you have a different PSU from another PC, try substituting it before doing anything else. I'd advise uplugging all your many hubs before doing this, but top marks for using connector savers. I got through several micro-D savers each year at work costing $200+ each.
A thought regarding your powered hubs. There's a vague chance the +5V supplies (from the power bricks associated with the hubs) are feeding back into the computer. This would happen if the supplies are not completely isolated (diode-ORed) inside the hub. You could check with a multimeter, but's not always easy to achieve.
The next time your computer fails, try unplugging the USB connections from all external hubs and drives and see if the Corsair rainbow lights go out. You might be forcing +5V from the USB hub power supplies back into the motherboard and keeping the rainbow lights illuminated. Stranger things have happened.
My personal preference is to avoid using USB hubs, so I mothballed a collection of 11 USB3 desktop hard disk drives several years ago. The 3.5" hard disks (2TB to 8TB) get too hot for my liking (up to +60°C) inside their plastic housings, with limited ventilation and no fans. If you extract the hard disks from WD Elements or Seagate housings and fit them in a desktop PC with direct fan cooling, the same drives run at +30°C to +40°C.
https://www.ifixit.com/Guide/How+to+Shuc...ive/137646
I feel much happier with nearly all my hard disks running cool inside PCs. A typical Fractal Design R5 case has bays for eight 3.5in hard disks. I have Lian Li cases with room for twelve to sixteen 3.5in drives. Better (to my mind) than a bunch of external USB drives running near their maximum temperature limit.
From 2018 onwards I've been using a couple of cheap second hand LTO4 SAS tape drives to archive files to 800GB tapes. The drives cost $100 each and some barely used tapes were $3 each. With the write-protect tab engaged, the tapes are pretty safe from accidental file deletion or Ransomware. Tapes are nowhere near as handy for fast file retrieval, but they're good for archives.
A USB hard disk is at risk all the time it's connected to a PC attached to the internet. I keep a minimum of 3 copies of all important files on different systems, sometimes up to 6 copies.
Tip.
I use Free File Sync for such tasks. If the copy process is halted or fails, Free File Sync can be used to check the difference between two folders or complete drives and resume file copying without having to start all over again.
All you need to do is Compare the contents of the two folders/drives. I place the Source folder in the left pane and the Destination folder in the right pane, click Compare, then click Synchronize. Easy.
https://freefilesync.org/
With this information, I concur with Misgar's opinion;
Check the PSU by swapping it out using your spare Corsair RM1000x; also remember to replace the modular power cables from the PSU, as Seasonic cables won't work with Corsair units.
After replacing the PSU and ensuring stable operation, I recommend returning the Seasonic device under warranty—you might receive a free replacement PSU.
Looking ahead,
I advise investing in an UPS to safeguard your PC. I can provide detailed guidance on choosing and using UPS devices.
Because main power outages can occur unexpectedly, a UPS protects all connected hardware, keeping your system running during blackouts so you don’t lose unsaved work.
Otherwise, you’re taking a risk; another grid failure could damage your PSU again soon.
POST = Power-On Self Test.
The initial system test is performed when the PC powers up. It prompts you to enter BIOS before the OS loads.
many thanks for the extensive reply!
I need to digest the info, before I read this I did buy a brand new 6-multiplug, with 2 USB A and 2 USB C sockets today, and put just the PC, monitor and one 10 hub on it from a different wall socket, next to the socket currently used.
after maybe more than an hour in the middle of an email reply, the click happened and the power cut from the PC. I will try that again and then to try your idea of disconnecting the USB hub.
its a bit of a performance to change the PSU, so there may be a delay before I get this completed!
what is the reason for the power switch at the top of the enthoo not working when the power cuts out?
or maybe its what you say in the following paragraph about the corsair getting its power as a backflow from the USB hub
so I will test that idea when it next happens.
I use the Sunguy extenders recommended by Aeacus last year:
sunguy USB 3 extender
because I junked so many USB 2 hubs where the contacts were worn, and eventually all USB2 sockets of my old 2010 PC were unusable! I then had to use a USB3 PCI card to get 2 USB3 sockets!
Spoiler:
expand for slightly off topic info about minimising wear of contacts
I first learnt about this problem from a Nextbase dashcam, where I was moving the SD card to and from the PC to access the recordings. and at some point vids were becoming corrupted, eg freezing with a green screen. Then I exchanged SD cards between the 2 dashcams, one at the front of the car, one at the back, and now no problem on the affected dashcam. so realised it was the SD card, and its just the contacts wearing out. if you have a chip and pin credit card, and use chip and pin a lot, the chip becomes seriously worn out and can then stop functioning. Since 2012, cards moved to contactless, but larger payments and every now and then as security they force you to use chip and pin! so now I never remove SD cards and use the USB cable instead. With the Nextbase, I cannot keep an extender as it collides with the fixture on the window. but when it wears out, I can then use the SD card just once! I bought a 2nd hand 3D camcorder recently, and have a permanent extender to connect to a permanent extender on the 3D bluray TV recorder machine, both by Panasonic. It takes 2 x 64GB SD cards, the large format. I use the remote control also instead of the touchscreen to minimise wear and tear of the screen! Pioneer no longer make the dual 3mos 3D camcorders, which have 6 CMOS's so you get super impressive colours, many 3D camcorders use just 1 CMOS for the 6 colour components of the 2 lenses! where they split the image to one sensor, which shares red, green blue.
with my new Samsung smartphone, I only use a magnetic charger, that way no physical wear of any socket. Also I transfer files to my PC wirelessly using Samsung Dex, but today since installing an update, Dex isnt functioning, and instead I transferred a photo as an email attachment!
with the PC in particular I never use the BIOS flashing USB socket except to flash the BIOS because if that wears out, you cannot flash the BIOS again!
would this only happen for things connected indirectly to the same wall socket?
ie 1 wall socket ---- multiplug --- multiplug --- multiplug ... items,
ie are the 2 wall sockets isolated from each other?
this is in Britain, where I have a modern "consumer unit", where its not fuses, but where switches trip if there is a problem, and you just flick the switch back. Older houses have fuses, but I got mine modernised as by law you need the new system for new installations of kitchen electrics. "consumer unit" is probably british jargon.
I will definitely try this, when it next happens.
most of the time I just use the 3 M.2 drives, I only connect other drives for things like backups.
I think the M.2 drives get seriously hot also.
Now I noticed that if I do a major backup of say M.2 drives, the fans come on full power!
so I think its when you access the drives that they really heat up. data is basically hot!
right now the 17 hour backup of a 2T magnetic 3.5" sata drive (WD Blue) in a USB2 enclosure has completed, and it has gone into hibernation! I dont know if the hibernation is from the USB enclosure or the PC or the sata.
I try to make backups to a different disk each time, with the backup disks powered down otherwise.
these are huge backups, eg compressed could easily be 300GB, thus making 2 identical ones is very time consuming.
Spoiler:
expand for further about necessity of a robust backup arrangement
this actually paid off recently when a 500GB Samsung SSD went completely defunct. And is where I think magnetic drives are more reliable, in my experience magnetic drives slow down before they go completely kaput, where you can salvage the stuff. but this SSD went from everything to nothing, so salvaging not possible.
the USB2 enclosures not so reliable, but maybe the ones I got are low quality ones, there wasnt much choice.
connected to sata it does appear, on Linux you can access a disk as if it is a file, eg /dev/sda or /dev/sdb etc, which you can read off from the Disks program. When I did this, attempts to access it, which require the sudo command, say it has 0 bytes.
luckily I have a sector clone as a file of the entire drive in 2021 which has other volumes also,
and a compressed archive of the main work directory from 2023, where I have only made some tiny changes since then. And I have regenerated the sector clone to a 1T T9 Samsung drive, and later will manually merge the archive with that, I need to study the timestamps to compare the 2 systems to merge them.
had I not been so paranoid about backups I could have lost a lot of my own works!
as regards 3 copies, what I do is archive each time to a different disk. so I have archives going back to 2006 for my work directories, which generally arent huge, as its mostly text based things, and for backups of drives, when a disk runs out of space, I delete the least recent ones of any where there are more than 2.
thus I have copies on different disks, but from different points in time. eg I might have a sector clone of the system drive from May on one disk, and from April on another disk. if I attempt a repair of Windows, I make a sector clone of the entire drive via Linux, because in the old days I didnt, and at least 2 repairs got stuck, where the windows installation became unusable! luckily I never keep work on system drives, as system drives are the most used, as they are used every time you boot the machine, so they wear out first.
I dont know about the newer volume formats, but with say 32 bit Windows XP, the volume format of hard drives begins at sector 0, so that sector gets worn out first! I had a Samsung drive, where the first sector was corrupted and unwritable. I think I could salvage the data, but the disk was no longer usable.
thus emails, photos, text, NEVER on the system drive, I dont use those "Documents", "Pictures", "Music" etc folders that Windows supplies! Always such things to other drives, and backups to drives which are powered down most of the time.
does it preserve the timestamps also?
the click happened whilst I was replying to you!
as I could no longer read what he said, I removed the USB3 hub's power socket, where that new multiplug just has monitor, PC and 10 hub connected.
but the rainbow lights continued, I waited many seconds, then switched off the mains to the PC and now the rainbow lights vanished.
afterwards I realised maybe I need to also disconnect the USB cable to the PC, where just 1 extender is used from the PC as the hub cable is relatively long. Next power cut click I'll disconnect it at the extender and at the mains to see if the corsair rainbow lights stop.
then whilst writing this, the power cut out again!
and I forgot to disconnect the USB cable from hub to PC.
this time I am posting from my laptop, with the PC tower booted up to windows 11 on the other table waiting for it to fault. I made a really small table for the laptop, and this laptop and mouse can handle 3 computers, so I can use the same wireless keyboard and mouse for both. one technical problem regarding this is the laptop mouse pad is worn out where clicks dont work properly, which on Linux makes it problematic to pair the mouse and keyboard. The only way out would be to use a wired mouse. Once paired, next session it remembers, but sometimes it doesnt! and it has to be re-paired.
this is what I am dreading!
not sure what you mean by RMA?
I may contact them first before exchanging PSUs, should I volunteer the info that there was the power cut?
or should I wait till they ask?
I wont lie about it, but with businesses sometimes you have to be careful what you say and what you dont say, as they can use either to your disadvantage. Thus often to only give info that is requested. With government officials also its like this, you should never volunteer info, neither positive nor negative.
whereas with this forum I will give all the info I can think of.
if they dont replace it, I may try the house contents insurance,
will the UPS protect the PSU from this kind of power cut damage?
as regards the saving, the problem is if the power cut is whilst I am asleep!
as I often backup disks whilst asleep! eg the last backup I did took some 17 hours!
some take 8 hours, some take 3 hours.
thus I may lose some unsaved work!
I often start a command last thing at night, then after breakfast I check whether it completed!
I can tell often also if the hard drive led is flickering that the commands havent finished!
I put some echo commands in the scripts to tell where it reached if a problem, eg
date > markerfile1.txt
shell_command1
date >markerfile2.txt
shell_command2
date >markerfile3.txt
I may be able to tell from the file sizes also if a command completed.
I had wrongly assumed a Seasonic would have its own shielding from this kind of problem.
but it does this test for whose benefit?
if the test fails, does it communicate this to the user somehow?
it is presumably not for the computer's own amusement?
especially with AI!
😕
this mobo does have an led with numbers which change continually during the boot, this is where the transparent door is a good idea!
also where it pays to print out the manual! or to have a 2nd pc.
some of my early PCs didnt have transparent sides, so if there was any signal one wouldnt know! I think this is a problem with laptops also!
Because you've already put the whole system together yourself, swapping out the power supply isn't such a challenge as, for example, installing MoBo into the case and connecting all necessary cables. Right now, we're not certain whether the problem is with the PSU (possibly damaged during a blackout) or with your USB hubs. To eliminate both possibilities:
1. Take OFF ALL USB hubs connected to your PC. Keep only the power cable for KB/mice, monitor, and the main power connection.
This might mean one of your powered hubs is sending power back to MoBo, as Misgar mentioned. Even if you turned it off from the control panel, it could still supply some power to the PC (the button may not work properly).
2. Swap in a spare PSU.
If the issue involves the Seasonic TX-1600, using a different PSU—especially one that's known to work—would immediately confirm whether the problem persists.
RMA = Return Merchandise Authorization
Further reading:
https://en.wikipedia.org/wiki/Return_mer...horization
This process allows you to file a warranty claim for the PSU and return it to the manufacturer if the fault occurred within the warranty period. You should describe your symptoms clearly, mentioning any blackout events that might have caused lasting damage. However, be aware that if you don't mention a blackout and hope Seasonic will replace the PSU, they may test it to determine the cause. They could find an external factor (like a blackout) as the reason for failure. After testing, Seasonic may decide whether to honor the warranty, since warranties protect against manufacturing defects, not external events such as power surges or environmental issues.
This situation is typical of what the UPS is designed to handle. It really doesn't matter if you're awake or asleep during a blackout—once power is lost, the PC can't stay running. The PSU can only maintain operation for a very short time (about 26.3ms), which is insufficient to register even a brief shutdown. Human reaction times are around 250ms for sight, 170ms for sound, and 150ms for touch. The Seasonic TX-1600 can sustain the PC for just 26.3ms, which meets the ATX standard minimum of 17ms.
If you had a reliable UPS, it could handle grid fluctuations (like power cuts) within 2–4 milliseconds—long enough for the PSU to keep the system running before switching to battery backup. Although the PSU has basic protections like this hold-up time, it lacks an AVR and a battery for sustained loss. For that level of protection, you need a UPS.
The PSU converts AC power into DC and distributes it across various voltage rails (e.g., +12V, +5V, +3.3V) for the PC's components. If all ATX PSUs were fully equipped to handle grid disturbances, there would be no need for a UPS. However, each PSU only protects against its own issues and not external factors like blackouts or surges.
Once the PC powers on, these steps are carried out in sequence before you can open Windows. Human reaction time is about 250ms for vision, 170ms for audio, and 150ms for touch. The PSU can only maintain operation for a fraction of a second during a blackout—insufficient to register any meaningful shutdown.
For more details, see the slide show describing POST behavior (links provided). It shows how errors appear on screen or through beeps, but often no visible output occurs during critical failures, leading to a complete system failure.