Can failing hard drive cause system to turn off, unable to turn back on?

[DarkEnergy]

I'm trying to repair a friend's computer and wow does it have issues. I've installed a new cmos battery (old one was dead), different ram, and tried another psu but still issues. This computer dates early 2000s, windows xp 32 bit with a mix of IDE and SATA drives. The issue is that the computer randomly shuts down, like completely dead. Power switch doesn't do anything, PSU switch doesn't do anything. I can unplug the system and plug it back in and it still won't turn on. After maybe a minute I can press the power switch and it does turn on. Windows tries to run chkdsk but the system hangs at stage 3 of 3 (CHKDSK is verifying security descriptors). It gets to 30 something percent and that's it, no hard drive activity or anything. That said, I can bypass chkdsk using a keyboard with a PS/2 connection. Logs into windows completely fine. sfc /scannow doesn't seem to find anything. I'm trying to boot some tools from CDs but nothing is working (perhaps a bios setting, will have to dig further). I'm thinking either a motherboard or hard drive issue. Any thoughts? The behavior of the system shutting off and not wanting to turn on right away is rather strange.

 

The shutdowns seems less frequent the more often I have the computer on/using.

 

Windows xp is installed on two hard drives in RAID 0 (I'm fairly confident). Crystaldiskinfo doesn't support RAID. 

It's not the power switch, screwdriver method turns it on fine but issues persist.

CPU is not overheating. 

Athlon x64 dual core

2gb ram

[manikyath]

well.. i see several things that could be at fault here, but first off: why? please tell me this is a "retro" system for good old time sake?

[DarkEnergy]

2 minutes ago, manikyath said:

well.. i see several things that could be at fault here, but first off: why? please tell me this is a "retro" system for good old time sake?

I wouldn't call it retro, more like a mid-range brick shit house. My friend helped me get a new graphics card when mine died so this is more like a favor in return. My next step it trying a different hard drive with windows 10 32 bit and see how/if that works.

[manikyath]

1 minute ago, DarkEnergy said:

I wouldn't call it retro, more like a mid-range brick shit house. My friend helped me get a new graphics card when mine died so this is more like a favor in return. My next step it trying a different hard drive with windows 10 32 bit and see how/if that works.

if its for actually using the thing.. dont bother.

 

you can get something core i5 with 4-8GB ram from corporate surplus between 50-150 bucks. saves you the time of troubleshooting a 15 year old relic.

[DarkEnergy]

That's probably what I'll end up telling him. I just want to test a few more things to see if I can get it working.

[manikyath]

28 minutes ago, DarkEnergy said:

That's probably what I'll end up telling him. I just want to test a few more things to see if I can get it working.

i'll tell you why you shouldnt bother. based on experience, here's some possible outcomes in 15 year old systems:

- instabile power supply (possibly grid power related)

- broken IDE hard drive

- bad IDE cable

- bad capacitors anywhere in the system

- temporary short circuit (for example a part that shorts out when hot)

- worn out power supply connectors, making voltage delivery inconsistent

- probably some more stuff i'm forgetting.

 

here's how you *could* go about testing this:

1. full disassembly and look over parts for obvious defects. (bent pins, burst caps, craters where your mosfet used to be, ..)

2. re-mount the cpu and cpu cooler, if you have a cpu without a lid, spread your (non-conductive!!) thermal paste for guaranteed full coverage.

3. put one stick of ram in (swap it out between tests until you've got a known good in that specific system)

4. check EVERY pin on the 24-pin connector to make sure all are still straight and tight

5. assemble a system with the bare minimal hardware necessary for a post, on a clean non-conductive surface

6. if the above boots, add hardware one piece at a time until either all hardware is connected, or you've encountered an issue.

7. either you are left with a full assembly, or you have suspect parts seperated out. analyse these suspect parts with a different machine

8. try other hardware on the same ports/slots as the suspect parts.

9. i'm sure you figured out what to do next at this point.

[Mark Kaine]

3 hours ago, manikyath said:

i'll tell you why you shouldnt bother. based on experience, here's some possible outcomes in 15 year old systems:

- instabile power supply (possibly grid power related)

- broken IDE hard drive

- bad IDE cable

- bad capacitors anywhere in the system

- temporary short circuit (for example a part that shorts out when hot)

- worn out power supply connectors, making voltage delivery inconsistent

- probably some more stuff i'm forgetting.

 

here's how you *could* go about testing this:

1. full disassembly and look over parts for obvious defects. (bent pins, burst caps, craters where your mosfet used to be, ..)

2. re-mount the cpu and cpu cooler, if you have a cpu without a lid, spread your (non-conductive!!) thermal paste for guaranteed full coverage.

3. put one stick of ram in (swap it out between tests until you've got a known good in that specific system)

4. check EVERY pin on the 24-pin connector to make sure all are still straight and tight

5. assemble a system with the bare minimal hardware necessary for a post, on a clean non-conductive surface

6. if the above boots, add hardware one piece at a time until either all hardware is connected, or you've encountered an issue.

7. either you are left with a full assembly, or you have suspect parts seperated out. analyse these suspect parts with a different machine

8. try other hardware on the same ports/slots as the suspect parts.

9. i'm sure you figured out what to do next at this point.

Yuuup... that's the thing there are probably several parts broken at this point.  Imo as long such a PC works it's fine and whatever but then troubleshooting and replacing parts? Not worth it at all. 

 

[manikyath]

5 hours ago, Mark Kaine said:

Yuuup... that's the thing there are probably several parts broken at this point.  Imo as long such a PC works it's fine and whatever but then troubleshooting and replacing parts? Not worth it at all. 

 

i'd say best case scenario you spend 2-3 hours of rebuilding and cleaning parts, and have a working computer without replacement parts. worst case scenario your motherboard has gone unreliable and you'll have to either repair or replace.

 

it's a fun hobby project, but it's *not* a thing to keep running as a main rig anymore.