PSU Destroying SSD's??

[SpellowHouse]

Very curious about this.
I have had two Samsung SSD's fail in one PC in six months. Both exactly the same way. They apparently checkout just fine but are very, very slow. Samsung Wizard thinks the discs are fine, but Windows knows there is a problem but can neither say what's wrong or fix it. Attempting to delete a file and it just sits there doing nothing forever. The PC is extremely slow to shut-down. The disc apparently has data on it (about 10% full) but the files are zero. Applications have trouble creating and deleting files. All problems disappear if I just replace the drive with a new one.
I wouldn't be concerned but after the first SSD failed like this I replaced it and darn it the replacement has failed in six months.
I can't think what it could be, except maybe the PSU????

 

[Semple]

Why the PSU? There's lots of components in-between the PSU and the SSD that could cause a number of issues.

If the PSU was on its way out, I'd be expecting lots of other components to be acting up.

 

[Armageus]

Have you tried them in another PC - sounds more software related

 

[Tetras]

Very curious about this.
I have had two Samsung SSD's fail in one PC in six months. Both exactly the same way. They apparently checkout just fine but are very, very slow. Samsung Wizard thinks the discs are fine, but Windows knows there is a problem but can neither say what's wrong or fix it. Attempting to delete a file and it just sits there doing nothing forever. The PC is extremely slow to shut-down. The disc apparently has data on it (about 10% full) but the files are zero. Applications have trouble creating and deleting files. All problems disappear if I just replace the drive with a new one.
I wouldn't be concerned but after the first SSD failed like this I replaced it and darn it the replacement has failed in six months.
I can't think what it could be, except maybe the PSU????

There's a known problem with Samsung SSDs, mainly the 870 Evo, but bad firmware can apply to other drives too and they 'die' in exactly the same way you describe, where the quality of the data deteriorates and eventually they go read only (massive errors start showing up in their SMART data usually between 10TB - 15TB, including bad blocks).

 

[SpellowHouse]

There's a known problem with Samsung SSDs, mainly the 870 Evo, but bad firmware can apply to other drives too and they 'die' in exactly the same way you describe, where the quality of the data deteriorates and eventually they go read only (massive errors start showing up in their SMART data usually between 10TB - 15TB, including bad blocks).

Ah, they are 870 Evo's.

I have another one too. I suppose that will die soon. Is there any fix? I am pretty certain they had the latest firmware.

 

[SpellowHouse]

Have you tried them in another PC - sounds more software related

Yes. It's very strange the way they die. As Tetras says, they almost become read only. The odd thing is they don't really report errors. Windows has a fit, but doesn't know why.

 

[Tetras]

Ah, they are 870 Evo's.

I have another one too. I suppose that will die soon. Is there any fix? I am pretty certain they had the latest firmware.

It isn't clear.

Some of them are fixed by updating the firmware (the damage is still there though, it doesn't restore the data or free the reserved blocks being used), other drives still die anyway. It might depend how much damage has done, I'm not sure.

You can confirm it by looking at the SMART data in something like CrystalDisk and see what the ECC errors and bad blocks are like.

 

[SpellowHouse]

It isn't clear.

Some of them are fixed by updating the firmware (the damage is still there though, it doesn't restore the data or free the reserved blocks being used), other drives still die anyway. It might depend how much damage has done, I'm not sure.

You can confirm it by looking at the SMART data in something like CrystalDisk and see what the ECC errors and bad blocks are like.

It declined really rapidly. A few days ago it was showing errors. Yesterday it had pretty much brought Win 11 to a halt. I might add it was only a data drive, but Windows really hated it!

 

[Tetras]

It declined really rapidly. A few days ago it was showing errors. Yesterday it had pretty much brought Win 11 to a halt. I might add it was only a data drive, but Windows really hated it!

Can you take a pic from crystaldisk, curious to see what the smart data looks like and how many TB have been written.

 

[SpellowHouse]

Can you take a pic from crystaldisk, curious to see what the smart data looks like and how many TB have been written.

I can't, I am afraid, it was still under warranty so it has been returned. I am about to return the others too. It seems the 870's are a disaster.

The latest failed drive wasn't very old. Only six months. I think the amount written was 8.3TB. I'm really not sure about that though. Samsung Magician thought the drive was fine. Even tested it and it worked fine - except of course it really didn't because the test took ten times longer than it was supposed to.

 

[Eddie99]

I can't, I am afraid, it was still under warranty so it has been returned. I am about to return the others too. It seems the 870's are a disaster.

The latest failed drive wasn't very old. Only six months. I think the amount written was 8.3TB. I'm really not sure about that though. Samsung Magician thought the drive was fine. Even tested it and it worked fine - except of course it really didn't because the test took ten times longer than it was supposed to.

If Samsung magician didn't show any errors then the drive was probably fine, bad blocks would've shown up in the diagnostic scan. The 870 EVO had mass failures mostly in drives manufactured in 2021, and they revised the manufacturing process in November 2022.

 

[Murphy]

You'd typically expect a bad PSU to go *bad in a binary manner, as in it either goes bang and stops working/takes components with it or it doesn't.

Something that's always been the next go to for me after ruling out actual problems with a SSD/HDD is checking the RAM, all data that's read/written to permanent storage goes through your RAM so if that's creating errors those are going to get written to storage.

*Excluding not being able to deliver enough power and shutting the system down.

 

[SpellowHouse]

If Samsung magician didn't show any errors then the drive was probably fine, bad blocks would've shown up in the diagnostic scan. The 870 EVO had mass failures mostly in drives manufactured in 2021, and they revised the manufacturing process in November 2022.

Yes, I got the impression it wasn't storage on the drive, rather it was one of the secondary chips. Who knows what? Whatever it was, it introduced massive delays and in some cases reported the wrong information about what was on the drive.

 

[Eddie99]

Yes, I got the impression it wasn't storage on the drive, rather it was one of the secondary chips. Who knows what? Whatever it was, it introduced massive delays and in some cases reported the wrong information about what was on the drive.

It's very unlikely 2 SSDs failed in the same unusual manner without showing any errors in a diagnostic scan. I would be testing the memory and CPU for errors, and also make sure I didn't have a bad SATA cable or SATA port.

 

[Tetras]

It's very unlikely 2 SSDs failed in the same unusual manner without showing any errors in a diagnostic scan. I would be testing the memory and CPU for errors, and also make sure I didn't have a bad SATA cable or SATA port.

I'd have definitely been more comfortable RMAing them after confirming those error values, because the mass-faulty SSDs don't pass the diagnostic scans, as you said.

 

[SpellowHouse]

It's very unlikely 2 SSDs failed in the same unusual manner without showing any errors in a diagnostic scan. I would be testing the memory and CPU for errors, and also make sure I didn't have a bad SATA cable or SATA port.

The problem is that it's fixed by replacing the SSD and it goes away for six months or so. In the case of the latest failure, it started to go one day, then I changed the motherboard, but the failure continued until the drive was unusable. I suppose it could be the cable. Anyway, I guess it doesn't matter since I have returned all the drives anyway. I have never had a problem with any of the other drives in the PC.

 

[Tetras]

The problem is that it's fixed by replacing the SSD and it goes away for six months or so.

You said that it was 8.3 TB? Passing the diagnostic test isn't consistent, but that is definitely the kind of TB written that it starts being problematic and the read/write errors are noticed.