Samsung 870 EVO - Beware, certain batches prone to failure!

[ACME]

Bought this 870 EVO 1TB in nov. 2022. SVT02B6Q. Production date: 2022.04.08. Should I be concerned about those errors in red?

Can RMA be done? or the values are too low? I didn't found any info about RMA conditions for a SSD.

 

[SPDIF]

Please post a screenshot with HD sentinel or CrystalDiskInfo, thanks! Some things i should know does not shop up here.

 

[ACME]

 

[SuperConker]

There is 1 bad Block on it already, i would hurry up and backup your data while you still can
(the 870 EVO will not let you backup data once it starts getting a lot of ECC/CRC errors).

 

[ACME]

Thanks. I already did a backup. Before this 870 EVO I had an Seagate SSHD who suddenly died so I learned my lesson.

I haven't checked the SSD for errors since I bought it in 2022. But a few days ago I wanted to open a file and it didn't work, what the hell? I restarted the PC and then I was able to open the file. Then I checked the SSD and saw those errors.

I also have two 860 EVO from 2018 and they are great, no errors whatsoever.

I don't know what changes they made on the 870 EVO, but it's regression.

 

[SPDIF]

It's just the beginning, but yeah sorry prognosis is not good... It begin to make reallocated sector's, bad news. It will become worse and worse until you lose data, Backup and replace ASAP.
RMA back to Samsung. Good luck and don't wait with it, it's their fault, you have the right for a new one.

The 860EVO are the best, they don't make quality like this anymore, now it's all QLC and TLC. I still have an 860PRO and like yours it's made purely of MLC NAND.
The older SSD are much better of quality, even the newest NVMe drives are now all made with TLC.

 

[chrcoluk]

Thanks for that! At least my old 830 drive is still going strong, I suspect the reallocated sectors are just they hit some write limit... no BBs... about 9 years 8 months Powered On and 118TB written BTW...
View attachment 328693
The 870 is somewhat younger...
View attachment 328694

Wow I never seen a SSD with such high write cycles.

My 128 gig 830s have cycles close to the mid 100s, which I considered really high, the 830s are of course from an era when write amplification was higher, but at an average cycle rate of over 2700, I too wonder if those reallocated sectors are related to cycle exhaustion, but it is also possible that they are premature failure before the spec'd 3k cycle limit.

Here something of interest to you. Your host writes on your 870 EVO are 7273 GB, on my 860 EVO they are 7231 extremely close. However your drive is double the capacity, yet has the same erase cycles, mine has 5 erase cycles. The write amplification on my 860 EVO is extremely good right now close to almost zero amplification.

Thanks. I already did a backup. Before this 870 EVO I had an Seagate SSHD who suddenly died so I learned my lesson.

I haven't checked the SSD for errors since I bought it in 2022. But a few days ago I wanted to open a file and it didn't work, what the hell? I restarted the PC and then I was able to open the file. Then I checked the SSD and saw those errors.

I also have two 860 EVO from 2018 and they are great, no errors whatsoever.

I don't know what changes they made on the 870 EVO, but it's regression.

Yep, backup and then RMA it. Thats on its way out.

 

[ACME]

I sent them an email, let's see what they say.

But I found a message on reddit that doesn't sound too encouraging...

"I went through the same email process, and asked if I could just use the warranty to replace the drive instead of sending mine in for repairs, they said no, it has to go through repairs first. Just a heads up if anyone else is concerned about that."

 

[Jon555]

Wow I never seen a SSD with such high write cycles.

My 128 gig 830s have cycles close to the mid 100s, which I considered really high, the 830s are of course from an era when write amplification was higher, but at an average cycle rate of over 2700, I too wonder if those reallocated sectors are related to cycle exhaustion, but it is also possible that they are premature failure before the spec'd 3k cycle limit.

Here something of interest to you. Your host writes on your 870 EVO are 7273 GB, on my 860 EVO they are 7231 extremely close. However your drive is double the capacity, yet has the same erase cycles, mine has 5 erase cycles. The write amplification on my 860 EVO is extremely good right now close to almost zero amplification.

View attachment 328855


Yep, backup and then RMA it. Thats on its way out.

Note the 870 is very nearly full (42.7MB free as I type - should be a lot more though but I dump some copies of new things there to keep them safe until the next NAS backup and haven't deleted any older ones for a while. Also contains a number of large disk images (7) and a pagefile (as I wanted to save the 830 some exercise by having most of the pagefile on a younger drive). Not sure how relevant that is, but FYI.

 

[ACME]

I always look to the SMART data table to know if one's disk is still okay, that count's for any program you use. No need to pay anything.

Well, Samsung requested a Diagnostic Scan which looks Good, they didn't care about SMART. By the way, did you send your ssd to Hanaro Europe?

 

[Zeromis]

Figured I might as well give a status update. It's been over a year and a half since I had 2 2TB 870 Samsung Evo start to fail in the same way as everyone else, I remember posting about it on an earlier page of this thread. Luckily I caught it early on so I didn't end up losing much in the way of data. At the time I only had one receipt since the other was given to me by work since it was no longer being used, so I could only RMA one. The new one I got from the RMA has been going fine, no errors on it at all.

The old one which I couldn't RMA I did end up keep using(for non-important data like steam game installs). At the time I installed the new firmware and did a full wipe of the drive getting rid of anything that got corrupted, while also "revealing" the bad blocks. By the end of that the drive ended up with around 33 bad blocks, and around 1500 uncorrectable errors. To this date those numbers haven't gone up, and I've not had any files on the drive show issues or errors since. So I do believe the firmware did indeed fix the problem in the end.

I'd definitely recommend people RMA the drive if they can, you shouldn't put up with a drive that's lost blocks due to faults beyond your control. However in the unfortunate situation you can't, if the initial damage wasn't too bad, the drive seemingly can be somewhat recovered. Though I'm not sure I'd use it for important data ever again, hah.

 

[chrcoluk]

Note the 870 is very nearly full (42.7MB free as I type - should be a lot more though but I dump some copies of new things there to keep them safe until the next NAS backup and haven't deleted any older ones for a while. Also contains a number of large disk images (7) and a pagefile (as I wanted to save the 830 some exercise by having most of the pagefile on a younger drive). Not sure how relevant that is, but FYI.

Yeah that will be having an impact for sure.

My 860 has 223gig free, and is 10% unformatted. I dont use my 830s anymore (128 gig), they were momentarily going to be put in to my proxmox setup as its boot ZFS mirror pool, that would have likely got them into the 1000s easily, but I ended up needing more space so replaced them with a couple of larger 860s.

 

[SPDIF]

Yeah that will be having an impact for sure.

My 860 has 223gig free, and is 10% unformatted. I dont use my 830s anymore (128 gig), they were momentarily going to be put in to my proxmox setup as its boot ZFS mirror pool, that would have likely got them into the 1000s easily, but I ended up needing more space so replaced them with a couple of larger 860s.

I do that also always with any SSD i have, leave 10 percent free unformatted and as RAW partition. Best to protect your SSD not to write it full. This way you can write it full without thinking or checking it every time, there is always 10 percent free space for the controller to do it's job good. All SSD made before the 870EVO where of superior quality, they don't make them like that anymore.

 

[Jon555]

Yeah that will be having an impact for sure.

My 860 has 223gig free, and is 10% unformatted. I dont use my 830s anymore (128 gig), they were momentarily going to be put in to my proxmox setup as its boot ZFS mirror pool, that would have likely got them into the 1000s easily, but I ended up needing more space so replaced them with a couple of larger 860s.

The 830 has 23.8GB unused but I used all the 870 space as I hadn't originally planned to use it for anything intensive, just working data files for some programs and a fast place to copy data to temporarily (the pagefile idea came along later). I may shrink it a little after I've made some space. I have a new 12TB drive in the post - one good side-effect of all this SMART screen grabbing is I found my 5TB WD Red had a weak (aka pending) sector, which is now two weak sectors (it's fully backed up, as is the backup), so it's heading into retirement (I'll probably map them out so it could be used in emergencies).

 

[TPCEA]

For those who don't yet know, this issue was reported in the first half of last year on several places, including Jayz2Cents (check his video for more info). Anyone who gets certain 970, 980 and certain other SSDs from a specific batch and time are likely to have this issue because the firmware has a fatal error in it that corrupts the drive, eventually leading it to lock itself into read only mode permanently. If you have bought or own one, immediately use Samsung Magician to update the firmware. This will correct the programming error, but not the damage done.

 

[SPDIF]

Not with the 870, that was due to bad nand used in the time of the pandemic, and no one could find chips on the market.

There is an thread about the firmware and the problems off the 890 and 990Pro series here;

FirmWare Updates - for WD Black and Samsung

The list in the first post is not complete. The new Samsung 990PRO 4TB uses 0B2QJXG7 as firmware version. The D changed in an G as part of the new NAND process used it the 4TB versions. For older drive versions of the 990PRO it was version 0B2QJXD7, and the fixed version firmware is 3B2QJXD7...

www.techpowerup.com

And not all of them are bad, there are also many fake ones around now.

 

[TPCEA]

Not with the 870, that was due to bad nand used in the time of the pandemic, and no one could find chips on the market.

There is an thread about the firmware and the problems off the 890 and 990Pro series here;

FirmWare Updates - for WD Black and Samsung

The list in the first post is not complete. The new Samsung 990PRO 4TB uses 0B2QJXG7 as firmware version. The D changed in an G as part of the new NAND process used it the 4TB versions. For older drive versions of the 990PRO it was version 0B2QJXD7, and the fixed version firmware is 3B2QJXD7...

www.techpowerup.com

And not all of them are bad, there are also many fake ones around now.

Thanks for the clarification!

 

[E.S]

Samsung says that the cause of the 870evo problem is the firmware.
Someone posted about it on this forum.

No one said that bad nand was the cause.

 

[SPDIF]

Samsung says that the cause of the 870evo problem is the firmware.
Someone posted about it on this forum.

No one said that bad nand was the cause.

If it was the firmware, then now everyone would have the newest firmware installed... But you still can buy with older firmware and no update available.
Specially 4TB versions still come with SVT01 firmware and there are NO updates, production date 2023.

And all the bad ones are from the time of the pandemic, check the whole topic. Also crucial and WD had bad driver in that time. It's not because SOMEONE say this it's true.
I did sell then and now hundreds of them in my shop, how about you? I have the info from REAL experience here in my shop.

Here as proof, a new 870EVO with the old firmware, made 10/2023

I even have 4TB versions with SVT02 and SVT03 firmware, all made in 2023. No update's available for them!!!

 

[E.S]

If it was the firmware, then now everyone would have the newest firmware installed... But you still can buy with older firmware and no update available.
Specially 4TB versions still come with SVT01 firmware and there are NO updates, production date 2023.

And all the bad ones are from the time of the pandemic, check the whole topic. Also crucial and WD had bad driver in that time. It's not because SOMEONE say this it's true.
I did sell then and now hundreds of them in my shop, how about you? I have the info from REAL experience here in my shop.

Here as proof, a new 870EVO with the old firmware, made 10/2023

View attachment 329660

I even have 4TB versions with SVT02 and SVT03 firmware, all made in 2023. No update's available for them!!!

Is that SSD really manufactured in 2023?

For SSDs manufactured in 2023, the 8th digit of the serial number is "W".
Your SSD is "N".


・・・According to research, "N" is manufactured in 2020.
The manufacturing date of that SSD is "10/2020".

 

[Jon555]

Is that SSD really manufactured in 2023?

For SSDs manufactured in 2023, the 8th digit of the serial number is "W".
Your SSD is "N".


・・・According to research, "N" is manufactured in 2020.
The manufacturing date of that SSD is "10/2020".

Interesting - Looking into this it seems (!! I'm not certain) that generally for Samsung:
If the serial number has 14 or 15 characters, the 8th character is the year of manufacture and the 9th character the month.
If the serial number has 11 characters, the 4th character is the year of manufacture and the 5th character the month.
The first character is the product type.
The Second and Third characters are the code of the factory where it was made
Characters from Ten onwards are the actual serial number
The character after the year is the month (1=Jan...9=Sept...A=Oct...C=Dec) and it seems the years are:
(used twice)R=2001
(used twice)T=2002
(used twice)W=2003
(soon to be used twice?)X=2004
Y=2005
A or L=2006
P=2007
Q=2008
S=2009
Z=2010
B=2011
C=2012
D=2013
F=2014
G=2015
H=2016
J=2017
K=2018
M=2019
N=2020
R=2021
T=2022
W=2023
(I assume)X=2024
So on that basis my 870 was manufactured in December 2021 and I bought it in June 2022, which doesn't seem unreasonable.
Also the 830 was made in April 2012 which seems reasonable as I bought it in June 2012.

 

[P4-630]

If it was the firmware, then now everyone would have the newest firmware installed... But you still can buy with older firmware and no update available.
Specially 4TB versions still come with SVT01 firmware and there are NO updates, production date 2023.

And all the bad ones are from the time of the pandemic, check the whole topic. Also crucial and WD had bad driver in that time. It's not because SOMEONE say this it's true.
I did sell then and now hundreds of them in my shop, how about you? I have the info from REAL experience here in my shop.

Here as proof, a new 870EVO with the old firmware, made 10/2023

View attachment 329660

I even have 4TB versions with SVT02 and SVT03 firmware, all made in 2023. No update's available for them!!!

N=2020

The Samsung 870 EVO is a solid-state drive in the 2.5" form factor, launched on January 20th, 2021.
(google)

So how is that possible then..

 

[Jon555]

The Samsung 870 EVO is a solid-state drive in the 2.5" form factor, launched on January 20th, 2021.
(google)

So how is that possible then..

Excellent point!!!
Perhaps they have changed their numbering scheme?
Looking through previous pages I see a drive bought in April 2021 with R1 for Jan 2021
Another with R7 also for 2021
One with W6 = June 2023 bought July 2023
It seems to stand up except for that case...

 

[ACME]

Launched on January 20th, 2021 but factory produced in 2020?

In my case T=2022. On the box is: Production date 2022 04 08

 

[Jon555]

Launched on January 20th, 2021 but factory produced in 2020?

In my case T=2022. On the box is: Production date 2022 04 08

Another good point: NB = November 2020 which would indeed seem reasonable for a Jan 2021 launch...