[SOLVED] SSD disappears after standby/shutdown/crash

Dec 10, 2021
8
0
10
SSD disappears after standby/shutdown/crash
I have 3 SSD. A Samsung EVO 480, EVO 860 and EVO 870, All connected per SATA cable. This problem only affects the 860 and 870 which are V-Nand.
The 870 was brand new and instantly had the same problem as the 860 already was having.


The problem:
Sometimes when I start my PC (this can be from standby, full shutdown or crash) those 2 drives (or only 1 of them) completely disappear, so not showing up in bios or anywhere.
I can bring them back while windows is booted by just unplugging both power and sata cable and reconnecting. (power first)


What i did:
Tried to mess with windows power settings
Sent one of the drives to samsung support. They reinstalled the firmware and sent it back but problem persists unchanged.
Changed cables and slots.
Tried to enable Rapid mode but OS doesn't boot anymore that way
Updated chipset drivers and some others (SSD firmware already up to date)
Updated Bios
Run some tests with Samsung Magician software, everything seems good as long as they are detected
Tested the drive in external housing (worked fine)


My speculation:
Since it also happens after a crash, im assuming it has nothing to do with windows
There has to be something different about the EVO 840, since it never had that problem. This might be key

My system:
Win 10
MSI x470 gaming pro
Ryzen 7 2700X
16GB Ram
Radeon RX Vega 56
be quiet, Straight Power 11 850W
Fireface UC
Some PCI usb controller
2 internal HDD, 1 external
 
Solution
You could pause the POST by entering BIOS setup and then exit after a minute or so. This would help to confirm the cause, but it obviously isn't a solution.

I would check the SMART report and look for an unsafe shutdown count (POR Recovery Count, attribute 0xEB). It could be that the SSD is rebuilding some firmware components during its POST (power on reset recovery), eg the Flash Translation Layer.

I have an 860 Evo 1TB attached to an old Core 2 Duo motherboard. It sometimes doesn't detect after a normal shutdown, and for a while I believed it was responsible for intermittent system crashes. It seems OK now, though.
You could pause the POST by entering BIOS setup and then exit after a minute or so. This would help to confirm the cause, but it obviously isn't a solution.

I would check the SMART report and look for an unsafe shutdown count (POR Recovery Count, attribute 0xEB). It could be that the SSD is rebuilding some firmware components during its POST (power on reset recovery), eg the Flash Translation Layer.

I have an 860 Evo 1TB attached to an old Core 2 Duo motherboard. It sometimes doesn't detect after a normal shutdown, and for a while I believed it was responsible for intermittent system crashes. It seems OK now, though.
 
Last edited:
Solution
This sounds very plausible but how would i go about fixing this issue?
There are various power management features in the BIOS that are specific to drives.
They are hard to find/buried some times.

If you wanted to get super crazy you could connect another power supply to the drives directly independently of the computer power supply. If you wanted to make sure and couldn’t find the bios stuffs.
 
You could pause the POST by entering BIOS setup and then exit after a minute or so. This would help to confirm the cause, but it obviously isn't a solution.

I would check the SMART report and look for an unsafe shutdown count (POR Recovery Count, attribute 0xEB). It could be that the SSD is rebuilding some firmware components during its POST (power on reset recovery), eg the Flash Translation Layer.

Seems to be a different problem then. Both SSD appear in windows. I shutdown PC, boot into bios and they already are both gone. I wait 1 min and save+reboot, they still are gone in Windows.
Any other ideas?

SMART looks good to me
View: https://imgur.com/a/eHQfhhs



If you wanted to get super crazy you could connect another power supply to the drives directly independently of the computer power supply. If you wanted to make sure and couldn’t find the bios stuffs.
Thats a great idea but not a great solution :/
Any hint what such options in bios might be called? I have been playing around with it for a while now but didnt find anything that fixes the problem
 
So i was able to narrow it down a little further.
All i need to do is unplug the power of the SSD and replug it (while in OS) and it comes right back to life.
This also works while in BIOS but it doesnt show up instantly. But if i boot into OS from there the drives are detected.
And the problem definitely doesn't happen at restart, when power is never taken away from the drives.

The mainboard should be able to control when the drives get powered. Is there any option to manually control that or set a different value with delayed startup or something?
 
SATA power pin #3 is either a POWER Disable pin or DevSleep.

The drive is in sleep mode when the voltage on this pin is +3.3V.

The drive is powered on when the voltage is 0V.

Can you measure this pin with a multimeter?

I expect that there will be 0V on this pin, but I'm not completely certain.
 
SATA power pin #3 is either a POWER Disable pin or DevSleep.

The drive is in sleep mode when the voltage on this pin is +3.3V.

The drive is powered on when the voltage is 0V.

Can you measure this pin with a multimeter?

I expect that there will be 0V on this pin, but I'm not completely certain.

There are 3.3V coming from the SATA power connectors first wire while in OS.
That would include the first 3 pins i assume, kinda didnt want to short anything at the tiny pins so correct me if thats not the case but i dont see why it wouldnt.

https://ibb.co/rtfWrBh

So the SSDs are put into sleep mode for some reason? 0.o
(and ig only my newer model SSDs have that feature)
 
Last edited:
It's only the third wire that is important. Try putting tape over the 3.3V pins.
You mean the 3rd pin? I can confirm per diagram, if the first wire has 3.3V then the first 3 pins also have it.
https://ibb.co/6B3fBxQ
I cant rly measure the individual pin without breaking the plug but it should definitely have 3.3V

The 3rd wire has 5V

Edit: Unfortunately blocking the 3.3V with electrical tape didnt change anything about the issue. Im not sure how to explain this signal but it seems like its not doing anything.
 
Last edited: