Rescue Data chips from a blown SSD in my G4 Cube

[EmilioCube]

When I resoldered my VRM and it blow my Cube Motherboard and CPU up, also the SSD IDE converter was blown up. It turned out that even the SSD is dead now and all my Data is lost, i just have a very old copy of it which was from like one year ago or so.
After opeing the SSD up (Intenso Top 128GB), the two flash chips stood out very clearly to me and do not look damaged, but one FET on the board is blown up. Is it possible to unsolder them with a proper heat gun and if I order a new SSD, replace them with the Data chips on the new one?

 

[philgxxd]

I would try to preheat the board. I once practiced with my rework station to get the sata connector off of a 2,5” HDD and was not able to do so.

 

[Needleroozer]

When I resoldered my VRM and it blow my Cube Motherboard and CPU up, also the SSD IDE converter was blown up. It turned out that even the SSD is dead now and all my Data is lost, i just have a very old copy of it which was from like one year ago or so.
After opeing the SSD up (Intenso Top 128GB), the two flash chips stood out very clearly to me and do not look damaged, but one FET on the board is blown up. Is it possible to unsolder them with a proper heat gun and if I order a new SSD, replace them with the Data chips on the new one? View attachment 2349648 View attachment 2349649 View attachment 2349650

It's certainly possible to reflow the NAND flash chips off with a hot-air gun, though a preheater would make it a lot easier like @philgxxd says. They should be easier to desolder than the SATA connector since their footprint is a lot smaller and they aren't connected as sturdily to the thermal mass of the power planes.

One thing to consider is whether any out-of-band data about what flash blocks are being used is internal to the controller rather than the NAND chips. This might mean that you wouldn't be able to mount the drive straightaway and get your data back in a recognizable format, but a lower-level data recovery tool *might* be able to piece a few blocks together (SSDs do not have to store blocks in sequential order; the controller may reorder them as it sees fit for wear leveling purposes).

I would probably start by trying to replace that burned SOT23-5 MOSFET/switching regulator and hope that it sacrificed itself to save your flash. If you can still read the marking code on top of it, you could try looking it up on a site like smdmark.com to figure out what it is exactly.

 

[EmilioCube]

In the last hour and a half I was trying to desolder the NAND chips with success. Soldering them back on wasn’t so hard either, just have to look out not to short any pins.

 

[EmilioCube]

One thing to consider is whether any out-of-band data about what flash blocks are being used is internal to the controller rather than the NAND chips. This might mean that you wouldn't be able to mount the drive straightaway and get your data back in a recognizable format, but a lower-level data recovery tool *might* be able to piece a few blocks together (SSDs do not have to store blocks in sequential order; the controller may reorder them as it sees fit for wear leveling purposes).

Thats exactly what I worried about. May it help to replace the Controller (the chip soldered in a 45 degree angle relative to the NAND on the first picture) as well?

And as seen in the picture almost the entire top of the MOSFET is dead and there is nothing to read.

 

[Needleroozer]

Thats exactly what I worried about. May it help to replace the Controller (the chip soldered in a 45 degree angle relative to the NAND on the first picture) as well?

There’s a chance that the controller has written this sort of metadata to extra blocks on the NAND that it doesn’t expose to the user (sort of like how a 120GB SSD may actually have 128GB of NAND, although that’s generally just overprovisioning so that it lasts longer with wear-leveling), so the drive may work with your NAND-only transplant.
If you are comfortable with the reballing, though, I’d swap the controller over as well just to be sure it’s all there.

And as seen in the picture almost the entire top of the MOSFET is dead and there is nothing to read.

Alright; worth a shot. I know those part numbers are hard to photograph well, so I figured it was worth mentioning it on the off-chance that it was still legible. (And I suppose because I think that hunting down and resoldering a SOT23-5 is simpler than triple-BGA rework - but that’s because I’m a wimp when it comes to BGA)

 

[EmilioCube]

Okay, I'll order a new SSD and hope they still use the same board revision because mine is already on year and a half old.

 

[LightBulbFun]

I would strongly advise not trying to swap the NANDs onto a new board, its highly like they are paired to the NAND controller also

I would just replace the failed FET if you really think thats all thats blown up

although from personal experiance, when you over-volt an SSD, the controller is what usually blows up first, and when that happens your best bet is to use a data recovery service, someone with an external NAND reader...

 

[EmilioCube]

I will try it myself and if that does not work then that's the way it is. Data recovery costs 690 Euro and that is a whole other Level compared to this 10 Euro SSD.