Intel Optane H10 dead, advice needed

[okton]

SSD Intel Optane H10 in my old HP laptop died recently. I have no strong urge to restore data in professional service and I have electronic knowledge so I want to do it myself.
Intel Optane H10 contains two drives, 512GB "normal" SSD with SMI controller and and 32GB "optane" SSD with Intel controller, connected to separate PCIe 2x buses.
Diagnosed so far:
1. 32GB part is functional, device dected, disk reachable, 512GB disk is not detected but PCIex device is visible under Linux/Windows
2. I checked board under microscope for broken components, checked restistance no shorts between power and grounds.
3. I compared my 512GB board with another 256GB H10 disk, all voltages are same around 1.0-1.2V depending on sections
4. Resistance on my board on controller power line is 800ohms, on 256GB variant is almost 4k, and this is only visible difference
5. I compared temperatures of 512GB and 256GB board with thermal camera, no differences, all chips heats up, no hot spots

Any advice what to check next?

[fzabkar]

Silicon Motion SSDs usually have ROM mode test points. If you short these two points with tweezers during power-on, and then release them thereafter, the SSD will enter "ROM mode" (aka "safe mode"). It should then identify itself with its generic factory name. This will tell you that the controller has basic sanity and that the problem is most probably due to a "firmware panic".

I suspect that the appropriate shorting points are the square and round pads at the bottom middle edge of the PCB. I'm not sure, though.

https://images.anandtech.com/doci/14249/IMGP4724.jpg

http://vlo.name:3000/hw/ssd/rommode/

[okton]

I checked the pins to make sure if shorting will no break the electronics. One is ground, on the other there were no pulses, so even if it could be debug connector it needs to be activated.
After shorting the pins, SMI controller switch to ROM mode as you suspected, it is recognized as "Silicon Motion, Inc. SM2263EN/SM2263XT SSD Controller (rev 01)" instead of "Intel Corporation Device 0975 (rev 03)" as it was discovered previously. Under linux /dev/nvme0 could be readable but returned about 1GB of data:
0x000000-0x004000 - block of almost only zeros, with string SM2262AA, there are couple of others non zero sequences like FF-FF-1F and few isolated non zero bytes
0x004000-0x008000 - copy of block 0x0
0x008000-0x00C000 - block of data with high entropy, like encrypted/compressed
0x00C000-0x010000 - copy of block 0x0
0x010000-0x01C000 - copy of block 0x8000
0x01C000-0x020000 - copy of block 0x0
0x020000-0x03C000 - copy of block 0x8000
0x03C000-0x040000 - copy of block 0x0
0x040000-0x05C000 - copy of block 0x8000
0x05C000-0x060000 - copy of block 0x0
On my another H10 module (size 256GB), the block layout looks same, block 0x0 also looks same, but block of data is totally different.
I had no time to check deeper the characteristics of this high entropy data or to find any visible pattern, but I see by the offsets the size is different in different location.
In hex editor, looking by naked eye, the data is same.
Questions:
1. does it means the controller is fine?
2. does it means the flash chips is accessable?
3. what is high entropy data, chip firmware or storage flash? how to get end-user disk data?
4. any further advices to restore controller firmware?

[fzabkar]

The controller is most probably OK. I don't know the answer to your other questions.

[okton]

I am thinking about "transplanting" SMI controller from working board to dead one, will it work?
Do you know if firmware is in SMI chip or Flash memory together with user data?

[osity]

okton,

Do you still have your INTEL optane Drive?
Was it recovered?

[okton]

okton,
Do you still have your INTEL optane Drive?
Was it recovered?

No change here, I abandonend this work for while, but I have still the dead and potential donor SSD.

[osity]

It's been a while. If you are still in need of the data we are able to recover these and we guarantee it or your pay nothing, so no attempt fees or evaluation fees etc.
Also, free shipping back.

A successful recovery outcome is assuming the item has not been physically damaged in some obvious way.
Feel free to contact me for info.
Thanks for your reply.

[okton]

It's been a while. If you are still in need of the data we are able to recover these and we guarantee it or your pay nothing, so no attempt fees or evaluation fees etc.

I wasn't asking for service but for a help how to troubleshoot it deeper or maybe even recover low-cost. It passed almost 2 years from my first post but I still keep this disk for sentimental or education purpose. Can you give some tips if you use your own developed solution or something from the market like PC3K.

[osity]

I do not have any low-cost method to share unfortunately. Good luck.

[okton]

I do not have any low-cost method to share unfortunately. Good luck.

I do not count that you will share me solution for free. I troubleshooted this for a while and checked pricelists sometime ago. Symptoms are suggesting firmware damage, those repairs were one of more expensive and for SMI chips for Intel support was poor. Intel Optane was rare in the past, then became obsolete. I wonder if repair tolls manufactureres might have solution, so I just asked if it was your own solution or something commercial.

[osity]

Many labs blame firmware. We do not believe that is the case with most of these. It is our own in house guaranteed solution. The solution is provided within a week.

[okton]

Just for curiosity, how much do you want for recovery?

[osity]

The charge for these if successful is typically $800.