HDD GURU FORUMS

Okay, I know I am asking for the impossible. The standard policy is to destroy the client’s device, or otherwise render the data unreadable before disposing of it. I tried asking for failing SATA drives before and came up empty. So I guess I don’t have high hopes for this. But I am asking anyway. Maybe someone will have a client that is okay with giving the device to someone else for the purpose of research and development. I can say that any devices I acquire will not leave my possession in any sort of condition where the data could be read. I am willing to sign some sort of agreement for this if needed.

I am looking for NVMe devices that are failing, but still read. The first type of failure would be bad sectors, where some reads fail. The second type of failure is a device that goes offline when reading, and requires some sort of major reset or power cycle to get it going again. I am not looking for devices that are dead or require special tools using vendor specific commands or methods to read. The device must read some user data, and not have any major firmware issues such as a translator issue.

I would be willing to pay a small amount for such devices, to cover shipping and your time and effort to ship it, but I am not willing to pay a large amount. I cannot afford to pay a premium price for something that I would not even be sure would help with my development, not knowing the exact condition until I work with it. Discussion of what any device would be worth to me would be on a case by case basis.

And if anyone is unclear about why I am looking for such devices, I am looking to implement a NVMe passthrough mode in HDDSuperClone, and potentially (hopefully) a direct NVMe mode. And the only way I can truly do that is to have some devices to test with. Without any failing devices, I will be limited on what I can accomplish.

Buy at a flea market-faulty SSDs
For example:
https://www.avito.ru/balashiha/tovary_d ... ion=621540
https://www.avito.ru/moskva/tovary_dlya ... ion=621540
https://www.ebay.com/itm/Job-lot-6-SSD- ... SwxMBgWKOg

Have you checked eBay? I had some luck buying 'lots' of USB flash drives with condition 'for parts/not working' checked for target practice with FE. I basically got 3 categories that was. 1. Utterly dead. 2. Controller 'dead', those are good for FE. 3. Utterly unstable but some times actually readable using USB Stabilizer or even using YKUSH power cycles.

The only source I've found is ebay and they're not going cheap. Very little supply of them yet although you may find some older apple drives.

You will have a hard time with this.

- NVMe with just bad sectors (rare)
- NVMe with barely functional state (rarely lasts very long before being completely unresponsive)
- NVMe in an unresponsive BSY state (common, but requires a LDR and the ability to virtually rebuild the translator to recover)

lcoughey wrote:You will have a hard time with this.

- NVMe with just bad sectors (rare)
- NVMe with barely functional state (rarely lasts very long before being completely unresponsive)
- NVMe in an unresponsive BSY state (common, but requires a LDR and the ability to virtually rebuild the translator to recover)

If the most common type requires a LDR and translator rebuild, then that might mean it isn't worth the effort for the direct mode, as I will never be able to do that. I have looked at the low level stuff in the NVMe documentation, and it makes my head hurt. I have to weigh the hours of work to figure that out against the benefit. If it really isn't going to be much if any better than using a USB adapter with a USB Stabilizer (or HDDSuperClone Direct USB mode), then there is not much purpose for it.

maximus wrote: I have looked at the low level stuff in the NVMe documentation, and it makes my head hurt. I have to weigh the hours of work to figure that out against the benefit.

..... and that's exactly the point I reached. I looked at the price of PC3K portable vs the work required vs just a straight DE or HDDSC clone and none made sense for a product with a current very low demand and lower recovery percent (i've not had one sent in yet). I don't know how others have justified it, I certainly cant.

I'm waiting for some bits from China (probably stuck in the mud somewhere) to continue my R&D though.

- NVMe with just bad sectors (rare)

If bad sectors are rare, I believe it can be simulated with write uncorrectable. That is not always the same as real bad sectors, but it can be enough to verify that the software can properly detect bad reads. So I guess I can work around not having a device with actual bad sectors.

I am currently playing around with nvme-cli, which is something I will need to examine the source code to work on making a passthrough mode. I have not yet achieved a good level of motivation, so there is not much progress yet at all, just playing so far. Sometimes I need to get a good picture in my head before I start making any progress on something.