I killed my USB flash.

[vlad777]

Hi,

Anybody into reviving USB flash drives?

My problem is that windows doesn't recognize it in any way (device descriptor request failed).

So my question is about the controller in this thing.
Is it like an empty micro-controller, where you don't expect it to do anything?
Or does it have ROM or hardware that would establish basic USB communication?

Here is some info:

Transcend
4GB
JF V85
(JetFlash)

-------
Chip:

Alcor AU6984 (controler)
Samsung K9LBG08U0M (flash)

http://www.alldatasheet.com/datasheet-pdf/pdf/219004/ETC2/AU6984.html

-------
Recovery software:

JetFlash Online Recovery
ChipGenious
AlcorMP


Many thanks.

[Aodhan145]

buy an identical one and replace the controller chip? (This is an idea i dont know what im talking about really)

[vlad777]

I do get that desperate, just to show it I'm smarter than it

(But other than that , it makes no sense in any way , not economically justified, why solder if you can just program it)

[rs20]

buy an identical one and replace the controller chip? (This is an idea i dont know what im talking about really)

+1 This is by far the most interesting option, and the most likely to succeed if indeed the controller chip (and only the controller chip) is fried.

[vlad777]

But that is my question.
Is it fried or just bricked?
(Someone may know if they are into this.)

[kc8apf]

Could be either. The controllers I've evaluated for work all had flash for firmware. They are just microcontrollers with specialized peripherals.

Sent from my Nexus 5X using Tapatalk

[edy]

I thought I had a dead USB stick and used a utility to reprogram it, and it came to life. I had to try different versions and most were in Chinese so I had trouble figuring it out, but they were from the website of the original chip controller which I found by using another program that reads ID codes.

By the way, the drive died while I was running a live Linux distribution off of it that I had booted from USB. Turns out that also the reported memory size was faked.

[Brumby]

Turns out that also the reported memory size was faked.

This is a danger.  I never trust anything out of the box these days, so I run H2testw.  It's slow, but thorough.

[Voodoo 6]

This is a tough nut to crack, Everyone on the planet knows that electros from country x have a very high failure rate, yet everyone keeps buying from that country. Basically sowing the seeds of our own destruction globally. You want crime to go down impose super strict regulations for it. Same for electronics, they drive our world, bad components put everyone in harms way. Drop a plane from the sky due to faulty component is that not proof positive? Cant sue them, they are immune and they know that. Its not country x that is the problem, its the people who are buying from them. They have elevate the fakes to an art. You want to have a successful industrial base go back to the German model pre 1990. The concept of giving a hoot about what you are making is a lost art.

 

[R005T3r]

buy an identical one and replace the controller chip? (This is an idea i dont know what im talking about really)

Could be a good idea, but you need proper equipment to replace it...

Actually, I would try it on a Linux environment to see if it changes anything. I don't think the controller is faulty: electronics these devices rarely fails... Maybe it's only windows that's not liking the usb drive (won't be the first time that happened to me, after I've lost 1 day I've decided to buy a proper hard drive for backups and always do that)

[R005T3r]

I thought I had a dead USB stick and used a utility to reprogram it, and it came to life. I had to try different versions and most were in Chinese so I had trouble figuring it out, but they were from the website of the original chip controller which I found by using another program that reads ID codes.

By the way, the drive died while I was running a live Linux distribution off of it that I had booted from USB. Turns out that also the reported memory size was faked.

Yes, especially USB sticks and sd cards are faked.This problem appear to be common on ebay $5 cards. Actually I was likely to get scammed by one of these, however I asked myself "how it's possible it's about $5 instead of $30" and after buying I've made a search founding that was a scam...

[plazma]

I have fixed many USB flash sticks. Check for cracks in the USB connector pads. If the controller chip is fried you can swap chips with an identical stick.

[rob77]

also you might try to read the flash chip only (with a micro with enough IO pins) and hope the disc sector allocation policy of that dead controller was linear.

[janoc]

also you might try to read the flash chip only (with a micro with enough IO pins) and hope the disc sector allocation policy of that dead controller was linear.

That would be pretty rare. Most of the flash controllers do (some) wear leveling and transparent bad block reallocation, otherwise the frequently written parts of the flash (e.g. where FAT resides) would die really quickly rendering the stick useless after only short time. 

I think Bunnie had a good article on the SD card controllers explaining how this worked.

[rob77]

also you might try to read the flash chip only (with a micro with enough IO pins) and hope the disc sector allocation policy of that dead controller was linear.

That would be pretty rare. Most of the flash controllers do (some) wear leveling and transparent bad block reallocation, otherwise the frequently written parts of the flash (e.g. where FAT resides) would die really quickly rendering the stick useless after only short time. 

I think Bunnie had a good article on the SD card controllers explaining how this worked.

i know how it works. but in this case there is separate controller and separate flash chip.
so after first write ,or few writes the allocation might still be linear - that's why i wrote "and hope the allocation was linear"

[vlad777]

I will try under Linux, good idea , i did try under XP.
So if this controller is firmware based , and the boot-loader is gone how am I suppose to program it?

Apart from D+ and D-  there is this pin : "20 FMWP In Reserved for firmware " . (Write Protect?)
Any ideas?

Edit:

Could this firmware actually be inside the flash chip? I know that some hard drives keep their fw on the platters.

[vlad777]

I think Bunnie had a good article on the SD card controllers explaining how this worked.

Is this  it ? https://www.bunniestudios.com/blog/?p=3554
Great find!
So I can use this stick as a micro with loads of storage, cool.
That is, if I can reverse engineer it.

[janoc]

I think Bunnie had a good article on the SD card controllers explaining how this worked.

Is this  it ? https://www.bunniestudios.com/blog/?p=3554
Great find!
So I can use this stick as a micro with loads of storage, cool.
That is, if I can reverse engineer it.

I think I have seen yet another article on this but as Murphy would have it, I can't find it right now. It might have been related to his work analyzing fake SD cards and flash sticks.

EDIT:
Here it is - it finally wasn't Bunnie:
http://joshuawise.com/projects/ndfslave

Should give you an idea of what is involved ...

[janoc]

i know how it works. but in this case there is separate controller and separate flash chip.
so after first write ,or few writes the allocation might still be linear - that's why i wrote "and hope the allocation was linear"

Check this article on reverse engineering an SD card data layout in order to recover them. Should give you an idea why your hopes are pretty futile:

http://joshuawise.com/projects/ndfslave

[rob77]

i know how it works. but in this case there is separate controller and separate flash chip.
so after first write ,or few writes the allocation might still be linear - that's why i wrote "and hope the allocation was linear"

Check this article on reverse engineering an SD card data layout in order to recover them. Should give you an idea why your hopes are pretty futile:

http://joshuawise.com/projects/ndfslave

once again   in this case there is a separate flash chip , here is the datasheet

http://downloads.qi-hardware.com/hardware/datasheets/qi_lb60/U9~K9GAG08U0M~~2GB-FINAL_NAND-NOT_ON_SOME_PROTOTYPES~~.pdf

it's describing also the initial invalid blocks etc.... so the hope is not so futile as you think

[janoc]

once again   in this case there is a separate flash chip , here is the datasheet

Did you actually check what is in that SD card being reverse engineered? Surprise - a controller and a separate flash chip.



So I am not quite sure what are you trying to achieve with this argument.

[rob77]

So I am not quite sure what are you trying to achieve with this argument.

probably to suggest yet another possibility how to recover data ?
the datasheet doesn't mention any kind of sorcery needed.... that flash chip has a well defined interface with linear address space (from interface perspective). so reading it out with a micro followed by substracting the invalid blocks (identified as per datasheet) would give you a very good basis for data reconstruction.  once you have the image read out, you can try to analyse it with some software (GetDataback for FAT) or analyse it manually - look for FAT table and start to manually reconstruct your important files.

[Aodhan145]

So I am not quite sure what are you trying to achieve with this argument.

probably to suggest yet another possibility how to recover data ?
the datasheet doesn't mention any kind of sorcery needed.... that flash chip has a well defined interface with linear address space (from interface perspective). so reading it out with a micro followed by substracting the invalid blocks (identified as per datasheet) would give you a very good basis for data reconstruction.  once you have the image read out, you can try to analyse it with some software (GetDataback for FAT) or analyse it manually - look for FAT table and start to manually reconstruct your important files.

Stupid ass method though. He hasn't even tried the other ones.

[VK5RC]

Sorry if I have missed this (dumb) suggestion, have you tried 1  another computer and  2 tried gently cleaning the contacts?

[janoc]

the datasheet doesn't mention any kind of sorcery needed.... that flash chip has a well defined interface with linear address space (from interface perspective). so reading it out with a micro followed by substracting the invalid blocks (identified as per datasheet) would give you a very good basis for data reconstruction.  once you have the image read out, you can try to analyse it with some software (GetDataback for FAT) or analyse it manually - look for FAT table and start to manually reconstruct your important files.

Yes, but the problem is you have the data scrambling on top of that, as well as dead block remapping and wear leveling. So by definition you are not going to have a continuous block of sensible data you can "just read out" unless you figure out this scheme. You will read only meaningless garbage. That is what that article I have linked to actually documents, just nobody seems to be bothered to read it