USB drive toasted by Dell workstation - can it be recovered?

[wibit]

Hello,

I am a new member of HDD Guru Forums, but I have enjoyed the articles and posts on and off for awhile. I appreciate the quality of this forum and its membership. Storage media and data recovery have been an interest of mine for years, during which time I've had a small number of DIY DR successes. No failures yet.

I have a PNY 128GB USB stick that failed immediately when it was connected to a high end, expensive Dell workstation running Windows 7. I connected a sacrificial USB stick to the same port and it developed the same problem. After booting a Linux O/S, I discovered over-current conditions in the system boot log for that USB port. I have no record of the drive's contents except that it contains 80+ GB of work-related files. Some of the data may be proprietary company information. Again, no clear record of that.

I do not see a fuse on the circuit board, nor do I see any physical defects under magnification. The controller is an Alcor AU6998SN with an integrated fuse and oscillator. I believe it may be damaged.

There are two 64GB ball grid array (BGA) NAND chips. I assume these are 8-bit. I'm not able to find information about these chips since they do not bear comprehensive markings. There are no markings underneath the paper sticker. I can get the chips off the board and onto a pogo pin fixture, but I assume that the chip pinouts are not standardized across the industry.

It's likely that successful DR will require chip-off procedures. I am familiar with the work of Bjoern Kerler, Joshua Wise and others. Countering that, I understand the excellent points made by Dmitry in his DIY post here http://forum.hddguru.com/viewtopic.php?f=16&t=12671 This should be a sticky. I realize there are complicating factors that differ with each job and the work should be left to DR professionals, but I do not have budget or clearance to farm the work out at this time.

A few questions for anyone who can spare the time to answer:

Does my analysis seem to be reasonable up to this point?
I have another controller chip (identical) but I assume that a controller transplant may be useless or even unwise, correct?
Is it possible to find data sheets, or even just pinout information, for the NAND chips used in this device?
This seems like a straightforward recovery, since the chips are intact. What might a commercial data recovery center charge for this?
Where would I find a competent DR provider in the United States, Canada, or Europe?
Are they better (training, experience, results) in some countries than others?
Why do DR agencies think it is their right to retain customer data? This is something I do not see enough discussion on.

Thank you for the interesting (sometimes fascinating) reading provided by this professional forum.

Kind regards,
wibit

Attachments

[HaQue]

you can try to image the drive first. Using a known good windows system with at least 130GB free , insert usb, open DMDE and see if the drive is detected there. ignore any windows messages, and in fact "this drive needs formatting" is welcome at this stage. Don't close the dialog box if it appears.

if drive appears, select drive (not any partitions, but the drive) and try and create an image of it. If it images, then use any recovery software like R-Studio or GetDataBack.

If this does not produce results, then Chip off is required. Without knowing chip it is going to be hard to give any advice.

You are welcome to send to me in AU, PM me for details.

You can try and troubleshoot electronics, but components are not marked well, and may be quite difficult.

transplanting controller is not an option. each one contains firmware configs for the specific nand chips on each device, "set up" individually during manufacture.

[fzabkar]

I would confirm the voltages present at the Dell USB port. This will tell you what killed the sticks. BTW, have you tried the "sacrificial" stick in another machine?

There is a 3.3 ohm resistor (R1) in series with the +5V input. The voltage drop across R1 will tell you the current drawn by the stick. However, I would prefer to measure the resistance between R1 and ground.

I would also measure the resistance of each capacitor (C3,C4,C5,C7,C8,C11). These capacitors would be filtering the +5V input and the +3.3V supply for the NAND chips. The 3.3V LDO regulator would be embedded within the controller. A low resistance across the 3.3V supply would cast doubt on the health of the NAND chips.

As for the flash pinout, ISTM that a "chip-off" recovery may be unnecessary, even if the controller is dead. I say this because AFAICT the PCB appears to be a dual purpose type that can accommodate NAND chips with either TSOP-48 or BGA pinouts. You could confirm the TSOP-48 pinout by starting with the known (?) pinout of the controller and then testing for continuity between the controller and the TSOP pads. You would then use a fine saw to cut through the PCB between the controller and NAND pair, leaving just the TSOP stub.

[arvika]

Controller transplant is not so save for data. Nands model in this case is BGA152. You can check electronic, like fzabkar suggest, but I think in this case chip-off recovery will be needed. This controller like use bad columns even in standard chip, so it could be tricky solution.

[Tomset]

but I do not have budget or clearance to farm the work out at this time.

Flew translator - 99%
Independent decision can be many times more expensive.

The first thing to do is to find out the ID chip.
See if you have solutions for these IDs.
http://www.pc3000flash.com/solbase/index.php?lang=eng
Get acquainted with the basic principles of recovery USB Flash.
http://www.acelaboratory.com/pc3000flash.php
http://www.youtube.com/user/PC3000datarecovery

On learning this process, it will take at least two years.

[Amarbir[CDR-Labs]]

Hello,

I am a new member of HDD Guru Forums, but I have enjoyed the articles and posts on and off for awhile. I appreciate the quality of this forum and its membership. Storage media and data recovery have been an interest of mine for years, during which time I've had a small number of DIY DR successes. No failures yet.

I have a PNY 128GB USB stick that failed immediately when it was connected to a high end, expensive Dell workstation running Windows 7. I connected a sacrificial USB stick to the same port and it developed the same problem. After booting a Linux O/S, I discovered over-current conditions in the system boot log for that USB port. I have no record of the drive's contents except that it contains 80+ GB of work-related files. Some of the data may be proprietary company information. Again, no clear record of that.

I do not see a fuse on the circuit board, nor do I see any physical defects under magnification. The controller is an Alcor AU6998SN with an integrated fuse and oscillator. I believe it may be damaged.

There are two 64GB ball grid array (BGA) NAND chips. I assume these are 8-bit. I'm not able to find information about these chips since they do not bear comprehensive markings. There are no markings underneath the paper sticker. I can get the chips off the board and onto a pogo pin fixture, but I assume that the chip pinouts are not standardized across the industry.

It's likely that successful DR will require chip-off procedures. I am familiar with the work of Bjoern Kerler, Joshua Wise and others. Countering that, I understand the excellent points made by Dmitry in his DIY post here http://forum.hddguru.com/viewtopic.php?f=16&t=12671 This should be a sticky. I realize there are complicating factors that differ with each job and the work should be left to DR professionals, but I do not have budget or clearance to farm the work out at this time.

A few questions for anyone who can spare the time to answer:

Does my analysis seem to be reasonable up to this point?
I have another controller chip (identical) but I assume that a controller transplant may be useless or even unwise, correct?
Is it possible to find data sheets, or even just pinout information, for the NAND chips used in this device?
This seems like a straightforward recovery, since the chips are intact. What might a commercial data recovery center charge for this?
Where would I find a competent DR provider in the United States, Canada, or Europe?
Are they better (training, experience, results) in some countries than others?
Why do DR agencies think it is their right to retain customer data? This is something I do not see enough discussion on.

Thank you for the interesting (sometimes fascinating) reading provided by this professional forum.

Kind regards,
wibit

Well,
A Lot of Foolish Computer Hardware Engineers " Generally And working With Gaints Like Dell " Connect USB Ports In Reverse " .+ They Connect To Ground And - They Connect To Positive .Nothing Happens To The Computer but As soon as you connect a pendrive hell breaks loose .I Do Not Think So You Can Handle This Until And Unless You Follow Franks Advice and Check The Points .You Should Also Use Low Ohm Mode " Has 0.25 Volts " Feature Few multimeters Have And Test For a Dead Short Between + And - . If It Is There Controller Is Fried .In That Case Removing The Chips And Connecting It To a NAND Flash Tool Like PC 3000 Etc Would Be Necessary .People Who Do This Daily Should Be Able To Handle The Same With Ease IMHO .

[fzabkar]

but I think in this case chip-off recovery will be needed.

Why?

[HaQue]

aside from what Arvika has already mentioned, there is added complexity with BGA chips. The pinouts of these can be mirrored with 1 set of control and DATA separate from another and in some cases 4 different sets of pinouts. the TSOP pads do not simply route to the BGA pads on a 1:1.

If a TSOP is soldered on, the Mass production tool is configured so, and appropriate discretes used in conjunction if required. If a BGA chip is used, then this can have other routed traces going to extra channels on the controller. many PCBs have various configurations of controller revisions, NAND chips and support components factored into the design. you see this by a TSOP/BGA mix board and unpopulated component pads.

The easiest method for this recovery is a flash recovery person to unsolder BGA chips and go from there.

some common BGA chips have some issues with reading that need to be addressed, only possible with commercial chip reading gear and their associated configuration settings, or I guess buying other flash reading tools and copying the commercial ones methods.

the data is going to need a XOR Key applied and mix of wear levelling sorted out, along with any corrections to blocks that are in conflict.

Does my analysis seem to be reasonable up to this point?.

I really don't see any analysis done yet, aside from no defects noticed when magnified. Rarely do you see a flash drive with components "blown" visually. And so far, it is quite unclear how many cases have been from failed components, or components out of spec. I am meaning any real provable evidence.

I have another controller chip (identical) but I assume that a controller transplant may be useless or even unwise, correct?

correct

Is it possible to find data sheets, or even just pinout information, for the NAND chips used in this device?

If you can identify what it is, you may be able to.

This seems like a straightforward recovery, since the chips are intact.

This statement is false, the chips being intact has no bearing on how successful the recovery might be.

What might a commercial data recovery centre charge for this?

Anywhere from $150(very low) to $800 at a rough guess, but there is no way to know without going through the process with a chosen company and getting a quote/price.

Where would I find a competent DR provider in the United States, Canada, or Europe?

look online, or look in the flash section of the forum and see members who appear to know what they are talking about and whittle down the list with the usual way you would choose a business. I know, not easy.

Are they better (training, experience, results) in some countries than others?

no, I know of reputable people in UK, Poland, USA, Russia, Asia, Australia.. etc

Why do DR agencies think it is their right to retain customer data? This is something I do not see enough discussion on.

I don't understand what you mean here.

[HaQue]

Well,
A Lot of Foolish Computer Hardware Engineers " Generally And working With Gaints Like Dell " Connect USB Ports In Reverse " .+ They Connect To Ground And - They Connect To Positive .Nothing Happens To The Computer but As soon as you connect a pendrive hell breaks loose.

Ive not sen that. In fact I have regularly been buying dozens of Dell & HP workstations and servers and deploying them in the Edu sector and I find them extremely well made and tested. The only real issues we have had are drivers during changing to different windows versions, and the odd compatibility between various 3rd party devices. Also it is quite hard to connect USB ports wrongly due to the keyed nature of the internal cable connector.

.I Do Not Think So You Can Handle This Until And Unless You Follow Franks Advice and Check The Points .You Should Also Use Low Ohm Mode " Has 0.25 Volts " Feature Few multimeters Have And Test For a Dead Short Between + And - . If It Is There Controller Is Fried .In That Case Removing The Chips And Connecting It To a NAND Flash Tool Like PC 3000 Etc Would Be Necessary .People Who Do This Daily Should Be Able To Handle The Same With Ease IMHO .

never say "handle with ease" for flash. There is NO WAY to tell how a recovery is going to go based on anything physical. Controllers go nuts and make things difficult. Controllers with same markings can have wildly different algorithms. NAND chips can be refurbs, etc etc.
Actually NAND chips that are not marked with a vendor and part number are a very big red flag to a refurb/crappy chip

[Amarbir[CDR-Labs]]

Well,
A Lot of Foolish Computer Hardware Engineers " Generally And working With Gaints Like Dell " Connect USB Ports In Reverse " .+ They Connect To Ground And - They Connect To Positive .Nothing Happens To The Computer but As soon as you connect a pendrive hell breaks loose.

Ive not sen that. In fact I have regularly been buying dozens of Dell & HP workstations and servers and deploying them in the Edu sector and I find them extremely well made and tested. The only real issues we have had are drivers during changing to different windows versions, and the odd compatibility between various 3rd party devices. Also it is quite hard to connect USB ports wrongly due to the keyed nature of the internal cable connector.

.I Do Not Think So You Can Handle This Until And Unless You Follow Franks Advice and Check The Points .You Should Also Use Low Ohm Mode " Has 0.25 Volts " Feature Few multimeters Have And Test For a Dead Short Between + And - . If It Is There Controller Is Fried .In That Case Removing The Chips And Connecting It To a NAND Flash Tool Like PC 3000 Etc Would Be Necessary .People Who Do This Daily Should Be Able To Handle The Same With Ease IMHO .

never say "handle with ease" for flash. There is NO WAY to tell how a recovery is going to go based on anything physical. Controllers go nuts and make things difficult. Controllers with same markings can have wildly different algorithms. NAND chips can be refurbs, etc etc.
Actually NAND chips that are not marked with a vendor and part number are a very big red flag to a refurb/crappy chip

Well,
Not All USB Ports Are Keyed And This Does Not Happen When They Make it In The Factory ,Generally Happens When a Company Authorized Service Providers People Mess it up .I Was Big Time into computer assembling one point of time and i have seen it too many times to forget .Sometimes if i was assembling i always used to measure voltage on usb ports with my multimeter if the cable was not keyed " as in many assembled third class cabinet " .

[fzabkar]

aside from what Arvika has already mentioned, there is added complexity with BGA chips. The pinouts of these can be mirrored with 1 set of control and DATA separate from another and in some cases 4 different sets of pinouts. the TSOP pads do not simply route to the BGA pads on a 1:1.

Thanks. I now notice that arvika mentioned BGA-152. Your post prompted me to look more closely at the OP's AU6998SN controller. I see that it has 64 pins and is the GIL version whereas I had been looking at the 48-pin GHL version:

http://gif.datasheetbank.com/image/Unsp ... 139859.gif

The latter supports TSOP and LGA flash chips, and its 8-bit reference circuit (http://www.users.on.net/~fzabkar/AU69XXGHL_ref_cct.rar) does appear to have a 1:1 correspondence. I can't find a circuit or pinout diagram for the GIL version, though.

That said, I wonder if it would still be easier/safer to remove the controller and gain access to the NANDs via the controller's pins.

I also notice that there are two supplies, 3.3V and 1.8V, rather than a single 3.3V supply.

[wibit]

you can try to image the drive first. Using a known good windows system with at least 130GB free , insert usb, open DMDE and see if the drive is detected there. ignore any windows messages, and in fact "this drive needs formatting" is welcome at this stage. Don't close the dialog box if it appears.

Good advice. I already tried to mount the drive on a Linux system to image it there, and I couldn't get it to mount. Same type of problem with DMDE in Windows--the drive is never ready. DMDE allows a parameter to not wait if the drive is not ready, but the 'OK' button on that window is grayed out (because the drive is not ready). The drive can't be mounted.

Attachments

[wibit]

A Lot of Foolish Computer Hardware Engineers " Generally And working With Gaints Like Dell " Connect USB Ports In Reverse " .+ They Connect To Ground And - They Connect To Positive .Nothing Happens To The Computer but As soon as you connect a pendrive hell breaks loose .I Do Not Think So You Can Handle This Until And Unless You Follow Franks Advice and Check The Points .You Should Also Use Low Ohm Mode " Has 0.25 Volts " Feature Few multimeters Have And Test For a Dead Short Between + And - . If It Is There Controller Is Fried

I must agree with Amarbir about poor product quality for Dell, and not just for this case. This machine was less than one week old when the incident happened. But it was the third machine delivered by Dell because the first two were built wrong compared to the order. The workstation has other problems in addition to the USB port (which is now sealed against further use). Video problems, hard drive problems.. our enterprise sees low quality (but high priced) Dell workstations all the time. This one is a T3610, not supposed to be a cheap machine.

I would confirm the voltages present at the Dell USB port. This will tell you what killed the sticks. BTW, have you tried the "sacrificial" stick in another machine?

There is a 3.3 ohm resistor (R1) in series with the +5V input. The voltage drop across R1 will tell you the current drawn by the stick. However, I would prefer to measure the resistance between R1 and ground. ... I would also measure the resistance of each capacitor (C3,C4,C5,C7,C8,C11). .. As for the flash pinout, ISTM that a "chip-off" recovery may be unnecessary, even if the controller is dead.

I did try the sacrificial stick in another machine. It came up as an unformatted unknown device. Same as this one. I haven't connected anything else to that port ever since, but I can check the voltages on it.
From my DMDE screenshot, the Alcor controller is recognized. So not completely dead as I thought it was, but I get a Device Not Ready error constantly. Same problem in Linux--the device can't be mounted.
I will also take some resistance readings as suggested.

Attachments

[wibit]

but I do not have budget or clearance to farm the work out at this time.

Independent decision can be many times more expensive.
The first thing to do is to find out the ID chip.
Get acquainted with the basic principles of recovery USB Flash.
On learning this process, it will take at least two years.

This is a good point. DIY is not guaranteed successful and not guaranteed inexpensive. Time is money.

[LarrySabo]

As Amarbir suggests, it's probably caused by someone reconnecting the USB-MB connector of the wires that lead to the front panel reversed, as I have done. Took out two USB flash drives before discovering the cause. It's amazing that these blocks are not keyed, but also that the connection design allows reversal of the connector to be data-destructive.

[wibit]

Thanks LarrySabo. Yes, Amarbir suggests checking out that port wiring. But also fzabkar's initial response points a finger at checking out that port (and hence the port wiring). And now you.
I know that the port caused the problem, but I just walked away from it. I will crack that machine open and check the wiring. If it's reversed it will not surprise me. Either way, I will never trust data to a USB port on that machine again. Once bitten, twice shy.
I have posted two other replies with screenshots and also commenting on the poor quality vs. high price of the workstation but they haven't appeared in this thread yet. They're pending moderator approval. In a nutshell, after receiving this workstation--the third attempt at a correct build from Dell--I found problems with the SSD, video card and monitors. Reported them all to my procurement team, and then I installed my own video card and HDD. The monitor problem... I just live with it.
I will check the wiring on that port.

[wibit]

5.08 volts at the USB port on the workstation. That is, +5 on pin 1, and Gnd on pin 4, per USB specification.
I already have three posts not showing up, pending moderator approval. Hope this one gets through.

[wibit]

DM Disk Editor can see the flash drive, and recognizes the controller (somehow). But the device is not ready and can't be made ready. I had the same problem when I tried to image the drive in Linux. It wouldn't mount.

Attachments

[HaQue]

IMHO, this is just a chip off recovery.

[fzabkar]

AFAICT, PNY drives should report a USB Vendor ID of 0x154B whereas Alcor's VID is 0x058F. I have no knowledge of flash devices, but whenever a USB bridge IC reports the bridge manufacturer's PID and VID rather than the OEM ID of the device, this usually indicates a firmware or hardware problem.

I would measure the voltage across each of the capacitors. You should find +3.3V, +1.8V and +5.0V. The voltage across the 3.3 ohm resistor would be a useful indicator also. If either the 3.3V or 1.8V supply is missing, then test for a shorted capacitor.