Flash Extractor
Soft-Center
About     Shop     Downloads     Manual     Library     Forum     Services     Contacts
Reading Bad Quality Chips   Search  Register  Log in
Reply to topic
Author Message
jeremyb



Joined: 09 Dec 2008
Posts: 1531
Location: RecoverMyFlashDrive.com Fairfield, CT, USA

PostPosted: Sun Feb 16, 2020 0:53    Post subject: Reading Bad Quality Chips
Reply with quote

arvika wrote:
Currently there is a big problem with reading chips with good quality dump.

What are some IDs of NAND chips that are having issues?
Has adjusting the Vcc helped?
Has Read Retry helped?

My antidotal notes and thoughts not backed up with fact:

WL Toshiba / SanDisk:
Erasing a block doesnt set all bits to 1 or 0, instead most of the bits are set to 1 or 0. (See picture).. FE gets stuck on these trying to process the data.

My solution: Count the state of bits in the first 128bits of a block, if its more than 85% either way disregard the block and fill the block with set bits.. its a variable percentage but my reasoning is XOR distributes the number of bits, if the distribution is bad the underlying data is probably bad. This greatly speeds up FE on applicable cases (eg: SanDisk)

Variable Voltage: Do Toshiba/SanDisk NAND work with the Power Adapter? From memory every time I've tried I get an ID of 0x00ffffff, ?? Do Toshiba Chips NOT start up Tri-State ??.. From memory covering Vcc pin on the NR and supplying external voltage (eg: 2.55v) only works for monolithics, especially monolithics where the controller powers up when external 3.3v is applied sending spurious data to the NR..

Constant Voltage: Some Toshiba/SanDisk NAND chips need voltage applied to pins not supported by TSOP-48 adapter v3... eg WP or 16bit I/O.. This must be done manually.

Hynix:
Many of these chips need Read Retry Modes and the applicable registers vary by ID however its implementation would be complicated. Specifically, unlike SanDisk or Intel where a Read Retry mode is generally valid for an entire bank.. Hynix Read Retry modes are only valid for a page (upper, middle, lower), block, or plane. If a bad page is encountered every read retry mode is cycled through to find a good value. Currently the only option is ECC combine however it's not efficient. I haven't had any luck using different voltages but I haven't done any extensive testing with Power Adapter v3 so I might be wrong..

My Solution: Use Read Retry Codes in Segments and merge the files together..

Intel / Micron:
Variable Voltage - Use Power Adapter

Also sometimes I think it must be acknowledged the the chip might just be bad.

Does anyone else have any antidotal input on bad quality chips?
Capture.JPG
Erased block where most bits are set to 1
Display posts from previous:   
Reply to topic All times are GMT + 4 Hours
Page 1 of 1

 

Last added
82-00709-A0   45 3a a4 93   1x1 PS2251-09-26   98 3a a4 93   1x1 SM3257EN L BA   45 de 98 92   1x1 Monolith USB 60   ad 5e 28 22   1x1 PS2251-07-V   98 3a a4 93   1x1 SM3269M AA   2c a4 64 32   1x2 Monolith USB 59   98 3a 98 a3   2x1 Monolith USB 58   2c d7 94 3e   2x1 Monolith USB 57   ad de 14 a7   1x1 IS917   98 de 94 93   1x4
News
10.03.2020 RR for ad 3a 18 03
01.12.2019 Power Adapter v3
20.04.2018 Sector Number Slow
17.03.2018 Adapter BGA-272
22.02.2018 Bad Bytes Editor
Other products
Soft-Center ltd.