Dumps are completely wrong. 01_01 at the beginning have 3 repeat pages. Later dump is bad, too much zeroes, you can look on HEX, you should see it Second dump do not have repeats, just too much zeroes.
It could be wrong RR values.
jeremyb
Joined: 09 Dec 2008 Posts: 1876
Location: RecoverMyFlashDrive.com Bridgeport, CT, USA
Posted: Thu Aug 25, 2022 4:43 Post subject:
Uploaded dumps using the new methodology, they shouldn't have that issue anymore..
Upload Complete
jeremyb
Joined: 09 Dec 2008 Posts: 1876
Location: RecoverMyFlashDrive.com Bridgeport, CT, USA
Posted: Thu Aug 25, 2022 19:05 Post subject:
Methodology:
I’m going to call this phenomenon “Page Bleed” where by modifying the Read Retry registers will cause the contents of one of the WL pages to replace another. In WL mode there are lower, middle and upper pages. In my experiments if all the registers are written to, even with a value of zero, it will cause page upper to “bleed” into page lower causing it to have an identical read to page upper.. (ie: page lower = page upper).
This chip (ad 3a 18 03) uses Hynix v3-7 which has seven registers which control the following pages:
Page Lower : 2, 6
Page Middle: 1, 3, 5
Page Upper : 0, 4
In my opinion (I might be wrong about this) the registers are a 16bit number with most and least significant values. The MSB values appear to be 6, 5, and 4. In my case I was able to ignore the rest of the registers (2, 1, 4, 0).
To find the correct values:
1. Reconnect the reader to flush any previous settings
2. Load test retry with an initialization value of
Hynix v3-7 / -- -- -- -- -- -- 00
3. Use Tune with a Regs of 6 (Page Lower), Delta of 128, and Repeat of 1
4. Repeat for Regs 5 and 4 (not necessary in my case) modifying the position of the register in step 2 and register value in step 3.
Once done load the values, eg: Hynix v3-7 / -- -- -- -- -- -- -- fc into the initialization value and do a Test Read, the read should look good.. DON’T copy the updated value into NAND reader as it was loaded the flash memory chip by the last step, just read it normally.
From experience it seems like page upper always reads perfectly with these chips with no adjustment, page lower and page middle need the adjustments but using Find Saddle or Find Base will cause page bleed to occur, so the good value from page upper will transfer to page lower making it appear that you have a good read when you don’t.
In this case page middle had very little corruption, it was all in page lower. To get the best read modify the initialization value +/- 4 for each register and re-read/ecc combine.
I also read this chip at 2.2v using my own voltage adapter, it gave me a slightly better read.
Second dump do not have repeats, just too much zeroes.
