PC-3000 Flash. How to work with the problem memory chips

admin

There are a lot of different memory chips with reading problems such as not corrected ECC errors, dump filled by “FF” or other pattern etc.

In this article we would like to talk about different problematic memory chips and solutions for them.

TLC Samsung

EC D7 98 CA

In PC-3000 Flash SSD Edition version 6.4 was added method for such memory chips – Read Retry.
Using a Read Retry method in this case allow to solve the problem of bad reading quality with out decrease the power or cooling. However may be required using alternative methods, ’cause Read Retry option is not guarantee the 100% result.

Alternative methods it’s decrease the power or cooling.
For decrease the power is recommended use the external power supply swich allow to work in the range 1.50-3.3 V with voltage regulation with 0.01 V step.
Impossible to perform decrease the power procedure for memory chips reading by the PC-3000 Flash v.1, v.2 and v.3 reader regular means and we advice to work with this case on PC-3000 Flash reader v.4 or on PC-3000 Flash reader v.3 with Power supply adapter.

The reason why need to use 0.01 V step:

…now I understood while 0.01 step is important thing.
below 1.57 V a lot of errors
1.57 – corrects nothing but no errors
1.58 – very fasr corrects
1.59 – almost nothing corrects

The same situation on rage 1.58-2.5 V.

EC D7 98 CE
EC DE 98 CE
EC A5 98 CE

These are another group of problematic chips.
Unfortunately cooling procedure is not help for such memory chips.
The only way for these memory chips is decrease the power.
A feature of these chips is that ID can’t be read if voltage below 2.7 V.
You can get a good result for these chips somewhere on the border of the minimum voltage, that’s why need to check a different voltage value with 0.01 V step.

Also, there are very rare cases whichs have a lot of ECC errors and described methods are not get a good result.
Such cases can be fixed by 4.1 V power supply value.
The idea according to the datasheet the memory chip should work only in the range between 2.7-3.6 V (3.3 +/- 10%), but with larger value mem chip can be damage, but at practice it can work with 4 V power supply value (but increasing to larger value is danager).

Let’s summarize:
Read Retry work in range between 3.6-3.9 V but sometimes need to set 4 V value and ECC will be corrected perfect.

TLC Toshiba/Sandisk

98D79892
45D79892
98DE9892
45DE9892

…and other which are read by the WL algorithm.
Most of sch memory chips allow to use hardware Read Retry option, which is help to improve a reading result with out cooling and decreasing the power.
Note, that cooling and decreasing the power procedures are not allow to improve a result and the only way is re-read the problem sectors from submap with active Read Retry option.

Also, please note that Toshiba and Sandisk memory chips have a different Read Retry types and wrong Read Retry procedure can not be used:

1. Read Retry from Toshiba mem. chips can not be used for Sandisk mem. chips and vice versa
2. Do not confuse technological standards for running a specific Read Retry type.
If you can’t to determine the technological standards of memory chip production let’s check the following reference material about 6 bytes chip ID decoding:

For example we have two memory chips: 98DE98927656 and 98DE98927657.
Looks like they are the same but but it is not. Differences have not only in the Read Retry type.

56 = 1010110
57 = 1010111

Highlighted bits responsible for the definition of the technological process of Toshiba and Sandisk chips, and if you look at the table…

000 – 130 nanometers.
001 – 90 nanometers.
010 – 70 nanometers.
011 – 56 nanometers.
100 – 43 nanometers.
101 – 32 nanometers.
110 – 24 nanometers.
111 – 19 nanometers.

…then becomes obvious, 98DE98927656 is require Read Retry Toshiba 24 nm, but 98DE98927657 is require 19 nm.

32 nm TLC Toshiba и Sandisk.

List of memory chips (Read Retry ID 16):

98D798827600
98DEA8827A55
45D7988276D5
45D798827600
45C798B276D5
45D798B276D5

These memory chips are also have a some problems, ’cause:

1. If you read the memory chip with Bad Column Managment active option chip is not allow to read Addings table as result need to remove them manually by utility option.
2-bytes or 1-bytes mode is also selecting manually, the new chips can have both addings types.

2. Basically these mem. chips have a lot of bit errros, that’s why PC-3000 Flash utility have a special Read Retry ID 16 procedure which allow to re-read sectors by submap.
Alternative metods (cooling, changing the power) are not bring a good result.

MLC Toshiba/Sandisk

Among MLC chips from these manufacturers there are some problematic. example98EE9532, 98DE9532 etc.
You can separate such mem. chips from TLC chips by page size value.

For MLC it is 8676, 8640, 8832
for TLC it’s 9216.

Besides you can separate MLC and TLC by third byte in Chip ID.

For example 98DE 98 92 and 98DE 95 32:

98 = 10011000
95 = 10010101

Highlighted by blue color – bytes are responsible for the count of data banks,
highlighted by red color – bytes are responsible for memory chip type (SLC, MLC, TLC, QLC) and can be:

00 – SLC
01 – MLC
10 – TLC
11 – QLC

Based on this we get the following result:
98DE9892 – it’s TLC and 98DE9532 it’s MLC mem chips.

Cooling procedure is not help for Toshiba and Sandisk MLC memory chips and decreasing the power procedure help to improve a reading result of memory chips only with 24 nm and 32 nm technological process (and it is not always).
Pc-3000 Flash have two types of Read retry for such memory chips:

– Read Retry ID 7 (Toshiba Mode 0) – for MLC Toshiba 32, 24, 19 nm.
– Read Retry ID 14 (Sandisk Mode 2) – for MLC Sandisk 24, 19 nm.

Usually Read Retry option for these memory chips allow to get 100% successful result.

MLC/TLC Hynix

ADDE94D2

It’s very problematic memory chip. Lowering the power procedure is almost doesn’t help , but in the last Pc-3000 Flash software versions you can use the hardware Read Retry option.

It allow to get 100% (or around 100%) successful result.

MLC/TLC Intel/Micron

By and large, the Intel company does not produce NAND Flash chip itself, they bought them from Micron’s, hence the chip is essentially identical in all aspects.

MLC and TLC Micron memory chips have a problems caused by cells wear as result dump will have a lot of errors pages, but
cooling and decreasing the power procedures are not help.

For Intel and Micron mem chips you can use one of two Read Retry modes:
Read Retry ID 1 and Read Retry ID 2.
Here is list of memory chips Chip ID for Read Retry options:

Read Retry ID 1 (L84A/C)

2C 64 44 4B
2C 84 C5 4B
2C 88 28 5F
2C 88 08 5F
2C 64 64 3C
2C A8 09 DF
89 88 24 4B
89 88 28 5F

Read Retry ID 2 (L74A)

2C 88 24 49
2C 88 24 4B
2C A8 05 CB
2C A8 25 CB
89 88 24 49
89 88 24 4B
89 88 04 4B

ECDE98CE and EC3AD9CE

Very problematic memory chips with lots of nuances.

First of all pay attention to the fact that these chips with the first 4 bytes have ID with a different technological standards and therefore with a different Read Retry and different reading parameters. PC-3000 Flash utility have an EXT ID for these chips, for this reason, select the type Read Retry manual is not necessary, but just for note:

ECDE98CE 74C3 – 24 nm, Read Retry ID 21.
ECDE98CE 74C4 – 19 nm, Read Retry ID 20.

EC3AD9CE 74C3 – 24 nm, Read Retry ID 21.
EC3AD9CE 74C4 – 19 nm, Read Retry ID 20.

For some revisons of ECDE98CE 74C4 and EC3AD9CE 74C4 need to use Read Retry ID 19, but it is rare, basically – ID 20.
If these memory chips have a very damaged memory cells there is a good way to re-read problematic sectors by submap with additional heating of memory chip with soldering Iron up to 150-250 *C.
Also pay attention to the fact that some controllers using these chips can write a some garbage data at the beginning and end of the dump that are not re-read under any circumstances, that basically normal behavior.

Basically the controller chip is IS916EN/IS916D.
Also can be EN2685F controller chip (using on the modern SD cards).

How to use a Read Retry option

EC1AA8DE 88C5 and EC3CE9DE 88C5

These memory chips have differ block size and differ planes count from ECDE98CE and EC3AD9CE mem chips.
So, these memory chips have a not standart block size 256 pages, in other words it’s not a multiple to 3 (while original TLC Samsung mem.chip have 192 pages block size) also it have a differ count of planes which introduces some nuances for Typical Interleave Elimination procedure.
Generally Typical Interleave Elimination for memory chips with 4 plance is (can be some differences):

Split by Blocks (block size * sectors number in page)
Join by Blocks/Pages for pair (sectors number in page)
Split by Blocks (block size * sectors number in page*2)
Join by Blocks/Pages for pair (sectors number in page*2)

Basically for this mem chip need to use Typical Interleave Elimination with such parametrs:

As we can see it’s 2 Typical Interleave Elimination preparations.
Regarding reading: Quality of these memory chips approx the same like of ECDE98CE.
PC-3000 Flash have a special Read Retry ID 17 for these memory chips.
Similarly as for ECDE98CE current memory chips reading result can be improved by heating up to 150-250 C*, but sometimes to 300 C*.

Problem with reading Chip ID of TSOP memory chips with 9 and 109 power schemes.

At present time some memory chips have the problem with reading Chip ID.
Here is the list of problematic memory chips:

ECDE98CE74C3
ECDE98CE74C4
EC3AD9CE74C3
EC3AD9CE74C4
45DE94937657
45DE949376D7
454CA8927657

So, what need to do for getting Chip ID:

1. Close the Flash software and remove the memory chip from the reader.
2. Repluging PC-3000 Flash USB cable.
3. Open (create) Flash task.
4. Insert the memory chip in to the reader.
5. Launch reading Chip ID and press “Skip” button.
6. Remove ticks from all power schemes except 9 and 109.
7. Press “Read” button.

Problem can be solved on PC-3000 Flash reader by the last firmware.
Manual for updating to the last f/w version is available in your UpdateBox.
(Update Flash Reader.pdf document)

Problems with dumps reading from potentially faulty chips.

There are cases with MLC memory chips with reading problem when after reading procedure whole dump is filled by “FF“:

Why it can be?

1. The most common is repaiting of flash pen drives by manufacture tools and ignoring warnings that data will be loast. For example Transcend Online Recovery is not always can to save the users data before to start repairing attempt.

2. Firmware (controller chip) failures. Such problems are not a rare and most often onSM3255EN/SM3257EN controller chips, also can be on IS902. Can be and on cases based on other controller chips.

3. Deterioration of memory chips (failures in the logic circuit or memory cells problems).

Let’s take a closer look.
As you know PC-3000 Flash reader read the data directly from the memory cells to the dump.
Bsically utility get a correct (true) data from the memory cells but in some cases inner logic of memory chip have a big timeout and work slowly. As result we receive a package gaps, mem. chip is freezed and transmit pages with “FF” pattern or filled by last byte which got from reader.
Such issue can be fixed by increasing and decreasing power supply method, and also by changing Timings value.
In reading settings you can find the window for changing Timings values:

On the appointment of each parameter you can read in ONFI specification document (also available in your UpdateBox), just note that the proper selection of reading options can eliminate the problem of filling dump by “FF” or other constant.
Basically this problem can be solved by following parametrs (but can be and others):

! Please note that increase timings would increase the reading time.
After re-reading procedure with these timing values dump get a correct (true) users data:

New 19 nm SAMSUNG memory chips. Technique with out heating on method.

! Only for users with PC-3000 Flash reader v 3.0 + Power supply adapter + external TSOP adapter and users with PC-3000 Flash reader v 4.0.

This method can be used for following memory chips:

EC0000DE00C5
ECD598DE94C5
EC3CE9DE88C5
EC3AD9DE94C5
ECAEB8DE86C5
ECD798DE94C5
ECDE98DE94C5
EC1AA8DE88C5

1. Set the values of Power supply like on screenshot:

2. Set the timimgs:

3. Set the values for re-reading:

…then launch re-reading procedure.
Finally we have a good reading result:

Also please note that Read Retry is not an universal procedure for getting a good result of ECC problems.