128mb chip not reading correctly about snander

[McMCCRU]

Hi! Is it a desoldered chip? You read flash in mode On-die ECC NAND? Use flag -d? I reflashed more than 30 chips of this name, never had any problems!

[jacksonium]

Yes, it is a desoldered chip and I put it on a WSON8 to DIP8 adapter to use with CH341A. I read it with the -d flag.

Do you use windows/linux/mac primarily for this?

[McMCCRU]

If the problem data was multiple of the page sizes of 2Kb, then it would be possible to understand the problem, but in a few bytes, this is more likely your problem on your computer. The programmer reads everything into memory, and only then saves it to a file. Try using another computer. I tested on Linux and Windows this chips, no problems..

[jacksonium]

Comparing files 8.bin and 9.BIN
00374EE1: 73 63
00859059: DB 5B
0114DEBF: 0C 4C
02A183E3: EA FA

This is an example. If i read the adress 02A183E3 by specifying:

SNANDer.exe -d -a 0x02A183E3 -l 10000 -r try1.bin

Out of 10 try. I would get 5 EA and 5 FA as the first byte of the file.

[McMCCRU]

Is the difference always at the same addresses? Try to solder capacitor on power supply of the flash chip at 0.1-0.47uF. You can also try commenting out several lines in the int enable_pins (bool enable) function CH341A_CMD_UIO_STM_OUT | 0x37 in ch341a_spi.c to reduce CS latency...

[jacksonium]

you mean commenting out all the lines with 0x37 in func enable_pins() ?

[McMCCRU]

No. In turn, at the beginning one, then the second and so on, but at least one must remain.

[jacksonium]

like this?

uint8_t buf[] = {
	CH341A_CMD_UIO_STREAM,
//	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_OUT | 0x36, // CS low (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_DIR | (enable ? 0x3F : 0x00), // Interface output enable / disable
	CH341A_CMD_UIO_STM_END,
};

and then this?

uint8_t buf[] = {
	CH341A_CMD_UIO_STREAM,
	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
    //	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_OUT | 0x37, // CS high (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_OUT | 0x36, // CS low (all of them), SCK=0, DOUT*=1
	CH341A_CMD_UIO_STM_DIR | (enable ? 0x3F : 0x00), // Interface output enable / disable
	CH341A_CMD_UIO_STM_END,
};

[McMCCRU]

No. You commented out one line, checked, if it didn’t help to solve the problem, commented on the next one, leave the previous commented out, and so on!

[McMCCRU]

I strongly doubt that the problem is with my software, several people have been using it for over two years, no one has had a problem. Most likely the problem is in your programmer, a problem with the power supply, you may need to solder pull-up resistors for signals or a strong ripple in the power supply. Perhaps you need, on the contrary, to increase the delay time, add lines to the recommended function. I don’t know, didn’t face such problem. Perhaps your chip has bad blocks in the places it is trying to read ...

[jacksonium]

just tried on another computer and it is the same. Maybe as you say, the chip has bad blocks...

I have a question though, will the ECC in the chip itself take care of these bad blocks during runtime?

[McMCCRU]

The NAND controller of your device, from where you desoldered the chip, uses its own work ECC algorithm. Most likely, he created a backup copy of the broken blocks and stores a table of these blocks somewhere on the flash. If you enable the mode of the internal ECC algorithm in the chip, then most likely the device will not work for you. NAND controller has its own view of data storage. Make a flash memory backup with the -d option, I think it will work for you even if there are bad blocks.