SD card memory problem

[andrewg]

I don't know what mmcsd_read_ocr() does, but it is described as a "non-user function". I would ignore all "non-user functions" and just consider "user functions". mmcsd_crc7() and mmcsd_crc16() are also "non-user functions". Ignore them too.

I didn't expect you to test the functions individually, just develop your project as normal, but include error checking of every mmcsd function you call.

If mmcsd_write_block() isn't returning, then there is a fault in the comms between the pic and the card, or a fault in the card itself. Try slowing down the SPI clock speed. Try a different SD card.
_________________
Andrew

[ckielstra]

[Jim90]

I used a component like this :
http://www.babelduck.com/sd-card-module/
IS this ok or maybe the problem is there ?

I used the following code

[ckielstra]

The garbage characters in your RS232 communication are most likely caused by the clock being driven from an RC-oscillator. Stable RS232 requires the clock to be within 3% error margin. Given that most capacitors have 10-20% tolerance I'm surprised you receive any readable data at all.
Then, with temperature changes the clock will change a lot too.

The only way to get reliable R232 is to use a crystal or external clock driver.

SPI is not sensitive to variations in the clock speed so this is not what causes your SD card problems.

I couldn't find a schematic for your SD card interface module. There are a lot more resistors mounted than I would expect. Most likely these are the required pull-up resistors for the clock and data lines. That's fine, as long as you didn't add extra pull up resistors on your own board as well.

The only reason for "HELLO" not being printed in your modified mmcsd_write_block() function I can think of is that you have made a copy of the C-file and that the compiler is still using the original file and not your modified copy. To prove this, make a syntax error in the file and see if the compiler generates an error message.

[Jim90]

ckielstra i tried to implement your advice but unfortunately without success.

I use the fuses below :
#fuses HS, NOWDT, PR, NOPUT,NOPROTECT
#use delay (clock=20M)

and a 20M crystal with resistor and 2, 27pF.

also i did the change below (after the compiler's message)

/*#use spi(MASTER, DI=MMCSD_PIN_SDI, DO=MMCSD_PIN_SDO, CLK=MMCSD_PIN_SCL, BITS=8, MSB_FIRST, MODE=3, stream=mmcsd_spi,BAUD=4800)
*/

#use spi(MASTER, DI=MMCSD_PIN_SDI, DO=MMCSD_PIN_SDO, CLK=MMCSD_PIN_SCL, BITS=8, MSB_FIRST, MODE=3, stream=mmcsd_spi,BAUD=9600)



Can you suggest me what to do ???

[Ttelmah]

and, with that done, have you tested that the unit is running at the expected frequency?.
Simple basic test. Pulse a line on ()feeding an LED). have it stay high for (say) ten seconds, and then drop. Time this with a stopwatch. Are you actually getting 10 seconds (not 9.5, or 10.5)?.
This is the core of all debugging an testing. Do one thing at a time, and _prove this is working rioght_, before trying to move forward.

Best Wishes

[andrewg]

I thinks there's some confusion between SPI and RS232. The #use SPI should not have baud=9600 (or any baud at all).
_________________
Andrew

[Jim90]

Ttelmah i fix the problem (it was a bad connection of resistor to the crystal)

andrewg i replace the

[andrewg]

The code you've quoted seems to show calls mmcsd_write_block and mmcsd_read_block are duplicated, first not checking the return code, then checking the return code. You only want the call that checks the return code.

To assist in determining what the problem is, what is the error code returned by the calls?
_________________
Andrew

[andrewg]

I've also checked your method of calling the function. The ",10" part of the parameters "(4096, 512, 10)" does not seem correct. That should be a pointer to a buffer. Your code should be more like:

[Jim90]

andrewg did you use the code that you wrote and test it ?

If yes what was the results ?

I tried the code and again i receive the message :

write block fail :-128

(128)10 = 80 in Hex


In enum MMCSD_err is define that

RESP_TIMEOUT = 0x80

what does mean this ? (communication problem between card and microcontroller ?)

How to fix it ?

[Ttelmah]

The error is printed with %d. The format only has ':'. The -, is a negative sign. -128, is 0xFF hex.

If you look at the code, the returned error is from the defined values if the functions themselves return errors, but if the SD card returns anything that and's with 0xA, and is non zero, you get back the byte returned by the card.

Best Wishes

[Jim90]

Ttelmah, i have a little confusion.

If i replace the
printf("write block fail:%d\n", err);

to

printf("write block fail:%x\n", err);

and then i receive the value 80. (Ttelmah, according to what you described i expect it to receive -80).

[andrewg]

"%x" will print err as an unsigned hex value, e.g. "80" just like you see. If you like, you could change the printf to:

[Jim90]

andrewg, you said that :
"I've also checked your method of calling the function."

Did you try to compile the functions of mmcsd_write_block,mmcsd_read_block and program an SD card ?

If yes did it working ?(and if is working what are the differents from my code/design?)