Go Down

Topic: SD card read/write with Arduino (Read 219819 times) previous topic - next topic


Jan 12, 2009, 04:14 pm Last Edit: Jan 12, 2009, 06:26 pm by wiji Reason: 1
Is it possible to change the Arduino pins that are used by the library? I can't seem to find any reference to the pin numbers in the library source code.



I've found out you need to use the SPI pins, so cannot change the pins in use by the SD card reader... time to get rewiring :D


Hi everyone,

Earlier, ducttapepro6 wrote:

Mr. Orange-

I had no trouble getting your code up and running in 15 minutes - happily logging, saving, and reading the input to analog1.

However, after hitting the arduino reset button, I found that the read function returns junk(from the SD card).

Shouldn't the SD card retain data after a power cycle? or, is it a function of the code that you so nicely supplied to get me started - and I just don't understand it, yet?

Along with Daniel, I'm hoping that someone can explain why the card starts up with the same junk data each time, if the first thing I do when the sketch runs is hit 'r'.

agent_orange says that it's like having a 1G (or whatever) EEPROM, which is IDEAL for my situation. My arduino will be sitting in a box, on top of a pole, powered by a solar panel for up to 3 months at a time. I have it successfully writing the date and max & min temperatures to the EEPROM, which I can read back out as desired. The EEPROM data is only overwritten after a new sketch is started. This is the functionality that I require from the SD card, but the supplied code appears to wipe the card each time the sketch is started. (I tried commenting out the if(!sd_raw_init()) but that didn't help).

agent_orange, thanks so much for supplying this work, it's fantastic & there's no WAY that I'd be able to work this out on my own. But if anyone can show me a way (even if it's a seperate sketch) to access the data on the card without destroying it, it would be GREATLY appreciated!!

I guess the other option would be if someone could recommend a Windows-based program which could access the data on the SD card?

Thanks for reading everyone!!


    I am using the same simple logging code that you have posted with just the raw read and write.  The initialization seems to go fine, and then it acts like it is reading and writing even when i have the SD card removed from the slot.  I enter the 'r' command, and it prints the same info, not zeros or junk. when i reset, it loses the info. It is as if it is just storing it in on board memory until it resets, like it isn't even using the sd card at all, and then loses the data when it resets. 

Any ideas?

Thank you Agent Orange for working on this project.  


Same here... at first it looks like it's working fine, it appears I can sample data to the card and read it back. But once I've reset the board and try and read the data back I am presented with all zero's.  :(

I too can confirm that reading and writing appears to work even without a card in the slot, despite the fact that it says initialisation failed and doesn't show any card statistics.  :-/

With a card in the slot I get reasonable looking card statistics (size: 125698048 for a 128MB card) which suggests it can see the card.

Which again raises the question, what are we missing here? Any ideas?

I will be sure to post if I make any further discoveries.

Thanks all! :)


Hi there.

I found another solution that I think is much better for storing large amount of data and it is Fat16+32 compatible. You can communicate with the card with normal serial, read more: http://www.roguerobotics.com/ uMMC datastorage.

Regards /Daniel :)


Hi Daniel,

That's certainly a nice-looking solution! Might be a bit of overkill for me though; using a 16Mb SD card that came with my camera, I worked out that if I store 10 bytes of data every minute, I've got over 2.5 YEARS worth of storage capacity...

Just for my curiosity, how would you communicate with this uMMC device? I'm guessing that you'd have to Serial.print() commands like "O 1 R /LOGS/JANUARY/JAN03.LOG{cr}" to it in it's own language?



Okay, I am getting somewhere...  

I thought I would try the uFat library that was mentioned fairly early on in this thread, which straightaway gave me a directory listing of all the files on this card. But hang on a minute, I thought I had formatted the card! So I popped it back into the computer, deleted all the files and created the required data.bin file, but to my amazement my new file disappeared and all the old ones came back! No amount of deleting formatting or mortifying seemed to affect this card any more, so I have scrapped it and got myself a new one.

uFat seemed very temperamental with my new card, failing to initialise nine out of 10 times. I put a simple loop round the initialising statement with a delay, so it would retry if it failed instead of just showing an error and halting. This seemed to work a treat, no matter how much delay before or after it always seems to fail once and then initialise fine on the second attempt without fail. Reading and writing then worked perfectly :D

I think I am going to stick with uFat over the raw library as it isnt any bigger and I still can't get the raw library to work even with my new card (same problems as described before), also I like being able to read my data back on the computer. :)

The only downside is that the code for using uFat isn't as friendly as the raw library, and I can't seem to get it to write one byte at a time, it has to be written in 512byte sectors, but I can live with this.

Hopefully this information helps someone out, I am thinking to get the raw library working perhaps something needs to be inside a retry loop much like I had to do with the uFat library to get it working.

Thanks everyone who has contributed, especially agent_orange for the circuit diagram (which needs no modification to work with uFat), and sirmorris for the uFat library.


Hey bobemoe this sounds good. I've tried the RAW program and haven't had any luck. I'm currently attepmting SirMorris's code and am not getting anywhere there either. I get some errors when I try to compile.

o: In function `microfat::initialize(unsigned char*)':
multiple definition of `microfat::initialize(unsigned char*)'hardware\libraries\uFat\microfat.o:C:\arduino-0010/hardware\libraries\uFat\microfat.cpp:17: first defined here

o: In function `microfat::locateFileStart(char const*, unsigned long&, unsigned long&)':
o: In function `main':

Any sugggestions?
I have the required header files and I'm trying to run the microfat program.

My apologies if this has already been resolved somewhere else 8-)


Jan 23, 2009, 06:43 pm Last Edit: Jan 23, 2009, 06:44 pm by bobemoe Reason: 1
Hi smelios,

It doesn't look like you are using the latest version, try downloading uFat2.zip rather than mmc1.rar, hopefully that will fix the problem.


Fresh code mmmmmm.

Still having problems. Bobemoe, did you encounter an error like this?

error: 'directory_entry_t' was not declared in this scope In function 'bool showDirectory_walkerfn(directory_entry_t*, unsigned int, void*)':
In function 'void showDirectory()':

Or did yours compile with no problems?

Sorry to keep asking you questions i'm a noob.  ;D
Your help is greatly appreciated!


Mine compiled fine.

I can get a similar error if i remove the #include lines from the top of the example which makes me think its a problem with your includes.

Did you delete the files you copied to /hardware/libraries/ for version 1? they may be conflicting.


Found my error, I was on arduino10. Morris said he developed this in Arduino12 so i downloaded that and it complied. Now i just can't init the sd card. Oh well work for another day :P
Thanks again for helping


Excellent, now you are at the same stage that I was. Try making it retry the initialisation.  This is what fixed it for me, I changed:
Code: [Select]
 if (mmc::initialize() != RES_OK)
   error(PSTR("mmc init failed.\n"));

Changed to:
Code: [Select]
 while(mmc::initialize() != RES_OK){
   pprint(PSTR("mmc init failed. retrying...\n"));

I find it always fails once. If that doesn't fix it I am all out of ideas. Good luck.


Hi again,

After further reading (ie that uFat basically takes up the same space as the sd_raw instructions) I've headed down the uFat track. I've downloaded the uFat2 files from SirMorris's blog (http://groups.google.com/group/micro-hacker
) and have integrated them into my program, although I have a few problems...

First, the sketch seems to 'interfere' with my DS1307 real time clock data. With the SD card connected as per the diagram at the beginning of this thread (and the clock connected to analog pins 4 & 5), the clock output is 'mixed up', e.g. the day comes through as the hour, year comes through as day of the month, etc. Leaving the hardware connected and loading up a sketch which ONLY displays output from the RTC, I get the correct output from the clock.

Secondly, I'm not too sure as to exactly how the reads/writes work. Am I correct in thinking that if (RES_OK == mmc::writeSectors(sector_buffer, sector, 1)) simply checks that a 1-byte write to sector (which=0) of sector_buffer works? Is it just the digitalWrite(13, (millis() / 1000) & 1) which writes sector_buffer out to the SD card? If that's correct, can anyone explain how this works for me? I'm very much learning as I go...

I'd like to be able to write data out to the SD card 10 bytes at a time (is this possible, or does the buffer have to be 512 bytes?), and be able to read them back & display them when prompted. Is this also possible, or am I only going to be able to check the output by checking the data.bin file on the PC?

Thanks again for everyone's thoughts...



Hi JB.

The line:

if (RES_OK == mmc::writeSectors(sector_buffer, sector, 1))

writes 1 sector's worth of data stored at sector_buffer in the arduino's SRAM to the specified sector on the card. Once this routine returns your data is stored. The line:

digitalWrite(13, (millis() / 1000) & 1)

has nothing to do with the workings of the program. It simply flashes the LED on port 13 to let you know all's done.

All reads and writes in uFat are 512 bytes. This is the size of a sector on the card. It would be possible to write a function that allows arbitrarily-sized writes but that would add to the size.

Do you only want to ever have 10 bytes stored? Or are you saving a stream? When you want to read it back would you be showing all the data stored thus far? What's the expected data rate? If it's a slow dribble of data to the card that would warrant a different approach than if you're going for realish-time.

If you can describe your program requirements I'll see wht I can do to help you out.

There's no obvious reason why uFat should interfere with your RTC. It sounds like you're using 2wire for the clock. I have a board that uses I2C components simultaneously with hardware SPI without a problem, I may need to look over your code.


Go Up