SD Card oddities

Guest · Posted: Mon Nov 10, 2008 5:07 pm

I'm using PCD 4.079 and a dsPIC30F4011. I'm using sparkfun's MMC/SD breakout board, and voltage dividers on the SDO, SCK, CS, and VDD lines. Using ex_mmcsd.c, I've been able to successfully write and read data. But it seems like the first byte either isn't written or isn't read. If I do this:

Guest · Posted: Mon Nov 10, 2008 8:03 pm

Edit: In the two program scenario, the sequence stored is 0,1,2,3,4, not 1,2,3,4,5.

asmallri · Joined: 12 Aug 2004 Posts: 1634 Location: Perth, Australia

Guest · Posted: Mon Nov 10, 2008 8:51 pm

I know the signals are going to the right pins. It wouldn't work even intermittently if they weren't.

Yes, I am creating ~3.3V from 5V with a 3k and a 1.6k. I do know that my wall wart powering my devboard, and therefore the sd card, is limited to 200mA. I guess I assumed that it couldn't work intermittently. I have no problem using a regulator, I was really just doing a proof of principle, and keeping it simple.

I will try a 3.3V regulator, a beefier power supply and a 10k pullup resistor.

asmallri · Joined: 12 Aug 2004 Posts: 1634 Location: Perth, Australia

In this case I am surprised you got it to work as well as you did.
_________________
Regards, Andrew

http://www.brushelectronics.com/software
Home of Ethernet, SD card and Encrypted Serial Bootloaders for PICs!!

Guest · Posted: Tue Nov 11, 2008 2:35 pm

I've put an adjustable regulator, set to 3.3V and a 15k pullup to 5V on the Data out pin from the SD card. I still get the same error, a bad first byte read back.

I've also noticed that I have to run the code twice to get output. The first time it runs, it hangs. But on a reset, it works. Another reset makes it hang again, and yet another makes it work again. So I kept advancing an output_high(PIN_B1) statement (turns on an LED) to see how far the code makes it, and where it hangs.

What I found was that on the times is doesn't work, the if (g_MMCSDBufferChanged) statement in mmcsd_flush_buffer() evaluates to true, while on the times it does work, the if statement fails.

So, when the buffer has changed, the buffer is flushed, and mmscd_write_block() is called. Inside this function, towards the end, there is a while() statement that writes 0xFF to the SPI bus and won't exit until a 0 is returned. My problem stems from the 0 never being returned, and the while loop becomes infinite.

Once I commented out the while() statement, it works every time. This does not correct the need for the extra dummy write at the beginning though.

So I have two workarounds to make it work, not waiting for the SPI line to return to low, and writing a dummy byte. But I would still like to get to work without the workarounds.

Any ideas?

asmallri · Joined: 12 Aug 2004 Posts: 1634 Location: Perth, Australia