SD card > 1GB working !

[Sebastien Lelong]

Hi guys,

Trev, from Jaluino, has made several tests with SD cards, running on Jaluino Bee. He's been able to make 2GB SD Card working, see below. Matt, can you tell us more about this SD_DELAY_AFTER_WRITE constant ? When true, it just waits, when false, it does a lot more than just wait ?

Pull-ups configured with this test: /SS and SCK

Sample: https://code.google.com/p/jaluino/source/browse/trunk/samples/jaluino_bee_sd_card.jal

with modified SD_DELAY_AFTER_WRITE to false (in lib jaluino.jal)

Cheers,

Seb

---------- Forwarded message ----------

You might recall issues I had with some uSD cards not working with the Bee and your test SD write/read program.

Well, having purchased a single 2Gb uSD card off eBay and finding it worked and was branded ADATA I asked the Chinese vendor if I bought more, would they also be ADATA branded. He said yes, so I bought six more. However, when they arrived they were generic, unbranded cards. Three of them worked and three of them didn't work with the Bee.

Dang! So tonight while I was re-testing the three non-working cards, I removed a card after the Bee had "written" but after it got stuck and did not do any "reading", and then put it back almost immediately. The Bee then started reading the crad and flashing the LED. Hmmm. So I decided to play around with the code.

The first thing I did was add a delay after the stop_write() and before the start_read(). This made no difference. The next thing I did was add a second stop_write() immediately after the first one... this worked!

So now I turned my attention to the sd_card.jal file.

if SD_DELAY_AFTER_WRITE == TRUE then
   _usec_delay (50_000)
else
   for 50000 loop
      sd_chip_select = low  -- enable the sd card

      var byte reply
      send_command(SD_STOP_TRANSMISSION,0,reply) -- stop current transmission
      sd_ready()                                 -- wait till sd card is ready
      send_command(SD_GO_IDLE_STATE,0,reply)     -- command 0, Resets card to idle state
      sd_ready()                                 -- wait till sd card is ready

      sd_chip_select = high  -- disable the sd card

      if !(reply == 0) then
         exit loop
      end if
   end loop
end if

I first commented out the "if...else" and "end if" and ... it worked! Looking at the code, I'm not sure I understand how simply delaying for 50,000us is supposed to be a delay compared with all those commands in the else clause trapped in a 50,000 loop!

But be that as it may, the ultimate fix was to track down the SD_DELAY_AFTER_WRITE constant in the jaluino_bee.jal file and make it FALSE although I still suspect there's something not quite right about that sd_card.jal code which I don't understand because it seems to me that the else clause causes an even bigger delay!

Anyway, it now works for all the 2Gb uSD cards that I have (although I can't find the 2Gb Kingston branded one which first caused the problem more than a year ago).

Cheers,
TREV.

[mattschinkel]

Hey Guys,

 

http://justanotherlanguage.org/content/jallib/tutorials/tutorial_sd_card

 

"Some sd cards may require a 10ms (or more) delay every time you stop writing to the sd card, you can choose weather or not to have this delay. If you are doing many small writes and are worried about speed, you may set SD_DELAY_AFTER_WRITE to FALSE."

 

I had an issue with writes quite a while ago when I wrote the lib, and so I made this 50,000 delay. It seems that some cards take longer then others to do writes. 50ms is plenty long enouf to do a write. I see most of my files still have it set to TRUE. Using this should be safe as it gives the card more time, but apparently not in this case.

 

I actually wrote the FALSE part later on, and it seems to be stable now. I've been thinking of removing the TRUE.

 

So, since setting SD_DELAY_AFTER_WRITE  to FALSE solved his issue, I am guessing that the card is taking longer then 50ms to become ready which is rediculusly slow.

 

I'm not sure how long the 50,000 loop takes. The loop is not supposed to equal 50ms, it's just supposed to allow the loop to exit eventually if there is no reply. An error should be set here if no reply from the card, but I never got to that.

 

The question is... how is it exiting that loop with SD_DELAY_AFTER_WRITE = FALSE? There are 2 ways to exit. Please find out.

 

I guess this constant is hidden from jaluino users?

 

We're still waiting for high capacity cards > 2gb to work :)

 

Matt.

[Sebastien Lelong]

Hi Matt,

Thanks for your reply. What's weird is the TRUE case just wait, and FALSE case does a lot of things. I'd expect TRUE case to wait, then execute the FALSE case (that is, no "else").

In our case, setting it to false doesn't mean it waits 50ms, it just loops in the else case, performing stop_write(), etc... Having it set to true doesn't call stop_write() at all. What Trev found is calling stop_write() a second time fixes the issue.

About high capacity card, is it a matter of timing, or is it about a different protocol to implement ?

Cheers,

Seb

--
You received this message because you are subscribed to the Google Groups "jallib" group.
To view this discussion on the web visit https://groups.google.com/d/msg/jallib/-/fKVNV4T0CrIJ.
To post to this group, send email to jal...@googlegroups.com.
To unsubscribe from this group, send email to jallib+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/jallib?hl=en.

--
Sébastien Lelong

[mattschinkel]

>> Having it set to true doesn't call stop_write() at all.

sd_stop_write() does get called, in the sample.

 

a stop write is just to send 0xFD, then wait till the card is ready with sd_set_idle(), and repeats. The card should have gotten the commands the first time. SD_STOP_TRANSMISSION, SD_GO_IDLE_STATE that are in set_idle.

 

Maybe these old cards require the clock pulses or can't listen while busy, but that is not in the datasheets. These are the only reason I can think of, and would explain why 2 calls of stop_write fixes it. weird.

 

I actually think I read somewhere in the datasheets that a disable of the select line would stop any reads/writes, and after some time the card would be ready again. This is of course not working here.

 

I suggest we try going with FALSE as default, and maybe eventually get rid of TRUE.

 

High capacity needs a much different init procedure, and it's smallest addressing is a block of 512 bytes. on standard capcity you can address bytes.

 

In sd_start_read() you can see this line:

address = address * SD_BYTE_PER_SECTOR

 

This line would be commented out on high capacity.

 

I think I got the init procedure working once and received a reply from a high capacity card (one out of my 2 cards), but I never was able to read/write anything. I can look around to see if I still have that code if you want.

 

Matt.

[funlw65(Vasi)]

Guys,

Try to study the sources of this project http://www.dharmanitech.com/2011/05/microsd-atmega32-datalogger.html, specially, sd_routines.h and SD_routines.c, is not much code, and it address also SDHC.