Add SD card library

[Kozma04]

I am just wondering if a DxCore-compatible library which allows reading and writing to an SD card will be available, just like the Arduino one (https://github.com/arduino-libraries/SD).

[SpenceKonde]

Has anyone tried it? I haven't.

[SpenceKonde]

What fails with it?

My initial attempt would be to take that one, and in Sd2PinMap.h find the line

#elif defined(__AVR_ATmega4809__) // Arduino UNO WiFI Rev2 follows

and change it to

#elif (__AVR_ARCH__ == 102 || __AVR_ARCH__  == 103 || __AVR_ARCH__ == 104)

then see where it next starts complaining.

[SpenceKonde]

Can antone wirth hardware handy install this one in place of the normal one and see what doesn't work? https://github.com/SpenceKonde/SD If it works on SPI0's default pins, try alt pin sets. Because I'm less optimistic about those/

[nabelekt]

I was able to get this working fine for my purposes. Basically did what @SpenceKonde suggested above. Changes are at https://github.com/SpenceKonde/SD/compare/master...nabelekt:get-working-for-DA128DA28.

Compiles and runs fine on the DA128DA28 with the chipSelect pin set to 7.

[sshwarts]

I also use the modified library just fine on a AVR128DB48 with SD on PF2.

[SpenceKonde]

The chip select pin shouldn't make a difference - I'm referring to the pinswaps to get the alternate sck/miso/mosi pins. I think one would need to set SPI.swap(1); before the call to SPI.begin() is made.

[SpenceKonde]

Can anyone confirm whetherthis works after swapping to an alternate pinset using SPI.swap(non-zero)? That is the thing I' am ionly 90% sure of. Ince that's confirmed, I'm gonna try to get Arduino to merge it in.

[SpenceKonde]

we need a test done with SPI.swap(SPI1_SWAP1); (SPI1 with alternate pin mapping, using the PC4~PC6 lines on a 48+ pin part).

If that doesn't work, the next step would be to see whether it's the alternate port (SPI1 vs 0) that broke it, or the alternate pinset, or if either of them will break it, SPI.swap(SPI0_SWAP_1); puts the SPI port onto PE0-PE2) on a 48+ pin DA/DB, while SPI.swap(SPI1_SWAP_0) puts it on PC0~2.

[microPaul]

I can try this on a DB64. I'll report back in a day or two.

[microPaul]

Here are some results of my SPI testing this afternoon.

I'm using a homebrew DB64 and an SD card module jumpered into the DB64's SPI0 standard pins (PA4, PA5 and PA6). Not that it matters but I'm using PA3 as chip select. I'm also doing serial download (not UPDI) on DxCore 1.4.4.

I'm running this sketch: https://www.arduino.cc/en/Tutorial/LibraryExamples/CardInfo

With a #define Serial Serial4 near the top of the sketch (because the console on my DB64 is Serial4) but I do get a warning that states `p:\Users\p\Documents\Arduino\DB64-sd-card-info-220102-01\DB64-sd-card-info-220102-01.ino:5:0: warning: "Serial" redefined

define Serial Serial4

In file included from sketch\DB64-sd-card-info-220102-01.ino.cpp:1:0: C:\Users\p\AppData\Local\Arduino15\packages\DxCore\hardware\megaavr\1.3.10\cores\dxcore/Arduino.h:425:0: note: this is the location of the previous definition

define Serial Serial0 //Error here? Check for missing ; previous line in sketch.`

be that as it may, I do get nicely formatted output on my Serial Monitor from UART4 (primary)

And I changed the chip select define to be: const int chipSelect = PIN_PA3;

The rest of the sketch is as per the website reference.

When I run the program with the SD module attached to PA4-PA5-PA6 (cs on PA3) all looks good. All the output on the console about about blocks, cluster sizes and volumes looks good. However I don't get a list of the four text files that are on the SD card. Not sure why that's missing.

I then put a SPI.swap(1); immediately before the line if (!card.init(SPI_HALF_SPEED, chipSelect)) { which looks like the first SD card command.

I recompile, serial download, and then my Serial Monitor just shows jumbled characters, with a lot of backwards ? marks, like the serial baud rate is now wrong. So it looks like SPI.swap(1) causes problems for serial.

If I change the argument of the SPI.swap function from 1 to 0 [ SPI.swap(0) ] then the serial output works correctly again.

[microPaul]

Well this is embarrassing. Of course SPI.swap(1) kills serial on my test board. I'm using PE0 PE1 pins for my console terminal (UART4 primary) and PE0 PE1 are part of the pinset for SPI0 alternate.

Let me try some other SPI alternatives that are not going to interfere with other things on my board.

[microPaul]

Ok, good news! I've tested all the SPI swap possibilities (based on the table below) on the DB64 and they all appear to work correctly.

Here's the table from: https://github.com/SpenceKonde/DxCore/blob/master/megaavr/libraries/SPI/README.md

`Pin Mapping	Pins	Parts	Swap-level name	Synonym	Value
SPI0 DEFAULT	PA4-PA7	All	SPI0_SWAP_DEFAULT	SPI0_SWAP0	0x00
SPI0 ALT1	PE0-PE3	48+ pin	SPI0_SWAP_ALT1	SPI0_SWAP1	0x01
SPI0 ALT2	PG4-PG7	DA/DB 64	SPI0_SWAP_ALT2	SPI0_SWAP2	0x02
SPI0 ALT3	PA0-1,PC0-1	DD/EA only	SPI0_SWAP_ALT3	SPI0_SWAP3	0x03
SPI0 ALT4	PD4-PD7	DD/EA only	SPI0_SWAP_ALT4	SPI0_SWAP4	0x04*
SPI0 ALT5	PC0-PC3	DD/EA only	SPI0_SWAP_ALT5	SPI0_SWAP5	0x05*
SPI0 ALT6	PC1-PC3,PF7	DD/EA only	SPI0_SWAP_ALT6	SPI0_SWAP6	0x06*
SPI1 DEFAULT	PC0-PC3	DA/DB only	SPI1_SWAP_DEFAULT	SPI1_SWAP0	0x80
SPI1 ALT1	PC4-PC7	DA/DB 48/64	SPI1_SWAP_ALT1	SPI1_SWAP1	0x84*
SPI1 ALT2	PB4-PB7	DA/DB 64	SPI1_SWAP_ALT2	SPI1_SWAP2	0x88*

`

Here are the statements I used to select the particular pins to use for SPI given the argument of the swap statement:

//SPI.swap(SPI0_SWAP_DEFAULT); // this works: SPI on PA4 PA5 PA6 SPI.swap(SPI0_SWAP_ALT1); // this works: SPI on PE0 PE1 PE2 // SPI.swap(SPI0_SWAP_ALT2); // this works: SPI on PG4 PG5 PG6 //SPI.swap(SPI1_SWAP_DEFAULT); // this works: SPI on PC0 PC1 PC2 //SPI.swap(SPI1_SWAP_ALT1); // this works: SPI on PC4 PC5 PC6 //SPI.swap(SPI1_SWAP_ALT2); // this works: SPI on PB4 PB5 PB6

Of course, only one of these statements would be uncommented for a particular test (read info from SD card).

In the case of SPI0_SWAP_ALT1 which uses PE0, PE1, PE2 (and clobbered my Serial4 console in my initial test)...for that test I changed the solder bridging jumpers on my board to make Serial3 (PB0, PB1) my console and reburned the bootloader accordingly. With the console serial port off PE0 and PE1, then the SPI port with SPI.swap(SPI0_SWAP_ALT1) worked as it should.

So, bottom line, it looks to me like all the possible SPI swap positions for the DB64 work correctly.

Also, I i didn't mention it before, but I modified Sd2PinMap.h as Spence suggested to change

elif defined(__AVR_ATmega4809__) // Arduino UNO WiFI Rev2 follows

to

elif (AVR_ARCH == 102 || AVR_ARCH == 103 || __AVR_ARCH__ == 104)

[SpenceKonde]

Wonderful, thank you! I will PR this back to the official library repo.

[SpenceKonde]

Gears are turning on that PR,

[SpenceKonde]

Unless the PR is rejected, there will be no changes made to DxCore and megaTinyCore will get it's copy of SD removed.