ATmega328 + SPI闪存

Question

前言：这个问题是关于我正在与大学教授合作的一个项目。这不是为了成绩，但我在这位教授中的声誉确实很重要。因此，虽然我在这个项目上的成功对我来说很重要，但我并不认为向 Stack Overflow 寻求帮助是不公平的。

也就是说，这是我的项目的高级概述。我有一个 ATmega328 微控制器。我有一个 Microchip SST 64 Mbit 闪存卡。 ATmega 有 SPI 的硬件实现。闪存具有 SPI 的硬件实现。

我的目标是在 SPI 主模式下使用 ATmega 从闪存芯片读取数据并向闪存芯片写入数据。内存以多重覆盖结构组织，这很适合擦除，但就我的目的而言，它基本上只有 32,768 页，每页 256 字节。

要写入数据，基本思想是发送一个指令字节，然后发送起始地址，然后发送数据。要读取数据，基本思想是发送一个指令字节，然后发送起始地址，然后发送一个虚拟字节，然后它开始向我发送数据。

以下是数据表：

微控制器：http:// www.atmel.com/dyn/resources/prod_documents/doc8271.pdf

Flash：http: //www.sst.com/dotAsset/40498.pdf

代码：

#include <SPI.h>
#include <Peggy2.h>

#define SS_PIN  16

Peggy2 frame1;
byte toDisp = 0;
byte checker = 0;

void setup()
{
    frame1.HardwareInit();
    pinMode(SS_PIN,OUTPUT); //set pin16 to output, SS pin
    SPI.setClockDivider(SPI_CLOCK_DIV2); //set the SPI clock to f/2, fastest possible
    SPI.begin();    //SPI lib function which sets ddr for SCK and MOSI pin
                    //MISO is auto input
                    //see SPI.cpp for more info

}

void loop()
{

    if(!checker){
                enableProgramming();
        programData();
        toDisp = receiveByte(0);
        checker = 1;
        frame1.WriteRow(0,toDisp);
    }
    frame1.RefreshAll(2);

}

byte receiveByte(unsigned long startAddress)
{
    //Begin High Speed Read Instruction
    //See p. 10 of SST data sheet
    digitalWrite(SS_PIN,LOW);
    SPI.transfer(0x0B); //high speed read instruction
    SPI.transfer(0x00); //next 3 transfers are address bits A32 - A0
    SPI.transfer(0x00); //So this will read the first byte on the chip
    SPI.transfer(0x00); //last address bits
    SPI.transfer(0xFF); //dummy byte is required to start sending data back to uP
    SPI.transfer(0xFF); //I'm hoping that if I transfer a bullshit byte, the flash
                        //chip will transfer it's data to me in the same time
    digitalWrite(SS_PIN,HIGH);
    //End High Speed Read Instruction   
    return SPDR;    
}

//will perform the read instruction starting from
//startAddress and will receive numOfBytes bytes in
//succession
void receiveBytes(int numOfBytes, unsigned long startAddress)
{
    digitalWrite(SS_PIN,LOW);
    SPI.transfer(0x0B);//high speed read instruction

}

//will perform:
// 1) Chip Erase
// and loop through:
// 1) Page Program
// 2) increment Page
//until the data has finished **note this can loop and over write beginning of memory
void programData(){
    //Begin ChipErase Instruction
    //See p. 17 of SST data sheet
    digitalWrite(SS_PIN,LOW);
    SPI.transfer(0x60);//chip erase instruction
    digitalWrite(SS_PIN,HIGH);
    delay(50);//spec'd time for CE to finish
                //don't bother polling because time to program is irrelevant
    //End ChipErase Instruction

        //Begin WREN Instruction
    //See p. 18 of SST data sheet
    digitalWrite(SS_PIN,LOW);
    SPI.transfer(0x06);//write enable instruction
    digitalWrite(SS_PIN,HIGH);
    //End WREN Instruction

    digitalWrite(SS_PIN,LOW);
    SPI.transfer(0x02); //page program instruction
    SPI.transfer(0x00); //first 8 address bits
    SPI.transfer(0x00); //2nd 8 address bits
    SPI.transfer(0x00); //3rd 8 address bits
    SPI.transfer(0xAA); //10101010 is the byte I should be writing
    digitalWrite(SS_PIN,HIGH);
    delayMicroseconds(3000); //wait 3 ms for page program


    /*
    //Begin Page-Program Instruction
    //see p. 13 of SST data sheet
    byte firstAddress = 0;
    byte secondAddress = 0;
    //this loop will write to every byte in the chips memory
    //32,768 pages of 256 bytes = 8,388,608 bytes
    for(unsigned int i = 0; i < 32,768; ++i) //long variable is number of pages
    {
        digitalWrite(SS_PIN,LOW);
        ++secondAddress; //cycles from 0 to 255, counts pages
        firstAddress = i>>8; // floor(i/256)

        SPI.transfer(0x02);//Page-Program instruction byte
        SPI.transfer(firstAddress); //increments every 256 pages i.e. at page 256 this should be 1
        SPI.transfer(secondAddress); //increments every 256 bytes, i.e every page
        SPI.transfer(0x00); //beginning of a page boundary
        for(int j = 0; j < 256; ++j) //number of bytes per page
        {
            SPI.transfer(2program[(256*i) + j]);//data byte transfer            
        }
        digitalWrite(SS_PIN,HIGH);
        delayMicroseconds(2500); //2500us (2.5ms) delay for each page-program instruction to execute
    }
    //End Page-Program Instruction
    */
}

//Will prepare the chip for writing by performing:
// 1) arm the status register
// 2) Write Enable instruction
//Only needs to be performed once!
void enableProgramming(){
    //Begin EWSR & WRSR Instructions
    //See p. 20 of SST data sheet for more info
    digitalWrite(SS_PIN,LOW); //lower the SS pin
    SPI.transfer(0x50); //enable write status register instruction
    digitalWrite(SS_PIN,HIGH); //raise the SS pin
    delay(10);
    digitalWrite(SS_PIN,LOW); //lower the SS pin
    SPI.transfer(0x01); //write the status register instruction
    SPI.transfer(0x00);//value to write to register
                //xx0000xx will remove all block protection
    digitalWrite(SS_PIN,HIGH);
    //End EWSR & WRSR Instructions

    //Begin WREN Instruction
    //See p. 18 of SST data sheet
    digitalWrite(SS_PIN,LOW);
    SPI.transfer(0x06);//write enable instruction
    digitalWrite(SS_PIN,HIGH);
    //End WREN Instruction

}

所以这应该是一个测试程序，它将 1 个字节编程到闪存上，然后读回并显示我拥有的 LED 阵列上的字节。如果您对 LED 阵列感兴趣，可以在这里找到：http://evilmadscience.com/tinykitlist/157

我相信我的读取功能有效，因为我第一次运行此功能时，所有 8 个 LED 都亮了。这对我来说表明它在闪存处于全 1 的出厂状态时读取了闪存。现在显然我把一些东西搞砸了，因为点亮的字节与我试图编程的字节根本不对应。

我还应该注意，我正在使用 Arduino 的默认 SPI 库，并且帧缓冲区功能可以正常工作。当我执行frame1.WriteRow(toDisp)时，它工作正常并且已经过广泛测试。

如果有人有时间或耐心找出我做错了什么，那就太棒了。

编辑：帮助调试：
LED 由同样使用 SPI 接口的驱动芯片驱动。我没有写那部分代码。在示波器上我可以看到 SCK 线是由那部分代码驱动的。然而，我在 MOSI 引脚上也有一个探针，如果我不点亮任何灯，它似乎永远不会变高。对我来说，这意味着我没有正确发送信息。又名...也许我的 SPI.transfer() 需要启用功能或其他功能？

Answer 1

对于仍然好奇的人来说，问题在于存储芯片对缓慢的上升时间极其敏感。安装施密特触发器后，一切正常。