SPI Mode 3 on Due doesn't work properly with new APIs i.e. using SPI.beginTransaction(SPISettings)

[arpita-agrawal]

When I try to set SPI in Mode 3 on Due the first write doesn't work all the time. Upon investigating more I observed that when Due is set in Mode 3 the clock polarity on the oscilloscope is wrong for the cases it doesn't work on the first write, it generates Mode 0 clock.

The issue can be reproduced by using the following sketch to write to an EEPROM, the write in the setup function doesn't write correct data when the Clock of Mode 0 is generated when the SPI_Mode is set to Mode 3. However, the data written in the loop is read correctly and printed on Serial monitor when the:

#include <SPI.h>
// Using simple SPI APIs for Due
#define CSPin 4
#define CLKRate 16000000
void setup() {
  pinMode(CSPin ,OUTPUT);
  SPI.begin();
  digitalWrite(CSPin,HIGH);
  Serial.begin(9600);
  Serial.println("SETUP");
  SPI.beginTransaction(SPISettings(CLKRate, MSBFIRST, SPI_MODE3));
  digitalWrite(CSPin,LOW);

  SPI.transfer(6);
  digitalWrite(CSPin,HIGH);
  SPI.endTransaction();
  delay(10000);

  Serial.println("SETUP:WriteData"); 
  SPI.beginTransaction(SPISettings(CLKRate, MSBFIRST, SPI_MODE3));
  byte dataWrite[8] = {2, 0, 0, 1, 3, 8, 4, 6};
  digitalWrite(CSPin,LOW);

  for(int i = 0 ; i < 7;i++)
  {
    SPI.transfer(dataWrite[i]);
  }
  SPI.transfer(dataWrite[7]);
  digitalWrite(CSPin,HIGH);
  SPI.endTransaction();
  delay(10000);

  Serial.println("SETUP:ReadData");
  SPI.beginTransaction(SPISettings(CLKRate, MSBFIRST, SPI_MODE3));
  byte dataWrite1[8] = {3, 0, 0, 0, 0, 0, 0, 0};
  byte readData[8];
  digitalWrite(CSPin,LOW);

  for(int j = 0 ; j < 7;j++)
  {
    readData[j] = SPI.transfer(dataWrite1[j]);
    Serial.println(readData[j]);
  }
  readData[7] = SPI.transfer(dataWrite1[7]);
  Serial.println(readData[7]);
  digitalWrite(CSPin,HIGH);
  SPI.endTransaction();
  Serial.println("ExitingSetup");
}

void loop() {
  // put your main code here, to run repeatedly:
  // Set BitRate to 4 MHz
  Serial.println("LOOP");
  SPI.beginTransaction(SPISettings(CLKRate, MSBFIRST, SPI_MODE3));
  digitalWrite(CSPin,LOW);

  SPI.transfer(6);
  digitalWrite(CSPin,HIGH); 
  SPI.endTransaction(); 
  delay(1000); 
  SPI.beginTransaction(SPISettings(CLKRate, MSBFIRST, SPI_MODE3));
  byte dataWrite[8] = {2, 0, 0, 1, 3, 6, 4, 5};
  digitalWrite(CSPin,LOW);

  for(int i = 0 ; i < 7;i++)
  {
    SPI.transfer(dataWrite[i]);
  }
  SPI.transfer(dataWrite[7]);
  digitalWrite(CSPin,HIGH);
  SPI.endTransaction();
  delay(1000);
  SPI.beginTransaction(SPISettings(CLKRate, MSBFIRST, SPI_MODE3));
  byte dataWrite1[8] = {3, 0, 0, 0, 0, 0, 0, 0};
  byte readData[8];
  digitalWrite(CSPin,LOW);

  for(int j = 0 ; j < 7;j++)
  {
    readData[j] = SPI.transfer(dataWrite1[j]);
    Serial.println(readData[j]);
  }
  readData[7] = SPI.transfer(dataWrite1[7]);
  Serial.println(readData[7]);
  digitalWrite(CSPin,HIGH);
  SPI.endTransaction();
  delay(10000);
}

[cmaglie]

Does the CSPin have a pull-up resistor?

During the bootstrap, just after the reset, all the pins are configured as INPUT, so it may be that the CSPin may be low for some milliseconds (confusing the eeprom).

[arpita-agrawal]

Hi,

I tried connecting a 10K pull-up to CSPin. But still the issue isn't resolved.

Thanks
Arpita