I2C Repeated start

[lpcware]

Content originally posted in LPCWare by bobi-one on Mon Oct 27 06:41:33 MST 2014
I Am trying to interface an  accelerometer with LPC1549 and trying to use the LPCOpen library.
Reusing the example for periph_i2cm_interrupt. I noticed that my chip requires repeated start when reading from it.
How can i modify  SetupXferRecAndExecutein order to do such thing?

static void SetupXferRecAndExecute(uint8_t devAddr,
   uint8_t *txBuffPtr,
   uint16_t txSize,
   uint8_t *rxBuffPtr,
   uint16_t rxSize)
{
/* Setup I2C transfer record */
i2cmXferRec.slaveAddr = devAddr;
i2cmXferRec.status = 0;
i2cmXferRec.txSz = txSize;
i2cmXferRec.rxSz = rxSize;
i2cmXferRec.txBuff = txBuffPtr;
i2cmXferRec.rxBuff = rxBuffPtr;

Chip_I2CM_Xfer(LPC_I2C0, &i2cmXferRec);
/* Enable Master Interrupts */
Chip_I2C_EnableInt(LPC_I2C0, I2C_INTENSET_MSTPENDING | I2C_INTENSET_MSTRARBLOSS | I2C_INTENSET_MSTSTSTPERR);
/* Wait for transfer completion */
WaitForI2cXferComplete(&i2cmXferRec);
/* Clear all Interrupts */
Chip_I2C_ClearInt(LPC_I2C0, I2C_INTENSET_MSTPENDING | I2C_INTENSET_MSTRARBLOSS | I2C_INTENSET_MSTSTSTPERR);
}

[lpcware]

Content originally posted in LPCWare by 1234567890 on Sun Nov 02 07:18:41 MST 2014

---

Quote:

I believe the functions work and all the problems are cause because of the data coming from the sensor.

---

Well, of course Sunday is the day to believe, but when your parameter is definately wrong...
And 'data_write' and '&data_write[0]' is the same in C, just for clarification. Have a look at the C-book of your choice; chapter 'arrays and strings' or so.
Using an address instead of a value cannot 'work' (and vice versa).

[lpcware]

Content originally posted in LPCWare by bobi-one on Fri Oct 31 01:34:24 MST 2014

---

Quote: 1234567890

---

Quote: bobi-one

  static void MPU9150_writeByte(uint8_t address, uint8_t subAddress, uint8_t data)
{
   char data_write[2];
   char data_read[2];
   data_write[0] = subAddress;
   data_write[1] = data;
   SetupXferRecAndExecute(address, &data_write, 2, data_read,1);
}

  static char MPU9150_readByte(uint8_t address, uint8_t subAddress)
{
    char data[1]; // `data` will store the register data     
    char data_write[1];
    data_write[0] = subAddress;
    SetupXferRecAndExecute(address, &data_write,1, data, 1);
    return data[0]; 
}

  static void MPU9150_readBytes(uint8_t address, uint8_t subAddress, uint8_t count, uint8_t * dest)
{    
    int ii;
    char data[14];
    char data_write[1];
    data_write[0] = subAddress;
    SetupXferRecAndExecute(address,&data_write,1,data, count);
    for(ii = 0; ii < count; ii++) {
     dest[ii] = data[ii];
    }
} 
 

---

And no errors or warnings from the compiler? At the first glance
&data_write
is wrong; must be only
data_write
because data_write is an array like data_read.

---

Oh, i just noticed those ampersands ( there was even a warning).
I believe the functions work and all the problems are cause because of the data coming from the sensor. 
Have somebody else worked with MPU9150? Because the driver I am familiar with is for linux kernel and it is hard to interface just like that?

Regards,
Boyko

[lpcware]

Content originally posted in LPCWare by LabRat on Wed Oct 29 19:32:19 MST 2014
It's not very difficult to use own functions with this sample...

As sample I've added EEPROM access functions.

//EEPROM access
#define BUFF_SIZE 64
uint8_t tx_buff[BUFF_SIZE];
uint8_t rx_buff[BUFF_SIZE];
uint8_t tx_data[BUFF_SIZE];
uint16_t address;
uint16_t len;

//read bytes from address
static int read_address(uint16_t address, uint16_t bytes)
{
 tx_buff[0] = (address >> 8) & 0xFF;
 tx_buff[1] =  address & 0xFF;
[color=#f00] SetupXferRecAndExecute(I2C_ADDR_7BIT, &tx_buff[0], 2, &rx_buff[0], bytes);[/color]
 if (i2cmXferRec.status == I2CM_STATUS_OK)
 {
  return 1;
 }
 return 0;
}

//write bytes to address
static int write_address(uint16_t address, uint16_t bytes)
{
 tx_buff[0] = (address >> 8) & 0xFF;
 tx_buff[1] =  address & 0xFF;
 memcpy(&tx_buff[2],&tx_data[0],bytes);
 [color=#f00]SetupXferRecAndExecute(I2C_ADDR_7BIT, &tx_buff[0], bytes+2, &rx_buff[0], 0);[/color]
 if (i2cmXferRec.status == I2CM_STATUS_OK)
 {
  return 1;
 }
 return 0;
}

//write data
tx_data[0] = 0x12;
tx_data[1] = 0x34;
tx_data[2] = 0x56;
tx_data[3] = 0x78;
address = 0x10;
len = 4;
ret = write_address(address,len);
printf("Write Data to Address  0x%04X-",address);
if(ret)printf("Success ");
else printf("Error ");
for(i=0; i < len; i++){printf("%02X ",tx_data);}
printf("\n");

//read data
address = 0x10;
len = 4;
ret = read_address(address,len);
printf("Read Data from Address 0x%04X-",address);
if(ret)printf("Success ");
else printf("Error ");
for(i=0; i < len; i++){printf("%02X ",rx_buff);}
printf("\n");

This EEPROM sample is accessing data bytes with a [color=#f00]2 byte address[/color].

I think MPU-9150 is accessing registers with 1 byte.
So a register write function should transmit 2 bytes (register_address + data) and receive 0.
A register read function should transmit 1 byte (register address) and receive 1 byte (data).
Usually SetupXferRecAndExecute is doing everything what's necessary, you just have to fill the buffers and use the correct I2C 7bit address...

[lpcware]

Content originally posted in LPCWare by 1234567890 on Wed Oct 29 13:01:39 MST 2014

---

Quote: bobi-one

  static void MPU9150_writeByte(uint8_t address, uint8_t subAddress, uint8_t data)
{
   char data_write[2];
   char data_read[2];
   data_write[0] = subAddress;
   data_write[1] = data;
   SetupXferRecAndExecute(address, &data_write, 2, data_read,1);
}

  static char MPU9150_readByte(uint8_t address, uint8_t subAddress)
{
    char data[1]; // `data` will store the register data     
    char data_write[1];
    data_write[0] = subAddress;
    SetupXferRecAndExecute(address, &data_write,1, data, 1);
    return data[0]; 
}

  static void MPU9150_readBytes(uint8_t address, uint8_t subAddress, uint8_t count, uint8_t * dest)
{    
    int ii;
    char data[14];
    char data_write[1];
    data_write[0] = subAddress;
    SetupXferRecAndExecute(address,&data_write,1,data, count);
    for(ii = 0; ii < count; ii++) {
     dest[ii] = data[ii];
    }
} 
 

---

And no errors or warnings from the compiler? At the first glance
&data_write
is wrong; must be only
data_write
because data_write is an array like data_read.

[lpcware]

Content originally posted in LPCWare by LabRat on Wed Oct 29 08:52:28 MST 2014
I2C protocol of MPU-9150 doesn't look exotic  :)

Would suggest to [color=#f00]read[/color] WHO_AM_I register (Type: Read Only)...

[lpcware]

Content originally posted in LPCWare by NXP_Paul on Wed Oct 29 08:28:22 MST 2014
The fact that you're getting values back from the slave device, and there are no I2C errors reported is a good sign.
Now the question is whether the software is corrupting the values, or is the data on the bus not what you really expected.
You really need to get a scope and observe what is being transmitted on the bus.  I prefer a scope to a logic analyzer for looking at the I2C bus since it provides much more detail.

Paul

[lpcware]

Content originally posted in LPCWare by bobi-one on Wed Oct 29 07:52:20 MST 2014
Hi,
Writing to the WHO_I_M register of the slave doesn't return the expected value( only a couple of times actually did), so i presume the communication with it is not  working correctly. I did not observe any error conditions from the I2C .
Regards

[lpcware]

Content originally posted in LPCWare by NXP_Paul on Wed Oct 29 07:08:16 MST 2014
The attached sequences are standard I2C communications, so these should not cause any issues.
What do you mean that your code is failing on simple WHO_AM_I tests? I am not sure I understand what you are trying to accomplish.  I assume you are trying to read back a specific register, and receive a specific response, but please confirm.
What is the result returned by the I2C functions (is there a NACK, or other error conditions)?

Paul

[lpcware]

Content originally posted in LPCWare by bobi-one on Wed Oct 29 05:52:23 MST 2014
Thanks for the reply Paul,
Those are the functions that i wrote. But it seems the data that is coming  isn't quite as expected ( I cannot put logic analyser at the moment to observe the communication)

  static void MPU9150_writeByte(uint8_t address, uint8_t subAddress, uint8_t data)
{
   char data_write[2];
   char data_read[2];
   data_write[0] = subAddress;
   data_write[1] = data;
   SetupXferRecAndExecute(address, &data_write, 2, data_read,1);
}

  static char MPU9150_readByte(uint8_t address, uint8_t subAddress)
{
    char data[1]; // `data` will store the register data     
    char data_write[1];
    data_write[0] = subAddress;
    SetupXferRecAndExecute(address, &data_write,1, data, 1);
    return data[0]; 
}

  static void MPU9150_readBytes(uint8_t address, uint8_t subAddress, uint8_t count, uint8_t * dest)
{    
    int ii;
    char data[14];
    char data_write[1];
    data_write[0] = subAddress;
    SetupXferRecAndExecute(address,&data_write,1,data, count);
    for(ii = 0; ii < count; ii++) {
     dest[ii] = data[ii];
    }
} 
 

I have attached the communication sequence required by the slave device. But my code is failing on simple WHO_AM_I reading ( returns different values).
Any suggestions how to fix my application?

[lpcware]

Content originally posted in LPCWare by NXP_Paul on Mon Oct 27 08:43:29 MST 2014
Hi
Generally, you only need to specify a value greater than zero for rxSize (i2cmXferRec.rxSz = rxSize).  This should provide the repeated start between the tx and rx messages.

Paul