Posted on May 17, 2011 at 14:09I use the code below to write to an LTC4151. Your application must be similar.
void LTC4151WriteBuffer(uint8_t* pBuffer, uint8_t NumByteToWrite)
{
/*!< While the bus is busy */
while(I2C_GetFlagStatus(I2C1, I2C_FLAG_BUSY))
{
}
/*!< Send START condition */
I2C_GenerateSTART(I2C1, ENABLE);
/*!< Test on EV5 and clear it */
while(!I2C_CheckEvent(I2C1, I2C_EVENT_MASTER_MODE_SELECT))
{
}
/*!< Send address for write */
I2C_Send7bitAddress(I2C1, LTC4151Address, I2C_Direction_Transmitter);
/*!< Test on EV6 and clear it */
while(!I2C_CheckEvent(I2C1, I2C_EVENT_MASTER_TRANSMITTER_MODE_SELECTED))
{
}
//send the reg select COMMAND byte
I2C_SendData(I2C1, 0);
/*!< Test on EV8 and clear it */
while(!I2C_CheckEvent(I2C1, I2C_EVENT_MASTER_BYTE_TRANSMITTED))
{
}
/*!< While there is data to be written */
while(NumByteToWrite--)
{
/*!< Send the current byte */
I2C_SendData(I2C1, *pBuffer);
/*!< Point to the next byte to be written */
pBuffer++;
/*!< Test on EV8 and clear it */
while (!I2C_CheckEvent(I2C1, I2C_EVENT_MASTER_BYTE_TRANSMITTED))
{
}
}
/*!< Send STOP condition */
I2C_GenerateSTOP(I2C1, ENABLE);
}
The fact you could fix your problem with just 5 NOPs implies a hardware timing restriction which might just be how long it takes the I2C ''engine'' to clock through to the next state.
(Our posts crossed - pleased to hear it's now working.)
