INA229: INA229 current sensor reading Manufacture ID

Hello Andrew,

Thank you for using the TI forum.

About the polarity and phase, usually it is that polarity = 0 and phase = 1, however some MCU architectures define polarity and phase differently causing the values to be different. The important thing is that the data lines line up with the clock per the datasheet description:

You appear to be getting the right values, but maybe it is a format/parsing issue? I recommend looking at the example code from SysConfig: https://dev.ti.com/sysconfig/index.html?product=ascstudio&module=/ti/sensors/currentsensor/INA229. Hopefully this helps you with your firmware issue. If not, message here again and hopefully we can find the issue.

Kang Andrew said:

for(i = 0; i < 2; i++)
{
// byte[i] = SpiaRegs.SPIRXBUF;
byte[i] = tmp_rx_data16[i] = SpiaRegs.SPIRXBUF;
}
return 0;
}

I'm not familiar with that MCU but the code above looks suspicous to me. Shouldn't you wait for new data to arrive between those back-to-back reads of SPIRXBUF? There's probably a companion function to spi_xmit() which reads data from SPI.

Hey Andrew, 

I'm adding the TMS320F28377D team to this thread to help with the MCU specific SPI question.

Hello,

A few observations just looking at the code and provided information:

The SPI is configured for 16-bit characters which will force 16 bits to be transmitted/received on each write to the SPI TX buffer. However, I can't tell from the scope plot if 16 bits are being transmitted in each chip select frame since the SPICLK is not included.

Kang Andrew said:

spi_xmit (tx_data); // Send
// while (SpiaRegs.SPIFFRX.bit.RXFFST != 1) {}

for(i = 0; i < 2; i++)
{
// byte[i] = SpiaRegs.SPIRXBUF;
byte[i] = tmp_rx_data16[i] = SpiaRegs.SPIRXBUF;
}
return 0;
}

This code doesn't look right to me. Since the SPI is configured for 16-bit characters, only a single16-bit read should be done for each SPITXBUF write, not two. The code checking the status of RXFFST should also not have been commented out, otherwise you risk reading SPIRXBUF before there is any valid data in the SPI read FIFO.

If you can, please also include the code for spi_xmit(). This function should be writing a single 16-bit character to SPITXBUF.

Ah good point, the INA device does send data in 8 bit chunks, so you they may need to configure it for 8-bit reading then read twice.

Mitch M said:

they may need to configure it for 8-bit reading then read twice

Mitch, correct. For reading manufacturer ID register I would recommend the following:

• Configure MCU SPI for 8-bit character size
• Write byte address (SPI simultaneously reads in 1 byte)
• Wait if RXFFST != 1
• Read trash byte from SPIRXBUF and discard
• Write dummy byte (SPI simultaneously reads in 1 byte)
• Wait if RXFFST != 1
• Read byte from SPIRXBUF, this is your MSB of your manufacturer ID reg
• Write dummy byte (SPI simultaneously reads in 1 byte)
• Wait if RXFFST != 1
• Read byte from SPIRXBUF, this is your LSB of your manufacturer ID reg

Kang Andrew said:

I was able to read the correct value by reading and writing with 8 bits.

I did forget to mention that in 8-bit configuration, you will need to shift the byte address << 8 before writing to the SPITXBUF. It looks like you may have figured that out already though. This is because the SPI will always shift out the MSB of the SPITXBUF first. During reads from SPIRXBUF, no shifting is necessary.

• Configure MCU SPI for 8-bit character size
• Write (byte_address << 8) to SPITXBUF (SPI simultaneously reads in 1 byte)

Kang Andrew said:

It seems that it was not possible to read and write in 16-bit.

It looks like the INA will expect a minimum of 8+16 clocks during any SPI transaction. If you configure the SPI for 16-bit characters and try to read the INA ID registers, then you will need to do two writes to the SPITXBUF which will generate 32 clocks which is more than what the INA expects. Since the INA SPI frame can vary between 8+16 up to 8+40 bits, it is better/easier to configure the SPI for 8-bit data.

Kang Andrew said:

In addition, may I ask a question about reading current?

I would recommend starting a new thread.