I2C problems, address nack

Hello Rasmus

You can refer to the following application note. Though it is for TM4C129x devices, but the addressing a slave is common between TM4C123x and TM4C129x devices. So you can find the correct address to write.

www.ti.com/.../spma073.pdf

I am using the driverlib, with the predefined read and write commands, and the read process sets the slave adress to write for write and read for reading. I have checked that this happens correctly, so the problem cant be with the "I2CMasterSlaveAddrSet" command.

Thank you though!

Hello Rasmus

There is only one place where the Slave Address is set for an access in the master nodule, If you believe that is correctly applied, then may be the address of the slave itself is in correct.

But since it does acknowledge The write commands prior to The read, this shouldn't be The case i guess? It does send 8 bytes prior to The read command.

Thanks!

Hello Rasmus

1. Are you writing the same slave address to the I2CMSA register via the I2CMasterSlaveAddrSet API call for both write and read?
2. Does the I2C Slave have different address for write and read operations (check the slave device datasheet)?

I am writing The same address, 0x24, The datasheet says 0x48 in 8 bit adress, giving 0x24 in 7 bit adress. The datasheet says write at 0x48 and read at 0x49, because of The read flag being set. I tried reading with 0x25, but that gives The same result.

Thanks!

Hello Rasmus

Do not change the I2C slave address for write and read operation. Write 0x24 to the 7-bit address with the direction value and it would work.

That is excactly what i do, but the address doesnt get acknowledged when reading. And i really dont understand why. Could it be a delay issue, even though everything runs on interrupts?

Hello Rasmus

Which slave device is this. Can you please put a link to the datasheet?

PN532

Datasheet: www.nxp.com/.../141520.pdf

Thanks!

I got my logic analyser and the signals looks good, except 2 things!

After the "Setup to write" command, and after the clock is pulled low, the sda goes high for a very short amount of time. And i cant see how that should happen. See this image:

And another problem is that between every byte send and recieved, there is a delay of around 10mS.

I am testing this up again an arduino where the PN532 works perfectly. I compare the signals of both boards.

Any ideas?

Hello Rasmus,

As long as the SCL is not high, the SDA going high is OK as it would not be sampled.

Are you using TivaWare to code the I2C access? I have used TivaWare and I see full entitlement of bandwidth even with CPU access without having delays of this order.

Yes, i use The i2cmread function, and it gives The delay. Ive gone through The code and i cant find a source of The delay anywhere.  I tried

Writing  The registers manually ane that avoided The delay, but The read didnt work.

I have actually read some of The data out From The unit now. Out of 10 bytes, i read The last 4 correctly, but I miss The first 6 bytes.

Thanks

Hello Rasmus

i2cmread is not a TivaWare API.

Thats true, it is the sensorlibrary, which is utilizing the driverlib i2c functions.

But that shouldn't be the problem, right? Or is that library too specific for the sensors in the boosterpack?

Hello Rasmus,

I2CMRead and I2CMWrite are functions called by the respective sensor driver file and it handles the statemachine for the sensor. If you need to add a new sensor then I would suggest writing the sensor driver file.

The other option is to use driverlib approach as explained in the application note.

www.ti.com/.../spma073.pdf