MSP432 I2C Master Problem

I forgot to say that the SDA/SCL screenshot came from the
MAP_I2C_masterSendStart(EUSCI_B3_MODULE);
and the second one
MAP_I2C_masterSendMultiByteNext(EUSCI_B3_MODULE, SLAVE_ADDRESS);
didnt send any messages out.

The first masterSend sends the Start and Slave Adress bit, the second one was only for testing

Hi Robert,

what am i doing wrong with the MSP432 I2C?

Becouse if i shift the Device Adress it Acknowledge, but if i dont it doesnt.

So here you can see it works, but i send 9 Clocks and the Last one muss be ACK/NACK but is not! it is 0xD0 >> 1 == 0x6A

Here you see the problem that i am facing, i am sending 0x72:

The first Clock is Empty and the follow 8 Clocks are for the Data. But what i understood the last clock should be for ACK/NACK?

Should i shift it to the right 0x72 << 1 ?

Why do i have keep shifting the SDA Data?

Is it a setting problem like MSB/LSB or Risging/Falling Edge? 

Thanks in advance,

Michael

I think you might be misreading the scope traces. The first image shows a "start" condition followed by the eight bit value 0xD0. That corresponds to a 7-bit slave address of 0x68 and a zero R/W bit to select write mode. After that the clock line stays low, presumably because the slave is clock‑stretching.

The next image begins with the slave releasing the clock line while holding the data line low to acknowledge its address to the master. That's immediately followed by the eight data bits 0x72 from the master, and again the slave performs clock-stretching. The acknowledgement for the 0x72 and the stop or restart condition will have happened later (not shown on the trace).

With I²C there are no options for MSB/LSB or rising/falling edge (or clock polarity) like there are for SPI. Also it's only the slave address that might require a 1-bit shift to convert between 8-bit addresses with the R/W bit already appended and 7-bit addresses without it. There's no need to shift data bytes before passing them to the eUSCI module.

This is interesting because I am having the very same problem that Michael describes.  I have a device at address 0xD4, but it doesn't give me a ACK unless I address 0x6A.   I think there is a problem with 10 bit addresses and Driverlib.  If I use a 7 bit address it reads correctly in my logic analyzer, but if I do a 10 bit address, it should give me a 0b1110 just after the start bit, to indicate this is a 10 bit address.    Michael, did you figure anything out?

Thanks,

Rob

Hi Rob,

sorry no it didnt got it to work. I had to shiftet it too.

Thanks Michael, and too bad!  I guess I'll have to write a fix for driverlib.  The docs say it should work, but it doesn't send the right starting sequence for 10 bits, and when we shift things over, I think it works only because that makes our 10 bit address look like a 7 bit address, and we were lucky that our address was even, because I don't think it wouldn't have worked if it was odd.  Again, thanks for the answer.  If I get it going, I'll post back here with my solution.

Thanks,

Rob

One thing I just noted is that the May 2015 Driverlib Users Guide says it only works for 7 bit addresses, but the November 2015 edition says it works for 7 and 10 bit addresses. I think I have the latest MSPWare (3_20_00_37), because when I check for updates, there are note, but the docs seem to indicate that this might have been a change somewhere along the line last year.

Hello Sir,

I am using MSP432P401R launchpad and trying to communicate FRAM 24CL04 using I2C.

Working on example code "i2c_master_rw_repeated_start-master_code.c", slave address of my device is 0xA0

but by shifting it 0xA0>>1 i.e 0x50 does not work(shows SDA low & SCL high). And if I pass slave address 0xA0, it shows waveform of

-> 1st start bit(high to low transition) then 0x40.

Pls suggest solution for this problem.

Hi Supriya,
looks like TI Driver shifts the slave address for you. Can you plot the digital logic when you set the address at 0xA0?
Is the SDA still low or are both high after the communication?

Hello Sir,

Please find below digital logic, if I set slave address - 0xA0.

CH1 (Yellow) - -> SCl

CH2 (Blue) - -> SDA

Slave address are 7bit long normally and the first bit is for READ/WRITE commands.
So if your slave address is 0xA0 than you will have to shifted it so would be 0x50.
The first SCL clock is the start bit followed from 7 slave address bits and after that the next clock is READ/WRITE and last clock is aknowledge.
The Master hold the SDA low until the slave with that corresponded address pulls it back up. There is no slave with 0x80 on the bus so shift the addres 0xA0 to 0x50 and please add for us the plot for that here too.

Hello SIr,

By setting slave address 0x50 instead of 0xA0, I am not getting any pulse on  SCL(remains high continuously) and SDA (remains low cont.).

Using 4.7k pull ups in H/w  for SCL & SDA.

Please help to solve this issue.

Hi Supriya,

you have to power cycle your board so SDA and SCL are High again. 

If they are not High at init, you may have a hardware problem or the 4k7 pull ups by SDA are not solded well.

Make sure you put a break point where you are sending the slave address and debug it until there and see if SCL and SDA are still high.

Then go over the i2c command and check if it accept the slave address or not. If not SDA will be put low.

Hi Sir,

Now It is been observed that, I am getting pulses only at power on, further  SDA becomes low  and SCL becomes continuously high.

See below plot getting only at power on(for few seconds).  why is this happening?

CH1-SDA

CH2-SCL

The slave address looks good now.
What are you sending on i2C read or write? which register value you want to read/write?
Can you screen shot channel 1 same level as channel 2 and only the first command.

I am transmitting multiBytes ->

TXData[10] = {0x00, 0x3A, 0x1E, 0x39, 0x1C, 0x79, 0x5D, 0x6C, 0x0C, 0x01};

in my code. But my aim is to Read/Write Float value at one location in external FRAM.

I have still not understood why the waveform is visible only at power on for few seconds??

well this depens on your code.
The I2C communication seems to work, the slave responds to its master.
You could you simplify your code for first and write to the slave in the 0x00 register this value 0x3A.
After that read the register 0x00 and see if you get 0x3A back.
If you want to write 10 values from TXData[10] you will need to say which register should each value be writting in.
Say you start at register 0x00 and you could just increment each register write commands at one.
Resgiter 0x00 would have value TData[0]
Resgiter 0x01 would have value TData[1]
Resgiter 0x02 would have value TData[2]
Resgiter 0x03 would have value TData[3]
.....
and so on.
By reading the values you read the proper register entry.

Yes Sir, I came to know where i was making mistake in my code.
Now I am getting waveform throughout.
Till now my code was just transmitting data,Now I will try to read data from slave.
Thank you So much for the support.

Hi Sypriya,

you are welcome. Next time you have an issue, just open a new question on the forum so its better for others to track and reply.

regards,

michael

OK Sir, Thanks :)

Hello Sir,
I had created new thread for my issue - I2c Master reception issue. Will you please help me out.