I2C - The real thing

mbalderas · Joined: 18 Apr 2013 Posts: 3

Hi,

This is the code I've been using for a while, for creating a fully functional I2C slave, where I have placed up to 8 slaves like this one with no issues.

Recently, a partner told me that there should be an easier way, where a missmatch of slave address doesn't drive the interrupt, which I told him it isn't. What do you think?

Ttelmah · Joined: 11 Mar 2010 Posts: 19499

This is down to how the hardware responds so 'read the data sheet'. You don't say 'what chip', so taking a chip at random, the 4520, the data sheet entry says:

"If the addresses match and the BF
and SSPOV bits are clear, the following events occur:
1. The SSPSR register value is loaded into the
SSPBUF register.
2. The Buffer Full bit, BF, is set.
3. An ACK pulse is generated.
4. MSSP Interrupt Flag bit, SSPIF (PIR1<3>), is
set (interrupt is generated, if enabled) on the
falling edge of the ninth SCL pulse."

Note the 'If the addresses match'.

Otherwise the chip does nothing, unless you have general call address enabled. If this is enabled, any of the addresses 0x0 to 0x10, will also be accepted.

Best Wishes

mbalderas · Joined: 18 Apr 2013 Posts: 3

Thanks for your support.

The device is a 16f690. The datasheet says that the interrupt would fire only if the address matched, but in practice it fires no matter if you are writing this device or another. My slave address is 0xAA.

How do I check for the general call address to be disabled.

Ttelmah · Joined: 11 Mar 2010 Posts: 19499

You need to read up on the difference between the SSP, and the MSSP.

Best Wishes

mbalderas · Joined: 18 Apr 2013 Posts: 3

Hi,

I actually have read the differences, and seems to me that SSP is enought for me, since I'm only writing the slave device. The master is a computer based controller, which takes care of speed of clock and sequencing.

My slaves are the issue. I mean, with the code that I posted earlier, I have had no problems, but I wish the devices didn't care whenever another address is being addressed.

[EDIT:]

Maybe... The reason of the interrupt is that the master generates a NACK if no one responds to the address being called, is my guess correct?

Ttelmah · Joined: 11 Mar 2010 Posts: 19499

The SSP, has a _firmware only_ implementation. It interrupts on every byte, requiring the code to check the flags to see what has happened.
The MSSP, implements interrupt only on address match.

Best Wishes