PIC18F46Q71 and I2C

I have been caught up in an I2C nightmare most of the week.  Basically a few years ago Microchip change their implementation of the I2C controller.  Instead of the MSSP which did both SPI and I2C, they now have separate SPI and I2C controllers.  Now it is true that the MSSP probably did not give as much control over the protocol process as one would like.  I am sure the MSSP worked great with all of Microhip's I2C products.  But as with any "specification" it is subject to interpretation, which probably led to a less than ideal mating with other companies products.  The new controller has more settings for SMBUS which is slightly different than I2C in timing and voltage levels.  And it has a more automated mode of working.  It may be better and more flexible. I dont know.  It wasn't that easy to get working in the master state. 

But the PICs I am moving to have this new controller.  There are very few working examples and the firmware that the MCC generator gives you is buggy and hard to understand how to actually implement. It is mostly a cooperative multi-tasking implementation with switch statements and lots of function calls to set one bit in a register.  Now I2C was never straight forward as a protocol, but the MCC code seems to make it more complex than it needs to be.  Sometimes you just want to read an external EEPROM memory and do a simple task.  Not have 10 peripherals running that need to be serviced on a regular basis.

What I found when searching the Internet for working example code was this example on a PIC18F46K42 and a Curiosity HPC.  Since I was working on an HPC this was perfect.  I downloaded the project and started working.  Using my Digilent Analog Discovery 2 as an I2C protocol analyzer, I could see when it worked and when it didn't.   Then I used the driver (.c) and header (.h) files as the starting point for my driver/header files.  Kept modifying these files in the example project until I could read and write the I2C EEPROM I had installed in one of the Click Module sockets on the HPC.  At this point I moved the files back to my project and the PIC18F46Q71 and started testing.  Some minor tweaks to the initialization were needed, but it seemed to work just fine.  

The I2C driver I have running is very basic and runs to completion.  That is it is blocking and will hang the system.  But for now that is OK.  I will need to implement some form of cooperative multi-taksing to get the LP5569 to work, since that is the way it is setup right now.   My only issus is when reading the EEPROM.  The sequence is to write the client address followed by the two bytes that hold the memory address.  this is followed by a restart condition and the PIC resends the client address with read set and starts reading the memory.  Here is the protocol analyzer's view of that sequence.

and just the restart area

The data sheet implies this is OK.  The stop bit is not long enough and all the client IC sees is the restart condition.  We shall see.

 I will leave with this comment form another developer, which was pretty typical of those who posted.

"When you make it hard to use your products, developers will move to other products."