I2C Init

[graemec]

Hi,
I have an instance of the I2C init macro hanging the program in HW. I want to perform a write to an external I2C eeprom 24LC256 but it doesn't get past the I2C init macro. Everything works ok in the simulator. When should the init macro be run? Right at the start of the program or just before the I2C start macro? Any ideas? I'm running version 4.2.3.58 and Im using a 18f4620 @ 20MHz. Thanks
PS If I run just the full I2C routine separate from the main program it works in HW also....

G

[Benj]

Hello,

You should be able to call the init function at any time without causing lockups.

Are you using the hardware or software I2C?

Also do you have external pull up resistors connected to the 2 signal lines? These will probably need to be around 4.7K.

[graemec]

Hi Benj
I do have the external pull ups at exactly 4k7 and I'm using the onboard I2C of the 18F4620.
I've experimented a little...
1) To make things simpler and to make sure I had no stack overflow issues I exported my I2C eeprom write macro from the main program and added some markers so I'd know where it was hanging and recompiled.
2) Ran the program and as before it got stuck at the I2C init component macro.
3) Moved the I2C init component macro to the very start of the main program and it now it does not lockup.

So it would appear that it does make a difference where you call the I2C init.

Any ideas on understanding what's causing the lock up?

G

[Benj]

Looking at the code there is nothing in the init function that could be blocking the code from running.

In the Start function there is however a while command that could potentially cause a block if the sspcon2 register is not updated correctly or becomes corrupted.

Maybe try switching to the software (bitbanged) mode if the lockups become a recurring problem.

Migght be worth putting your program onto the forum to see if we can replicate the problem here and get to the route of what is going wrong.

[graemec]

Hi Ben,
Here are two files. One works fine, the other locks up at the I2C init macro. The only difference is the position of the I2C init component macro. Both run ok in the simulation.
dm4_i2c_4620_lockup.fcf is the one with the lockup.
I'm using a 18F4620 @ 20MHz with a 24LC256 connected directly to the controller with 4k7 pullups on the SDA and SCL lines.
Thanks
Graeme

dm4_i2c_4620_lockup.fcf

(14.46 KiB) Downloaded 263 times

dm4_i2c_4620_test.fcf

(14.36 KiB) Downloaded 275 times

[graemec]

Hi Ben,
Have you looked into this I2C init issue any further?
Since my last post I've tried a different controller (18F4520), different memory (24LC02), different clock (4MHZ) and a different board.
In all cases I'm able to cause problems by moving the I2C init macro (corrupt data or lockups)
The place the I2C init Macro seems to work best ( doesn't lockup) is if it is the very first macro of the main program but even then something is not quite right.
In my application the first time I call a page read of the I2C eeprom it works ok but subsequent attempts to call the read routine either for the same or another memory address of the I2C eeprom cause a lockup of the controller this time at the I2C start Macro. FYI writing to the I2C eeprom works fine without problems.
Is there anything in the chip config that would cause problems?
Please let me know something.
Graeme

[Benj]

Hello Graeme,

Sorry I've not had chance to look at it yet I spend my weekend concentrating on the bug in the FAT16 component.

I don't think this is a configuration problem it must just be a set of specific events thats go together to form a crash. I will try and have a look at the cause of the problem as soon as I get some free time.