>I didn't see anything obviously wrong with your code, but I did not look
>closely at it. Some general things though:
I've added some delay's at a few points and changed the start condition...
Now it is working OK.

>Remember that I2C is a multidrop interface which should have a passive
>pull-up to generate logic 1's on both clock and data. In practice if you
>only have a single slave that does not clock stretch (EEPROMs in general
>don't) you can drive the clock high and low actively.

Don't know where I read this but it was say that you should use a 10K
resistor for 100Khz and 2k2 for a 400Khz bus.

>You should use an open drain to drive these lines which have passive
>pull-ups. This allows you to simply wire-or two pins together, configure
>one as input, and the other as output and avoid changing the tris register
>for every byte transfered. This takes four PIC pins instead of two, but
>the simplification in the code is worth it.

I don't have any pins left...

>An oscilloscope is an excellent investment for this kind of work. A DSO
>works best, of course.
Ehh, I'm just a student... I don't have the money to buy a scope...
Unfortually

>Get the phillips spec for I2C, it's much better than what comes with the
>EEPROM.

I'll have a look, Thanx.