I2C routines on PICs that don't have I2C built-in...?

scolyvas · Guest

Has anybody used the built-in I2C routines (e.g. I2C_POLL(), I2C_READ())on PICs that do not have dedicated I2C pins?

According to the PCW manual, the I2C_POLL() and I2C_READ() routines are only available on devices that have built-in I2C. I would like to use the I2C routines on a 16F648A which does not have built in I2C...

Thanks,
Spero.

Ttelmah · Guest

What do you actually want to do with I2C?.
The internal routines work fine to implement a _master_ in software. What they cannot do, is implement a slave. I2C_POLL cannot work, because without the hardware, there is no buffer to look at the status of. I2C_READ, also works fine _in master mode_. Again what cannot be done is to implement a slave, where a byte is automatically received and clocked into a buffer. Potentially a slave would perhaps be 'writable', but would require the use of other hardware resources (using an interrupt pin to control the transfer), at the very least, and as such would not be generic code.
So you can control an I2C peripheral, send it clocks, send it data, and read back it's replies fine with the supplied code, using software. If you need a slave, then I'm afraid it would be down to writing custom code yourself.

Best Wishes

valemike · Guest

Mark · Joined: 07 Sep 2003 Posts: 2838 Location: Atlanta, GA

Ttelmah · Guest

Yes. The problem actually is not really the polling interval, but the long latency on entering an interrupt. I'd suspect it could be done fairly easily for 'slow' I2C transfers, especially on the faster chips, but it is going to interfere with just about everything else while the transfer takes place...

Best Wishes

Mark · Joined: 07 Sep 2003 Posts: 2838 Location: Atlanta, GA

I have never tried this before, but several other Micro manufacturers insisted they could do a slave in firmware. SSP slave is a sticking point for us. I always doubted how well it would work as far as speed but if you are controlling the master as well, then you should be able to make it work. I'd just choose another micro that has one though.

Ttelmah · Guest

The PIC has enough trouble managing a SSP slave in hardware!. The latency on the interrupt response, really does mean the master has to be aware of just how slow the device will be on receiving the first byte. The actual transfers should be fairly easy, provided you are a little flexible about the actual sampling 'point', and the rate is low enough to allow a good number of instructions/bit...

Best Wishes