I did this a short while back. The device under test is powered by a
dropping capacitor, so isolation was required for the interface. It's not
quite the I2C protocol, but modified to reduce code size. It's a PIC12CE518.
The I2C on the measurement side (an 8051) ended up using three pins (clock,
tx, rx). The PIC device under test uses two pins with pullup resistors. Same
principle as I2C, though. I used three (3) 4N36 with the base grounded
through a resistor to increase turn-off times of the phototransistor. The
base resistor is a compromise between speed and current. So you might play
with that for your circuit. One for the clock, one for tx, one for rx. I
also ended up using a small NPN to interface to the circuit under test,
which has very low drive capability. The whole thing is powered by an audio
transformer driven like a half-bridge converter to provide the power for the
optos. No feedback to keep it simple. Since the PIC is powered by an Xcap,
it really has very limited drive, so this was necessary.

I can take a quick rise/fall time measurement of my configuration if you are
interested. Feel free to ask for more details.


> -----Original Message-----
> From: Alan B. Pearce [mailto:A.B.Pearce@RL.AC.UK]
> Sent: Friday, August 17, 2001 4:58 AM
> To: PICLIST@MITVMA.MIT.EDU
> Subject: [EE]: Isolated I2C
>
>
> I have a requirement to have an I2C link across an isolation
> boundary. The
> voltage requirements of the boundary are not high, it is just
> that I am
> needing to emulate some circuitry for testing purposes from
> the rest of the
> test equipment.
>
> --
> http://www.piclist.com hint: The PICList is archived three different
> ways.  See http://www.piclist.com/#archives for details.
>
>

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.