Re: TI-H: I2C Multimaster Routines revisited
[Prev][Next][Index][Thread]
Re: TI-H: I2C Multimaster Routines revisited
Christopher Kalos wrote:
>
> Ozone mentioned working on Multimaster I2C setup for a TI network, but
> bandwidth issues got in the way. What if we used a Token Ring Style
> setup? It would allow for more bandwidth (theoretically) to be
> available at once. End result: a workable, minimal overlap TI network.
> It would work best if each calc were assigned a number, kind of like IP
> addressing.
It was me who was (and in theory, still is) working on it. It's just
that I've had my summer holidays and hadn't had time to finish the
routines. The receive routines are in working condition (v0.92 available
at www.clinet.fi/~ozone/i2c ) but the send routines still need lots of
debugging. I'll see if I've got time for this project next week.
The connection style itself doesn't matter as long as all SDA pins (one
data pin of the link/SPinTerface port) are connected together, as are
the SCL pins (the other pin), and of course ground. Every calculator is
assigned an address pair between 2-3 and 254-255 (receive and send
address). Additionally, 0 is the generall call address which is received
and interpreted by every calculator on the network. Other I2C devices
like I/O expanders, temp sensors, EEPROMs and microcontrollers can also
be connected to the same network without interference, as long as every
device has its individual address.
-Ozone
--
*** Osma Suominen *** ozone@clinet.fi *** http://www.clinet.fi/~ozone
***
References: