-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Wed, Jun 20, 2007 at 09:46:42PM +0200, Hector Martin wrote:
> > Then again I could rely on the LEDs being off when communication needs
> > to happen... but that would require the cooperation of all devices on
> > the line.
> 
> - Master switches to soft pull-up
> - LEDs short input
> - Slaves detect this
> - Slaves turn off LEDs
> - Slaves and master wait until comms line rises again
> - Communication begins

How would the slave detect that the LED is what is shorting the input?

Are you just assuming that the second the data line goes low, the start
that sequence?

You could potentially have the situation where all the leds are off, and
therefore the data line stays high, so the sequence would need to be:


Master switches to soft pull-up and pulls line low
Master notes line is low, attempts to bring it high
Line doesn't go high, so master retries until it does

Meanwhile:

Slaves detect this
Slaves turn off LEDs
Slaves and master wait until comms line rises again
Communication begins

> Sure, all devices have to turn the LEDs off, but you don't have to set 
> up any funky synchronization or "slot" system. Is there a problem with 
> requiring devices follow the spec? If reliability is a concern, then the 
> original system won't work either - any slave stuck in a pull down data 
> line condition would kill the bus for everyone.

True, but that error mode can be prevented by the watchdog timer. It's
not perfect your right, but I think having the slaves responsible for
turning leds off, an "active" thing, is more likely to fail than hoping
the watch dog timer will kick in.


Of course either way I'll have to be ready to disconnect nodes,
squirrels chewing cables have never heard of a watchdog timer. :)

> Personally, I'd just use three wires, one for power. It makes things a 
> lot easier.

What would be a good way to get really cheap, three pair cable?

- -- 
http://petertodd.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFGebLd3bMhDbI9xWQRAhJ9AJ4yyqexwRl9TCw7DlQ6RYxWX3agLgCgsbXK
PkvDSB5jVuovbJThAmJOUL4=
=1OAj
-----END PGP SIGNATURE-----
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist