Path: utzoo!mnetor!uunet!cos!hadron!decuac!felix!dennisg
From: dennisg@felix.UUCP (Dennis Griesser)
Newsgroups: sci.electronics
Subject: Re: House Automation (other approaches)
Message-ID: <20522@felix.UUCP>
Date: 4 Feb 88 22:40:38 GMT
References: <745@PT.CS.CMU.EDU>
Sender: daemon@felix.UUCP
Reply-To: dennisg@felix.UUCP (Dennis Griesser)
Organization: FileNet Corp., Costa Mesa, CA
Lines: 71

According to recent postings, both Kevin Dowling and Brian Reid have incredible
home automation projects in the works.  I eagerly await progress reports and
additional information on both projects.  Cost information would be nice too.
How much does it cost to run 25-conductor phone cable per room...

One thing that made me sad was that we seem to be working harder but not
smarter on this kind of project.  Yes, I want to rip up the walls too.  But
isn't there a less dramatic way to do it?

For example, Brian has
> ...2 500-MHz video channels from the utility room to each room of the house,
> and 25 twisted pairs to the same locations.
> ...a bunch of random cables ... such as doorbell buttons, several thermostats,
> exterior sensors, etc.
I'm inclined to allow the video, but isn't 25 twisted pairs per room overkill?
Especially since the stereo gets piped in through other wires.
In a given room, the allocation might look like this:
	pair	function
	----    --------
	  1	unregulated DC for smart thermostats and such
	  2	telephone
	  3	thermostat high-point for this room
	  4	thermostat low-point for this room
	  5	intrusion sensors (perimeter switch loop)
	  6	command console
	  7	air vent control
        ...	and about five pairs RFU (Reserved for Future Use)

That's still overkill!  How about this one:
	pair	function
	----    --------
	  1	unregulated DC for smart things
	  2	telephone
	  3	twisted-pair LAN (Local Area Network)
        ...	and about four pairs RFU (Reserved for Future Use)

Each room would contain one or more LAN nodes.  The nodes would poll intrusion
sensors and command consoles and ship the results to the home control
center in the utility room.  Commands from the control center would be
received by the nodes and fan out to devices such as lamp controllers and
air vent controls.

The LAN nodes could be single-chip MCUs with very little support logic and
an operating program in on-chip EPROM.  The average node would have 8 digital
input lines, 8 digital output lines, and a channel each for analog input and
output.  Hang a thermistor on the analog input and you can chuck the commercial
thermostat.

The lower-level LAN protocol could be simple indeed, if the control center
simply polled each station in sequence.  The node command repertoire might
look like this:
	o tell me your digital input
	o present this digital output
	o tell me your analog input
	o present this analog output
This is sufficiently general that you can do anything without special node
software or hardware.  System configuration and I/O interpretation are the
duty of the control center.  The protocol would include all of the "good network
stuff" like error detection, timeouts, retransmission, ...

If you are careful, a DC bias could be impressed on the twisted-pair LAN cable
to power the nodes.  Sounds like Kevin's TRI/CON wiring is sounding better
and better.

Such a LAN node should be buildable for about $10 per node in quantity.  Perhaps
$25 each if you build them yourself.

Now, let's step back and try to avoid reinventing the wheel.  SOMEBODY must
have designed some such system before.  I've never heard of it, but perhaps
Kevin has.  What's this "CSBus" protocol?  What else is out there?  How much
does it cost?  What can it do?  Life, the universe, everything!