Searching \ for 'Intelligent Alarm Bus' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=intelligent+alarm
Search entire site for: 'Intelligent Alarm Bus'.

Truncated match.
PICList Thread
'Intelligent Alarm Bus'
1999\01\29@064913 by Michael Rigby-Jones

flavicon
face
The house next door was broken into very recently, and the police took over
an hour to turn up!  So I've decided to update my house security system from
the old clockwork system I currently have.

Obviously I could go out and buy a more modern system, but this is contrary
to the spirit of the PIC list....  I have a very tiny 386 board which I am
considering using, simplifying development.  Primarily I want the system to
page me when it goes off, I could beat the one hour response even if I was
at work 35 miles away.  I have investigated the TAP protocol which allows
alphnumeric paging via a modem and I have started writting some code for
this.

The problem comes with sensors, e.g. reed switches, PIR detectors.  There's
going to be quite a few which involves lots of cables.  I don't want to go
down the wireless route, but I thought that if the sensors could be made
slightly intelligent (with a PIC) I could use a serial bus system and cut
down on wiring.  Each sensor would have a unique address and the controller
would regularly poll all sensors.

What I need to know is what type of bus I can use.  I'd love to use I2C as
it's so simple to implement in software, but I know it's not suitable for
the long cable lengths involved, and noise immunity would be poor.  Is there
anyway to improve this, say by using a 12volt logic swing instead of 5v?  If
not, is there any other bus system suitable that can be bit banged from a
small PIC and is good for master-multiple slave use?

Thanks for any info

Mike Rigby-Jones
spam_OUTmrjonesTakeThisOuTspamnortelnetworks.com

1999\01\29@072821 by g.daniel.invent.design

flavicon
face
I have heard of entire hospitals being run with I2C,
just keep the speed down and include protection on the lines, shielding
would also help.

Alternativly you could check out the Dallas "MicroLan", I think it could
easily be adapted.
regards,
Graham Daniel.

Michael Rigby-Jones wrote:
{Quote hidden}

--
Steam engines may be out of fashion, but when you consider that an
internal combustion engine would require recovery of waste heat by
transfer just before top dead centre then fashion becomes rather
redundant, USE STRATIFIED HEAT EXCHANGERS ! and external combustion.

You heard it first from: Graham Daniel, managing director of Electronic
Product Enhancements.
Phone NZ 04 387 4347, Fax NZ 04 3874348, Cellular NZ 021 954 196.

1999\01\29@073654 by Nigel Orr

flavicon
face
At 11:36 29/01/99 -0000, you wrote:
> The problem comes with sensors, e.g. reed switches, PIR detectors.  There's
>going to be quite a few which involves lots of cables.  I don't want to go
>down the wireless route, but I thought that if the sensors could be made
>slightly intelligent (with a PIC) I could use a serial bus system and cut
>down on wiring.  Each sensor would have a unique address and the controller
>would regularly poll all sensors.
>
>What I need to know is what type of bus I can use.  I'd love to use I2C as
>it's so simple to implement in software, but I know it's not suitable for
>the long cable lengths involved, and noise immunity would be poor.  Is there
>anyway to improve this, say by using a 12volt logic swing instead of 5v?  If
>not, is there any other bus system suitable that can be bit banged from a
>small PIC and is good for master-multiple slave use?

Have a look at microlan- http://www.ibutton.com- it's a 1-wire protocol that looks
ideal for security stuff- I'm considering using it when I get round to that...

There are switch/digital input 3-wire devices available that go on the bus-
and each has a unique identity built in.  They also do an interface for a
PC serial or parallel port and assorted development software.  Prices seem
reasonable too.

Nigel

1999\01\29@074734 by keithh

flavicon
face
It is not a good idea to try kludging up I2C.
Whatever the voltage it is pulled up to,
it will be vulnerable to noise.
When a slave pulls the signal down, there's a
low-z path to ground. When floating high,
it is high-z and thus sensitive to noise.

This kind of thing has all been done before,
nearly a dozen times, for factory automation.
They're called field buses. Each has pros and cons.
Check put SDS Devicenet (based on CANbus)
or Profibus (based on high-speed RS485)
to get an idea of what's needed to do the job!

Unless you want to spend millions (really!) developing
protocols and hardware, I'd suggest a low speed RS485 network.

All the home security systems I've seen are pretty dumb,
in that the only output is an alarm.
I thought about designing a system where a central
low-power PC could log events like PIR sensors
going off in the back garden at 3 a.m.
Even if they don't break in, its worth knowing
when someone is casing your joint!

1999\01\29@080222 by keithh

flavicon
face
> I have heard of entire hospitals being run with I2C,

Holy shit! I'd feel very nervous about being sick there!

Hearing of something being done does not mean it is wise to do it!

I'd avoid any kind of bus that leaves signals in a hi-z state
for significant lengths of time. Such as I2C and Dallas "MicroLan".

I used to work with factory automation buses, and I strongly
state that while networks are simple in concept they are
non-trivial in practice!

1999\01\29@083442 by Michael Rigby-Jones

flavicon
face
Thanks for the quick response guys..

The only reason I thought I might get away with some kind I2C bus is that
data rates can be very low, say 50 bits/second would seem to be quite
acceptable, maybe lower.  I would however like to use some of the wiring
that is already in place which is the standard  6 core 7/0.2 stuff that
seems to be used everywhere for alarm systems, which obviously isn't
shielded.  6 cores would give power, ground and 4 others for signalling
which hints at using some kind of differential signalling.  RS485 is looking
good.  Shame the drivers aren't cheap, at least by I2C standards.

I found a Philips App note AN452 which describes how to use I2C up to a mile
or more, but shielded cable seems a definate requirement.

I may get some time to do a few experiments this weekend, I have several 50
meter reels of this cable.  I'll let you know next week if I find any
interesting results.

Anyone know where to get the combined PIR/Microwave detectors at a
reasonable price?  Preferably in the UK.  Thanks.  Oh yeah...is using an
electric cattle prod on theiving scumbags strictly legal?

Regards

Mike Rigby-Jones
mrjonesspamKILLspamnortelnetworks.com

1999\01\29@085608 by Jochen Feldhaar

flavicon
face
Hello Mike,

will you want to operate the cattle prod on a serial bus control line
also??
Or is it "user controlled"??

Greetings,
Jochen DH6FAZ
.....jfKILLspamspam.....detektor.de

1999\01\29@091641 by Harrison Cooper

flavicon
face
For several issues sometime ago, INK had articles on building an alarm
system, based on Steve's on design at his house.  I *might* still have
these, but they were spread across several issues, so maybe a search at
Circuit Cellars website?  This might not be what you want, or already know
about it, just thought I would pass along what little knowledge I have.

1999\01\29@114405 by Bob Blick

face
flavicon
face
On Fri, 29 Jan 1999, Michael Rigby-Jones wrote:
> reasonable price?  Preferably in the UK.  Thanks.  Oh yeah...is using an
> electric cattle prod on theiving scumbags strictly legal?

I believe it is in the great state of Texas. You might check your local
laws.

Cheers,
Bob

1999\01\29@122426 by keithh

flavicon
face
> RS485 is looking good.
> Shame the drivers aren't cheap, at least by I2C standards.
> ... how to use I2C up to a mile or more,
> but shielded cable seems a definite requirement.

Yep, its cable cost vs driver cost.

The cost difference between a mile of cable shielded well enough
to protect I2C and a mile of twisted pair will pay for X number of
RS485 driver chips. How many devices will you need, and will this
number be greater than X ?

> is using an electric cattle prod on thieving scumbags strictly legal?

The Cambridge (UK) police pushed a leaflet through my door entitled
"Beat the criminals!". I swear it's true!
Either they've taken on ex-South African staff or
they weren't quick-witted enough to see the double entendre.
Both very worrying deductions eh?

1999\01\30@071552 by jlz

flavicon
face
> The house next door was broken into very recently, and the police took over
> ....

       I have the same problem. I get some good experience with RS-485 for
       link devices (PC with industrial machines). But, there are a
       problem.

       The hipotetical smart-sensor (with a PIC and a DS3695 to convert
       +5Volts to RS-485 signals) need 2 wires to Power and 2 wires to
       communication. Total 4 wires around the house. It don't sounds very
       well. Another posibility is the smart-sensor with a power source but
       the problem is the same. 2 wires for 220V AC  around the house and
       a lot of power sources.

       I think in a IR-Link (2 wires to power and visual-air link) or
       communication across 220V AC  (2 wires to power and link) or
       RF-Link (2 wires to power and air link but in Spain the RF laws are
              very strong) but .... (to be continued)


P.D. I am sorry if there are something incomprehensible. For foreing
people is very dificult write in English language.

Jose Luis Zabalza
EraseMEjlzspam_OUTspamTakeThisOuTctv.es

1999\01\30@073838 by Glenville T. Sawyer

flavicon
face
Glenville T. Sawyer   ( the 48 year old )
Outback Communications.   South Australia
Theatre & Concert Lighting, Special Effects & Props. + more !
Also - Embedded Control systems.  http://www.gsawyer.mtx.net/emailpad.html

{Original Message removed}

1999\01\30@113435 by Donald L Burdette

picon face
I have some experience with long 1-wire busses, and I can say that it's
possible, but not trivial to make them work with 1000 feet of cable.  The
trivial case only works to 25-100' (depending on your cable).  If you
want to try it, read Dallas Semiconductor's white paper on it.  I believe
you can find it on their web site (http://www.dalsemi.com).  If not, e-mail me
and I'll send it to you.

There are three main problems with long wires.   One is capacitance,
which causes slow rise/fall times.  This can be overcome by using low bit
rates.

The second problem is noise.  Shielding, filtering, low source
impedances, differential inputs and twisted pair all help here.  You
choose your solution based on how severe the noise is and how  much you
want to spend to get the bandwidth up.

The third problem is transmission line effects.  Any cable looks like a
distributed RLC network.  For lengths in the hundreds of feet and bit
rates in the kilobaud region, some of these issues can be ignored.  But
ringing and overshoot can still be a problem.  I had one circuit where
the falling edge caused an undershoot of about 3 volts.  This in turn
caused an SCR latchup on my 'receiver' chip (let's just say it wasn't
intended for that use).

Here's my recommendation:  If you have some expertise and are willing to
research and/or play with the setup to make it work, you can make some
inexpensive modules to work on your network.  If not, you're a lot better
off coughing up the cash to get parts that are meant to work in this
environment (like RS-485).

Good luck, and let us know what you come up with.


                               Don

1999\01\30@142426 by Byron A Jeff

face picon face
>
> The house next door was broken into very recently, and the police took over
> an hour to turn up!  So I've decided to update my house security system from
> the old clockwork system I currently have.

This too is on my ever growing list of things to do. I've prototyped some
stuff including PIR and door sensors...

>
> Obviously I could go out and buy a more modern system, but this is contrary
> to the spirit of the PIC list....  I have a very tiny 386 board which I am
> considering using, simplifying development.

Actually development isn't going to be simplified that much. And the complexity
of both the hardware and software will make it a system with a smaller MTBF.

>  Primarily I want the system to
> page me when it goes off, I could beat the one hour response even if I was
> at work 35 miles away.  I have investigated the TAP protocol which allows
> alphnumeric paging via a modem and I have started writting some code for
> this.

Of course there is already code out there to do this. I've also set up a
system that uses DTMF to send numeric messages to numeric pagers. It runs
on my Linux box, but it wouldn't be too difficult to adapt to run directly
from PIC.

>
>  The problem comes with sensors, e.g. reed switches, PIR detectors.  There's
> going to be quite a few which involves lots of cables.  I don't want to go
> down the wireless route, but I thought that if the sensors could be made
> slightly intelligent (with a PIC) I could use a serial bus system and cut
> down on wiring.  Each sensor would have a unique address and the controller
> would regularly poll all sensors.

I happened to home run what I've wired to the basement, but I was also
considering this as an idea.

>
> What I need to know is what type of bus I can use.  I'd love to use I2C as
> it's so simple to implement in software, but I know it's not suitable for
> the long cable lengths involved, and noise immunity would be poor.  Is there
> anyway to improve this, say by using a 12volt logic swing instead of 5v?  If
> not, is there any other bus system suitable that can be bit banged from a
> small PIC and is good for master-multiple slave use?

Well the physical media and data transport are two different tasks. This is the
type of system that RS-485 was made for. Multi-drop. High speed. Simple and
cheap interfaces.

Data transport could easily be done using a standard serial prototcol that
is typically used for 9 bit addressing. One sends packets with the top bit
set and the address of the module in question in the rest of the bits. Everyone
receives the address packet and checks if it matches them. If not then ignore
all traffic from the network until another address packet comes along. If
you are selected, then receive the data (packets with the top bit unset) and
act upon it.

There are dozens of variations on this theme. One could reduce the number of
addresses from say 128 to 32 (which is the max that RS-485 supports without
repeaters) and have a 3 bit command. Use a call-response protocol with timeouts
so that the primary doesn't get locked out the system because a secondary has
wandered off in the weeds.

You could in fact even run I2C over RS-485. But my first thought is that a
self clocked network with no clock line would probably be a more appropriate
choice.

Just some thoughts. Good luck on your project.

BAJ

1999\01\30@201059 by Dwayne Reid

flavicon
face
Donald L Burdette wrote:

>I have some experience with long 1-wire busses, and I can say that it's
>possible, but not trivial to make them work with 1000 feet of cable.  The
>trivial case only works to 25-100' (depending on your cable).

It is POSSIBLE but you have to live within the constraints Donald mentioned.
I use such a beastie to tie multiple control panels togher in an industrial
setting - the buss uses a voltage swing of 15 Vdc (open collector with a 2K2
pullup) and operates at 512 bits per second.  Hi and lo periods are measured
within the interrupt routine and extraneous transitions are either tossed or
set the error flag, depending upon the type and severity of the error.

You have to pay a lot attention to error sources and the effect they can
have on the link.  At the minimum, checksum, CRC, or transmission of
multiple identical messages are required to ensure detection of errors.  My
implementation uses 2 of the above - a small CRC for all messages and all
commands are sent twice, with the second copy being bit inverted WRT the
first.  All individually addressable commands cause the receiving node to
generate an ACK packet - if the sender does not see the ACK within a
reasonable time, the command is resent.  Data packets sent in response to
the command rely on the CRC field for error detection - if an error is
detected, the sender resubmits the command.

I can get away with this because I don't need fast data transfers, I use
small packets (8 - 13 bytes) and response time is not an issue.  The link is
used for telemetry and updating of eeprom constants within individual nodes
- if it takes a couple of trys to get a packet thru, no problem.

I would NOT use such a link for anything where the communication link is
critical to the correct operation of the system.  Too many chances for
delayed packets.

The link I described above has worked well for the hundred or so systems
installed since 1997.  Within the limits described, it can be done.

dwayne


Dwayne Reid   <dwaynerspamspam_OUTplanet.eon.net>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
(403) 489-3199 voice     (403) 487-6397 fax

Celebrating 15 years of Engineering Innovation (1984 - 1999)

1999\01\31@185528 by Graeme Smith

flavicon
face
I was just wondering, not to long ago, why we weren't multiplexing
power and signal on the same bus....

It makes sense to me, that with a small power differential between
peak signal, and peak DC, we could send the power, and the signal
over the same line, and save a line.... The result would be a little
more electronics at the far end, but shouldn't be all that expensive.

Any Comments.... Might be good for an alarm bus.



GRAEME SMITH                         email: @spam@grysmithKILLspamspamfreenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.

1999\01\31@190148 by dave vanhorn

flavicon
face
At 04:54 PM 1/31/99 -0700, Graeme Smith wrote:
>I was just wondering, not to long ago, why we weren't multiplexing
>power and signal on the same bus....

Use two pairs, and two 1-1 transformers, centertapped. Put the DC in on the
center taps, and run clk on 1 xfmr and data on the other, or go async 422.
We did this commercially, and it worked quite well. The DC through the
transformers cancels out, so you don't saturate the core. If you multidrop,
your "feeder" may need to get large, or you can do a passive or active
repeater that powers a segment.

1999\01\31@195242 by Graeme Smith

flavicon
face
OK, let me see if I have this right...

You use the center tap on the "OUTPUT" side of the transformers, for
DC.

(This creates feedback through the secondary coil, but also acts as the
"Center" for it, essentially biasing it rather than saturating it.)

You put your "DATA LINE" and "CLOCK" line through the "Primary" thus
causing them to bias the basic OUTPUT voltage up and down. Since they
are on the primary, they are supposedly isolated from the DC voltage...

This creates a combination DC/AC signal referenced to the DC voltage...

Um.... it seems to me, you need 5 wires unless you couple the "Low" end
of the signal to the ground....

two for each signal, and one for ground...

on the other hand, if you ground the one secondary, and output off the
center tap, and the high line you could get by with three wires...

is that right?

                               GREY

GRAEME SMITH                         email: KILLspamgrysmithKILLspamspamfreenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.


On Sun, 31 Jan 1999, dave vanhorn wrote:

> Use two pairs, and two 1-1 transformers, centertapped. Put the DC in on the
> center taps, and run clk on 1 xfmr and data on the other, or go async 422.
> We did this commercially, and it worked quite well. The DC through the
> transformers cancels out, so you don't saturate the core. If you multidrop,
> your "feeder" may need to get large, or you can do a passive or active
> repeater that powers a segment.
>

1999\01\31@200112 by dave vanhorn

flavicon
face
At 05:52 PM 1/31/99 -0700, Graeme Smith wrote:
>OK, let me see if I have this right...
>
>You use the center tap on the "OUTPUT" side of the transformers, for
>DC.


      XFMR                                                    XFMR
CLK        +CLK--------------wire----------------+CLK          CLK
              +12V                                    +12V
GND        -CLK -------------wire------------------CLK           GND

      XFMR                                                   XFMR
DAT        +DAT--------------wire----------------+DAT         DAT
              GND                                     GND
GND        -DAT-------------wire-------------------DAT         GND


Like that.   +12 feeds out through both + and - clk, as a common-mode signal
CLK is differential, so it shows up on the CLK xfmr output

1999\01\31@201946 by Graeme Smith

flavicon
face
Hm... your still stuck with 4 lines..... I can see where ballancing
the + and - lines allows you to reduce the capacitative effects, etc.

But, that extra line is still expensive if it is needed on every path.

The application I was thinking of, was a prefab baseboard sensor array
with intelligent sensors. (Although, I know that the originator of this
thread had a different application). If you multiplexed the clock, and
data signals, as well, you might be able to keep this design, and drop
to three lines.... I'll keep it in mind ;)

                               GREY


GRAEME SMITH                         email: RemoveMEgrysmithTakeThisOuTspamfreenet.edmonton.ab.ca
YMCA Edmonton

Address has changed with little warning!
(I moved across the hall! :) )

Email will remain constant... at least for now.


On Sun, 31 Jan 1999, dave vanhorn wrote:

{Quote hidden}

1999\01\31@202953 by dave vanhorn

flavicon
face
At 06:19 PM 1/31/99 -0700, Graeme Smith wrote:
>Hm... your still stuck with 4 lines..... I can see where ballancing
>the + and - lines allows you to reduce the capacitative effects, etc.
>
>But, that extra line is still expensive if it is needed on every path.

4 conductor phone wire is about the cheapest thing going.
I used this for a multiplexed smart sensor system. 1024 sensors on one line.
(I used to do alarms professionally)

More... (looser matching)
- Last day of these posts
- In 1999 , 2000 only
- Today
- New search...