From: firstname.lastname@example.org (Peet Grobler)
Subject: Re: Alternative design to this?
Date: Wed, 25 Sep 2002 06:49:39 GMT
X-Newsreader: Forte Free Agent 1.21/32.243
On 23 Sep 2002 10:00:37 -0700, email@example.com (N. Thornton) wrote:
>firstname.lastname@example.org (Peet Grobler) wrote in message news:<email@example.com>...
>> I'd like to start off with home automation (my own version thereof).
>> First off, I'd like to have a computer monitoring the outbuildings (so
>> my wife doesn't kill me if something goes wrong).
>> I have 33 inputs, including all doors, windows, lights, etc. I want to
>> have a PC monitoring all these, and logging changes in them. Later on
>> I'll add temperature monitors.
>Not very with it today, so this may be off the mark... you're sending
>8 bit data decribing 1 bit situations, like lights on or off. So it
>appears you could combine many pieces of info into one single byte of
>8 bit data. That would reduce your 33 dramatically.
>Point 2 is you only _need_ one wire, not a pair, assuming you've got
>an earth wire already. It is possible that very occasionally you could
>get a data byte messed up, but not often enough to have any real
>effect in real world terms. If you never ever want that to happen,
>just tell the PC to average the result of a few successive readings,
>or to re-question the device if the reported value changes, or to only
>believe a result when 3 successive readings are the same.
Nope, you lost me. I only need one wire? So let's say the sensor gets
activated. Then what?
>There are more old fashioned ways to do it, using very cheap parts.
>One example would be that each detector interface has a voltage window
>detector. When the PC end questions it, PC end sends the relevant V
>corresponding to the target device, and target device responds by
>either doing nothing or shorting the line until the i is removed. Thus
>you have a lot of 1 bit data senders, and very simple ones too. One
>very IC could do that.
This would involve making changes to the existing sensors, which will
incur unneccesary cost and troubles. I'd rather use the standard
>The PC end questionsing voltage is developed by putting Rs on a PC
>port, as a simple D-A. The PC end listening device is a simple opamp,
>giving hi or lo output.
>** Much better **
>Then there are even simpler methods. Each detector connects to the
>line via an R: the set of 8 detectors on one wire gives you an 8 bit
>voltage value (or current). Very very simple. For 33 devices you'll
>need 3 signal lines using 8 bit, or more if you use a lower bit code.
>That sounds to me like a real winner. Ultra simple, ultra cheap, very
>expandable. When you add temp senmsors later, you can decide what the
>important temp break points are, then you'd only need say 4 bits per
>temp.. or go the whole 8bit level per temp sensor.
Yes... I understand this, and would definately investigate it further.
If I understand correctly, you have 8 different R's on one line, each
giong to a sensor. Feed this line into an A/D, and get a value (in the
software). Use this value to determine which one of the 8 devices got
triggered. This would speed up the entire scanning process, since I'll
be doing 3 scans to read everything, instead of 33. Also, I can check
8 devices at a time, if nothing changed, read on...
Hey, thanks. I'll experiment with this real soon. This is something I
haven't done before, that's why I'm so interested!