On Sun, 12 Jul 1998, Timothy D. Gray wrote:

> Sure :-) tell me more how the standard shock sensor doesn't do this when a
> bird leaves evidence on your car, or the wind blows, or a dog farts near
> it. I actuall like the standar car alarm... Go through a parking lot, and
> push down on the hoods, set's them off nicely!  (BTW, my car has a
> clearance of 3 inches.. I had 18 inches of blowing snow build up and the
> alarm never went off :-)  as for bottles and burger wraps.. only if ya
> live in detroit, chicago or NY.. most of america is quite clean.

Ok, my dream car alarm uses 2 of those new piezo accelerometer sensors
(Murata etc)  coupled to digital integrators (in the PIC) to implement a
crude INS. It has a single PIR sensor in the passenger compartment and the
usual switches under the hood(s), ECU cutout, power monitor etc, and a
single piezo shock microphone screwed to the car body inside somewhere to
pick up vandals. For the wheels, use special nuts (there are firms that
make them one-of-a-kind.  Want 3 prongs, 37.15 and 21 degrees apart,
neither of them centered, or round, or deep ? ;)

Advantages: NOBODY will move that car without the integrators picking it
up. Not even downwards or sideways. Nothing to mount under the car body
(no holes that void the rust-proofing warranty). Anyone going for the
wheel nuts with brute force trips the shock microphone, via noise in the
body. And, because of the INS, the alarm can make a difference between
someone bumping the car while parking (which should give a short honk or
two) and a vandal banging the car without bumping/moving it first.
Lifting/towing can set off a different alarm than any other type of
motion. Go on and on and on... Nice ?

The only question is, how smart a shock microphone and its filtering have
to be, to pick up delicate tampering with the special wheel nuts through
the suspension. My educated guess is, that 3 active filters (high, low and
bandpass), and an AGC amplifier with outputs measurable by A/D on a PIC
will do fine to find the difference between a 500 HP Diesel engine
vibrating one meter away, high-pitched pings and scratching from tools,
(and birds !) and the mid-range thunk of a baseball bat hitting the body
somewhere (or of chestnuts falling off a tree) ;).

I suppose that it would have to learn each car ;) Imagine the face of an
installing technician reading the instructions in the booklet: "Select
test mode 1 using the jig, then exit the car, close all its doors and
windows, and whack each main exterior body panel of the car three or more
times, using a tightly folded newspaper, weighing not less than 300 grams,
spending no less than 3 seconds for each panel, and pausing for 5 seconds
before proceeding with the next panel. Select test mode 2 and exit the
car, to throw the small heavy rubber balls (use the 3 golf balls supplied
in the installation kit) against each panel of the car." <VBG>

Unrelated, this reminds me of certain magic rituals required to put the
camcorders I work with into service mode. It's like in a fairy tale. Eject
cassette compartment. Press 2 certain buttons on the remote control while
turning the mode dial left. Then, press the zoom and the date button
together... and it goes on and on until you are 'in'. The rationale is of
course, that the user has a probability of about 1/((No_of_buttons!) ^
No_of_steps_to_be_taken) to hit the jig setup combination by accident.
With the things I'm working on, the number is somewhere around 1/(32!)e3
;)

Peter