>You can still use metal sensing without using an oscillator in
>the strict sense. Measuring the inductance of a sense coil does
>the same job. Inductance sensing can use many means (some of which
>uses oscillators). You can also buy commercial inductance sensors fwiw.

Hmm, thinking it through a bit further, having a device on each door, I
would be tempted to do the following.

Have a reasonable Q tuned circuit in each side. About once a second have a
transistor pulse a current through the tuned circuit. This will cause it to
ring at the natural resonant frequency, and the tuned circuit in the other
box should be able to detect it. It then does the same thing about a second
later, to be listened to by the first box. They keep on doing this "ping
pong" of pulses, and note the time if the other one does not answer, or no
ping detected after a timeout period after sending the last ping. This
minimises the current draw by pulsing with a very narrow pulse, and by
sharing the drive requirements between both boxes, the battery life can be
extended further, instead of having large batteries in one box, and minimal
current requirements in the other.

Construction wise, I would be tempted to have the coils mounted in a half
potcore, with the open face of the potcore epoxied into the face of the box
which is looking at the other box. This whole face may need to be something
suitable impact resistant like polycarbonate to resist attempts at
destruction by the packing depot, who may well be in cahoots with the
thieves. The potcore will then act as a magnetic field concentrator with the
one the other side of the door becoming the pickup unit. This should give
enough pickup signal to feed a micropower op-amp for detection.

If one wanted to be real smart, then it would probably be possible to detect
how long the LC oscillated after a ping, using the receive circuit, to
detect if someone tried to fool it by slipping a piece of metal between the
doors. However the thieves may also try and use this to give a "false
opening" too early in the transit time to fool you into figuring when it was
opened, if there is a gap that allowed such a piece of metal to be inserted
without opening the doors.

Another trap might be an accomplice slipping a piece of metal in at the
packing depot as the doors are shut and locked. Perhaps a way around this
would be to have an IR shine between the two about every half hour. However
you still need some means of knowing the doors have really been shut, and
not the unit being fooled into thinking they are shut or open.

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.