Here's a datasheet on the 555:
http://www.ti.com/lit/ds/symlink/lm555.pdf
Probably gonna be tough, if you have to do it from your parts box. Look at the app note for the LED.
I think RESET would have to be connected to a monostable with edge detection. (i.e. door closed). Door open would provide the trigger pulse.
Then again, you need edge detection to determine if the timer times out.
A FF would do the alarm on/off type stuff. if you have power all the time you could use a DPDT relay in the standard self-latching mode. One contact is connected so that it holds the coil on. The alarm pulse that is in parallel to the relay and the NC pushbutton that turns off power. A standard start/stop station.
These, hhttp://
www.fairchildsemi.com/datasheets/74/74VHC123A.pdf or similar chips have the NICE ability to do positive or negative edge detection.
Some info here:
**broken link removed**
There's always issues with squaring up the signals (Schmidt trigger) so the edge detector works. There are logic level translation issues; for instance CMOS is mixed with bi-polar logic. The new logic families fix that.
The 555 is 1.1 * R * C and at like 2 M and 2 uf, it's like 4 seconds. When you start adding electrolytic caps in the timing circuit, leakage and tolerances get to be an issue. Polyester caps are available in +-5%. Some electrolytics are +80%, -20%.
You didn't mention if your switch is Closed or open when the door is open, so that could require inversion.
Power-up issues have always been a problem. So, with stuff dependent on the state at power up, some sort of reset circuit is normally required.
I liked the now obsolete LM3905 timer.
So basically, I see something like.
Switch conditioning (Schmidt trigger)
Edge detection (I'll pick positive, for now) which triggers the timing interval
Edge detection (negative (closed door)) ; resets with a pulse (another monostable)
If the timer gets to the end (another edge detector), the alarm FF (relay latch) stays on.
Two resets you have to worry about: 1) power on and 2) alarm reset.
SOMETIMES an RC circuit works, but an RC followed by a Schmidt trigger is usually better. I purchased a single board computer that would not reset reliably with a linear supply but worked fine with a SMPS. I modified the power up reset circuit.
Edge inputs typically need to be square edges.