This is a followup to an earlier post as came out in the comments you can't wake up an xBee directly with the output from the motion sensor, it's the wrong polarity. It connects to Vcc when motion is detected and floats when there is no motion. Just adding a pulldown and a transistor didn't work at all to connect it to ground and wake up the xBee. Too many timing issues.
So I decided to go with an ATTiny85 chip that can sleep most of the time and be woken up itself when the sensor goes high, and it can then handle the timing of waking up the xBee and having it sample it's pins and send the data. This actually worked lovely until I tried to add back in the motion sensor.
It turns out these motion sensors just dont work below 3v or so. They have an onboard voltage regulator for the 3.3v that they need to work so you really must power them with 5v to get them going. There is some talk on teh intertubes that you can just short across the regulator on the board. I think those instructions were written before they switched to all surface mount parts on these boards which makes it much more difficult. I did try doing that though and it did work then at 3.3v quite well.
you can see my tiny little wire loop across the voltage regulator. Unfortunately at about 3v and lower it stops behaving at all which means that 2AA batteries are not going to power it for very long. And it also stops behaving above 4.5v which means that 4 batteries are a no go. The ATTiny has a good range of voltages in this area which makes it excellent for a long term battery operated product but the xBee is very sensitive to voltages above 3.3 volts. I wouldn't want to run it on 3 batteries without a regulator. Which means that would require a regulator which would waste a lot power and significantly reduce battery life.
I've looked around quite a bit, and there are some excellent canned solutions to motion sensing out there, including some fantastic chip/sensor combinations from Zilog (who also offer an SMT version of the Z80 CPU chip which somehow I find funny :) but they use too much current and still dont operate at a low enough voltage for a good battery operated project.
So we're left with separate power supplies for the sensor and the xBee or regulators or rolling it from scratch. I might experiment with a 3AA battery system where the sensor runs with it's regulator off 3 batteries and see what the current usage is and tap just 2 of the batteries to power the xBee. They will run down out of sync but that might still work and 3 batteries in the device isn't out of the question. Or it really might be time to start studying how these sensors work and roll one from scratch. The otherwise quite good except in radio range and reliability X10 motion sensors work for years on 2AAA batteries. I've glanced at their insides and I think they do a op amp comparitor thing between the 2 outputs of the motion sensor and then have a tiny PIC to make decisions on the output.
At least I understand why the newer ZWave motion sensors are so expensive and why the battery life in them sucks so badly overall.
But all this does mean that without the motion sensor the combination of an ATTiny and an xBee radio work great. So other sensors should be a piece of cake, something like a door/window contact closure device that might also sample temperature and light levels on a regular basis should be quite possible. I might concentrate on that for the moment and stew over the motion sensor problems a little longer.