My lack of experience shows here; my goal is long battery life, my decision is to use the "sleep" functions of the hardware, or just disable power via the voltage regulator enable pins.

I'm designing a battery powered board with a micro controller, a few sensors and a radio. Every 60-120 seconds, the board will wake up, read a couple of sensors, send a few bytes of data over the radio, then go back to sleep. The sensors will run at 5v, the radio runs at 3.3v, so I need 2 voltage regulators. I'm using 4 AAA batteries and want 6-months+ battery life. My fear is that the voltage regulators are going to eat power while the rest of the hardware sleeps (I'm still shopping for voltage regulators). So my thought was, use a real time clock to enable the voltage regulators, at which point the micro controller will boot and take over the "enable" line keeping the voltage regulators running. When the work is done, the micro processor will turn off the voltage regulators via the enable pins so nothing drains power. The real time clock will then expire and the process repeats. Assuming the real time clock is sufficiently efficient, is there a good reason not to do it this way?

 

You can do that.
How about using a button cell to keep the RTC running and disable all the uC peripherals during sleep

 

That was my thoughts as well, when it awakes it turns on the voltage regulators, a diode could be used to avoid power to the cells at this point.
The sleep mode of a Picmicro is 20Na.
Max.

 

Thanks guys, I was thinking exactly what you're thinking. We're using an Atmel 328p processor and an XBee radio, plus a small pile of supporting hardware for the various sensors. Controlling the power to everything with the RTC will simplify a lot of things, I just wanted to be sure I'm not making some obvious error that just happens to not be obvious to me.