I'm making a floppy drive emulator for the Epson PX-8 CP/M computer. It is basically an atmega1284 + hcms 8 char LED display + microsd socket + rs232 transceiver + a couple of buttons. It is a 3.3V project, but like the rest of the PX-8 system, it is going to run on 4xAA batteries. Someone can use alkaline or NiMH/NiCD (like the PX-8 does). I tested one of those linear replacement SMPS blocks and it failed miserably because it didn't say working below 5.5V very well despite that its datasheet says it can go down to 4.5V. It turns out that it needs a large load to do that. I gave up on that and decided to build a SPMS using a TI LMR16006YQ3DDCRQ1 after their website tool recommended it to me and said it could go down to 4.0V. I somehow soldered it together using pins and doing a component to component point soldering and it works wonderfully. It looks like a mess, but it works great. It starts up even with no load at 4.0V and it seems to operate even down to 3.5V which surprised me, load or not.
The two battery types I want to support, I have voltage ranges of 3.2V-6.4V (alkaline) and 4.4V-5.8V (NiMH/NiCd). I am not too concerned if the alkalines are not fully drained to 0.8V, but it would be nice if they were used down to 0.9V (3.6V for all 4) of them. It will also have the option of a wall adapter which will be combined with the battery source using a couple of Schottky diodes. I also plan to have a slow unmanaged charging option that charges the batteries at 0.05C which the NiMH or NiCd should tolerate just fine if someone leaves it plugged in.
Here is the question - the mega1284 will have the BOD set at 2.7V, but the only way the output of the SMPS will drop below 3.3V is if the input falls to 3.5V or below. If the AVR goes into reset, the LED display will also go into reset because the AVR must actively take the display out of reset to enable it. I guess what I am saying is that if the output drops below 2.7V, the AVR will go into reset and current will be reduced. There is an external oscillator however so even in reset, it may use 5mA.
There is also however a shutdown feature on the SMPS. There is a pin that when driven by a voltage divider that can cause a shutdown of the SMPS to just 1uA or so. You pick a voltage divider to turn your desired cutoff voltage into 1.23V. The voltage divider is on the input side, so it will consume current, but the datasheet said it could be int he 100K range, so 3.5V / 100K = 35uA. So implementing the shutdown feature takes 2 resistors and would cut the current from 5mA to around 35uA if someone leaves it on and forgets about it allowing the batteries to deplete endlessly.
Would you implement the shutdown feature to protect batteries from being drained too far (and possibly causing damage to NiMh or leakage with alkalines)? If so, what voltage would you pick as a cutoff given the battery chemistry options?