Problem: as solar power turns on in the morning it acts like a gradually increasing current source. If I try to power a microcontroller with it (e.g. AVR) then the MCU will hiccup (on and off) until enough current is constantly available. Unfortunately such a poor power up sequence can latch-up CMOS devices.
Left IO: connects the Battery to PWR (blue).
Right IO: connects the PWR node (blue) to LOAD (green).
Switch: is used to remove charge from 20uF, so the load latch can reset, it is not part of the real circuit. The 20uF is output side of a buck converter used to charge the battery.
Add some capacitance between the upper PMOS gate and source. The 1nF Cgs divides a quick voltage change on PWR (blue), such that 90% is on the 100pF Cgd. Unfortunately, 10% of 12.8V on Cgs is enough to turn on the PMOS for a moment. Increasing the Cgs reduces the instantaneous voltage drop, e.g. 100nF give a .1% drop of the instantaneous voltage change across the gate-source and that is not enough to turn it on.
5kOhm ramps the 100nF slow enough to limit overshoot in the load. While keeping the lower PMOS that connects battery at 1nF which allows it to be disconnected fast.
http://en.wikipedia.org/wiki/Power_MOSFET#Gate_to_source_capacitance
CC01 is 6V and does not need a Vgs capacitor (1nF is enough) but at 12V CC03 needs about 10nF.
Circuit board(s) I am working on with this:
http://epccs.org/indexes/Board/CC01/
http://epccs.org/indexes/Board/CC03/
